
From muenz@net.in.tum.de  Mon Mar  1 00:43:07 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 4F45228C27C for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 00:43:07 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id aFFs6cN59H33 for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 00:43:05 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 810643A7930 for <ipfix@ietf.org>; Mon,  1 Mar 2010 00:43:05 -0800 (PST)
Received: from [192.168.2.152] (ppp-93-104-74-113.dynamic.mnet-online.de [93.104.74.113]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 58EA62243A81; Mon,  1 Mar 2010 09:43:03 +0100 (CET)
Message-ID: <4B8B7E20.1020504@net.in.tum.de>
Date: Mon, 01 Mar 2010 09:43:12 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Atsushi Kobayashi <akoba@nttv6.net>
References: <4B59DAB8.8040705@net.in.tum.de> <20100126100448.0A02.17391CF2@nttv6.net> <20100209010851.F028.17391CF2@nttv6.net>
In-Reply-To: <20100209010851.F028.17391CF2@nttv6.net>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms090800040709090207020701"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] WGLC comments on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Mar 2010 08:43:07 -0000

This is a cryptographically signed message in MIME format.

--------------ms090800040709090207020701
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Atsushi,

Atsushi Kobayashi wrote:
> Hi Gerhard, and all,
>=20
> I find my following answer could not have been shared in list due to ou=
r
> network trouble.

Thank you for you comments. As far as I can see, there are only two=20
issues left:

1) Match Filters

>>>>>>> 4.2.3.  Filter Classes
>>>>>>>
>>>>>>>            +--------------------------+
>>>>>>>            | FilterMatch              |
>>>>>>>            +--------------------------+
>>>>>>>            | ieId/ieName              |
>>>>>>>            | ieEnterpriseNumber[0..1] |
>>>>>>>            | value                    |
>>>>>>>            +--------------------------+
>>>>>>>                          Figure 10: Filter classes
>>>>>>>
>>>>>>>    The Filter classes in Figure 10 contain the configuration para=
meters
>>>>>>>    of specific Filtering methods.  For property match filtering, =
the
>>>>>>>    configuration parameters are:
>>>>>>>
>>>>>>>    ieId, ieName, ieEnterpriseNumber:  The property to be matched =
is
>>>>>>>       specified by either ieId or ieName, specifying the ID or na=
me of
>>>>>>>       the Information Element, respectively. ieEnterpriseNumber M=
UST be
>>>>>>>       used for enterprise-specific Information Elements.  If
>>>>>>>       ieEnterpriseNumber is omitted or zero, this is Information =
Element
>>>>>>>       is not enterprise-specific but registered at IANA.
>>>>>>>
>>>>>>>    value:  Matching value.
>>>>>>>
>>>>>> Can the ranges of value be defined like hash value?
>>>>> No.  The reason is that min/max fields are not specified in the Sel=
ector=20
>>>>> Report Interpretation:
>>>>> http://tools.ietf.org/html/rfc5476#section-6.5.2.5
>>>>>
>>>>> In the case of FilterHash, we do min/max fields in the Selector Rep=
ort=20
>>>>> Interpretation:
>>>>> http://tools.ietf.org/html/rfc5476#section-6.5.2.6
>>>>>
>>>>>> And, I would like to specify its instruction, "match" or "unmatch"=
=2E
>>>>> This is not specified in the Selector Report Interpretation. Theref=
ore,=20
>>>>> we cannot allow configuring this.
>>>>>
>>>> Above all, they apparently would be required from viewpoint of usabi=
lity.
>>> I agree. But this is not the document to solve this problem.
>>>
>>>> Please see http://tools.ietf.org/html/rfc5475#section-6.1 that descr=
ibed
>>>> it as an extension of the basic model.
>>>>
>>>> Implementor will add above functions because only exact match is too=

>>>> poor. In that case, can IPFIX-CONFIG expand to add these functions? =

>>>> How does it expand?
>>> The model can be extensible by vendors or in later revisions of the m=
odel.
>>>
>> Ok, I understood what you would like to say. I am concerned that the
>> extensions are built indiscriminately.

To summarize this issue: PSAMP protocol only specifies simplistic match=20
filters, and the current configuration data model supports the=20
configuration of such filters.

If the IPFIX WG wants the configuration data model to offer more=20
configuration possibilities than simple match filters, the WG must be=20
aware of the fact that this Selector configuration cannot be reported=20
within a Selector Report Interpretation.

Therfore, I will not extend the configuration data model unless we have=20
reached consensus within the IPFIX WG that this restriction is not a=20
problem.


2) sourceIpAddress of an Exporting Process

>>>>>>>    sourceIpAddress:  This parameter sets the source IP address.  =
If this
>>>>>>>       parameter is omitted, the address assigned to the outgoing
>>>>>>>       interface is used.
>>>>>>>
>>>>>> I would like to make available in the case of TCP.
>>>>> We adopted this parameter because the Cisco CLI has it for UDP.  As=
 far=20
>>>>> as I understand, it is not used to specify the outgoing interface b=
ut=20
>>>>> the (forged) source address in the IP header only.
>>>>>
>>>> No. In current network operation, source IP address is important
>>>> parameter as a indicator of exporting device(router). Operators are
>>>> often confused that a source IP address is changed when route change=
s
>>>> happen. Exporting device may select source address as a optimal exit=
 points, or
>>>> arbitrarily.  Then, some collectors may fail to recognize identical
>>>> exporting device before and after route changes.
>>>  >
>>>> This function is necessary to avoid the confusion.
>>> I will ask Benoit how this parameter is used.
>>>
>>>>> How do you want to forge the IP address of a TCP endpoint?  It must=
=20
>>>>> belong to the interface.
>>>>>
>>>>> The parameter description remains unclear in the case of multiple=20
>>>>> addresses assigned to a single interface.
>>>>>
>>>>> I'm not sure what to do.
>>>>> We can remove the parameter.
>>>>> We can change to "...one of the addresses assigned to the outgoing =

>>>>> interface is used."
>>>>> We can try to find a generic solution that defines the outgoing int=
erface.
>>>> I propose the following.
>>>>
>>>> The following parameters are applicable if TCP or UDP is transport
>>>> protocol:
>>>>
>>>> sourceIpAddress:  This parameter sets the source IP address.=20
>>>> If this parameter is omitted, Exporter selects outgoing interface
>>>> according to routing policy, and then arbitrarily assigns the addres=
s.
>>> You cannot use an arbitrary source IP address with TCP.
>>>
>> I mean exporting device may select appropriate source ip
>> address as TCP client along with routing policy if the parameter is
>> omitted. The routing policy may change between before and after its
>> shutdown or link-failure, etc. This may cause in the case of TCP.
>>
>> In simple example, if you can use source IP address on loopback, it
>> avoid this problem.

After discussing this issue with you and Paul, I've learned that the=20
source IP address parameter can be configured independently from the=20
interface.

A note about how this works on *ix systems:
The IP address is configured with bind() while the interface is=20
configured with socket option SO_BINDTODEVICE. In the case of SCTP,=20
multiple source IP can be bound using sctp_bindx().

So, I propose to have the parameter sourceIpAddress for all transport=20
protocols. In the case of SCTP, it replaces localIPAddress and thus may=20
appear more than once.
In addition, I propose to add parameters ifIndex/ifName to optionally=20
specify the interface.

NEW:

4.4.1.  Destination Class

    +------------------------------------------------+
    | Destination                                    |
    +------------------------------------------------+
    | name                                           |<>-----+
    | exportMemberType =3D parallel                    |       | 0..1
    | ipfixVersion =3D 10                              |       |
    | transportProtocol                              |   +------------+
=3D> | sourceIpAddress[0..*] {[0..1] for UDP and TCP} |   | Transport- |
    | destinationIpAddress                           |   | Layer-     |
    | destinationPort =3D 4739|4740                    |   | Security   |=

=3D> | ifName/ifIndex[0..1]                           |   +------------+
    | sendBufferSize {opt.}                          |
    | rateLimit[0..1]                                |
    | timedReliability =3D 0 {SCTP only}               |
    | numberOfStreams {opt.} {SCTP only}             |
    | templateRefreshTimeout =3D 600 {UDP only}        |
    | optionsTemplateRefreshTimeout =3D 600 {UDP only} |
    | templateRefreshPacket[0..1] {UDP only}         |
    | optionsTemplateRefreshPacket[0..1] {UDP only}  |
    +------------------------------------------------+
                        Figure 14: Destination class


    sourceIpAddress:  This parameter specifies the source IP address(es)
       used by the Exporting Process.
       If the transport protocol is SCTP, this parameter MAY appear
       multiple times to specify the list of eligible local IP addresses
       of a multi-homed Exporting Process [RFC4960].  If this parameter
       is omitted and the transport protocol is SCTP, all locally
       assigned IP addresses are eligible local IP addresses.
       If the transport protocol is UDP or TCP, this parameter MAY appear=

       only once. If this parameter is omitted and the transport protocol=

       is UDP or TCP, the IP address assigned to the outgoing interface
       is used as source IP address.

    ifIndex/ifName:  Either the index or the name of the interface used
       by the Exporting Process to export IPFIX messages to the given
       destination MAY be specified according to corresponding objects in=

       the IF-MIB <xref target=3D"RFC2863"/>.  If omitted, the Exporting
       Process selects the outgoing interface based on local routing
       decision and accepts return traffic, such as transport layer
       acknowledgments, on all available interfaces.

Do you agree with these changes?

Regards,
Gerhard


--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms090800040709090207020701
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms090800040709090207020701--

From muenz@net.in.tum.de  Mon Mar  1 00:49:03 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 3077128C275 for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 00:49:03 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id XNslT07T33lr for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 00:49:02 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id C47883A7930 for <ipfix@ietf.org>; Mon,  1 Mar 2010 00:49:01 -0800 (PST)
Received: from [192.168.2.152] (ppp-93-104-74-113.dynamic.mnet-online.de [93.104.74.113]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 8718E2243A81; Mon,  1 Mar 2010 09:49:00 +0100 (CET)
Message-ID: <4B8B7F85.8010205@net.in.tum.de>
Date: Mon, 01 Mar 2010 09:49:09 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: "Romascanu, Dan (Dan)" <dromasca@avaya.com>
References: <EDC652A26FB23C4EB6384A4584434A0401F8CBE4@307622ANEX5.global.avaya.com>
In-Reply-To: <EDC652A26FB23C4EB6384A4584434A0401F8CBE4@307622ANEX5.global.avaya.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms040408060408050504000304"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] FW: [Technical Errata Reported] RFC5477 (2052)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Mar 2010 08:49:03 -0000

This is a cryptographically signed message in MIME format.

--------------ms040408060408050504000304
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi all,

If we accept errata 2052 [1], we do not need erratas 2053 to 2059 [2]=20
because the examples do not show any reduced size encoding.

[1] http://www.rfc-editor.org/errata_search.php?rfc=3D5477&eid=3D2052
[2] http://www.rfc-editor.org/errata_search.php?rfc=3D5476

Regards,
Gerhard


Romascanu, Dan (Dan) wrote:
> =20
> This one is Technical!=20
>=20
> WG members, please suggest a resolution.
>=20
> Thanks and Regards,
>=20
> Dan
>=20
> -----Original Message-----
> From: Paul Aitken [mailto:paitken@cisco.com]=20
> Sent: Friday, February 26, 2010 1:31 AM
> To: RFC Errata System
> Cc: Thomas.Dietz@nw.neclab.eu; bclaise@cisco.com;
> dressler@informatik.uni-erlangen.de; carle@informatik.uni-tuebingen.de;=

> Romascanu, Dan (Dan); rbonica@juniper.net; quittek@netlab.nec.de;
> psamp@ietf.org
> Subject: Re: [Technical Errata Reported] RFC5477 (2052)
>=20
> Errata errata, sigh.
>=20
> See inline...
>=20
>=20
> RFC Errata System wrote:
>> The following errata report has been submitted for RFC5477,=20
>> "Information Model for Packet Sampling Exports".
>>
>> --------------------------------------
>> You may review the report below and at:
>> http://www.rfc-editor.org/errata_search.php?rfc=3D5477&eid=3D2052
>>
>> --------------------------------------
>> Type: Technical
>> Reported by: Paul Aitken <paitken@cisco.com>
>>
>> Section: 8.1.2
>>
>> Original Text
>> -------------
>> Abstract Data Type:  unsigned16
>>
>> Corrected Text
>> --------------
>> Abstract Data Type:  unsigned32
>=20
> "unsigned64".
>=20
> P.
>=20
>> Notes
>> -----
>> Change selectorID from 16 bits to 64 bits per discussion on the IPFIX
> mailing list:
>>
>>
>> http://www.ietf.org/mail-archive/web/ipfix/current/msg05133.html
>>
>> Instructions:
>> -------------
>> This errata is currently posted as "Reported". If necessary, please=20
>> use "Reply All" to discuss whether it should be verified or rejected. =

>> When a decision is reached, the verifying party (IESG) can log in to=20
>> change the status and edit the report, if necessary.
>>
>> --------------------------------------
>> RFC5477 (draft-ietf-psamp-info-11)
>> --------------------------------------
>> Title               : Information Model for Packet Sampling Exports
>> Publication Date    : March 2009
>> Author(s)           : T. Dietz, B. Claise, P. Aitken, F. Dressler, G.
> Carle
>> Category            : PROPOSED STANDARD
>> Source              : Packet Sampling
>> Area                : Operations and Management
>> Stream              : IETF
>> Verifying Party     : IESG
>=20
>=20
> --
> Paul Aitken
> Cisco Systems Ltd, Edinburgh, Scotland.
> _______________________________________________
> IPFIX mailing list
> IPFIX@ietf.org
> https://www.ietf.org/mailman/listinfo/ipfix

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms040408060408050504000304
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms040408060408050504000304--

From paitken@cisco.com  Mon Mar  1 02:34:19 2010
Return-Path: <paitken@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 41AE53A8B35 for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 02:34:19 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id e+R2a4NBbASK for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 02:34:18 -0800 (PST)
Received: from ams-iport-2.cisco.com (ams-iport-2.cisco.com [144.254.224.141]) by core3.amsl.com (Postfix) with ESMTP id 195993A69BE for <ipfix@ietf.org>; Mon,  1 Mar 2010 02:34:17 -0800 (PST)
Authentication-Results: ams-iport-2.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AtQAALEmi0uQ/uCWe2dsb2JhbACbDhUBARYkBhyIG5lGl0eEewQ
X-IronPort-AV: E=Sophos;i="4.49,559,1262563200";  d="scan'208";a="3877089"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-2.cisco.com with ESMTP; 01 Mar 2010 10:01:47 +0000
Received: from [144.254.153.37] (dhcp-144-254-153-37.cisco.com [144.254.153.37]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o21AYHUx023612; Mon, 1 Mar 2010 10:34:17 GMT
Message-ID: <4B8B9829.5030505@cisco.com>
Date: Mon, 01 Mar 2010 10:34:17 +0000
From: Paul Aitken <paitken@cisco.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-GB; rv:1.8.1.22) Gecko/20090605 SeaMonkey/1.1.17 (Ubuntu-1.1.17+nobinonly-0ubuntu0.9.04.1)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <EDC652A26FB23C4EB6384A4584434A0401F8CBE4@307622ANEX5.global.avaya.com> <4B8B7F85.8010205@net.in.tum.de>
In-Reply-To: <4B8B7F85.8010205@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] [Sender: ipfix-bounces@ietf.org] Re: FW: [Technical Errata Reported] RFC5477 (2052)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Mar 2010 10:34:19 -0000

Gerhard,

> If we accept errata 2052 [1], we do not need erratas 2053 to 2059 [2] 
> because the examples do not show any reduced size encoding.
> 
> [1] http://www.rfc-editor.org/errata_search.php?rfc=5477&eid=2052
> [2] http://www.rfc-editor.org/errata_search.php?rfc=5476

Yes we do. As you can see by reading further down the mail, I had a 
brain fart and typed unsigned32 instead of unsigned64.

I've asked RFC editor to fix the errata.


The u64 plan was discussed on the mailing list last year:

http://www.ietf.org/mail-archive/web/ipfix/current/msg05134.html
http://www.ietf.org/mail-archive/web/ipfix/current/msg05135.html
http://www.ietf.org/mail-archive/web/ipfix/current/msg05136.html


Cheers,
P.

-- 
Paul Aitken
Cisco Systems Ltd, Edinburgh, Scotland.

From muenz@net.in.tum.de  Mon Mar  1 04:42:50 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 0FB9D28C2DA for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 04:42:50 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id pYlssa+vW1Fs for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 04:42:49 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id B1B7628C0DD for <ipfix@ietf.org>; Mon,  1 Mar 2010 04:42:48 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 4E6FC2243A84; Mon,  1 Mar 2010 13:42:35 +0100 (CET)
Message-ID: <4B8BB644.1020101@net.in.tum.de>
Date: Mon, 01 Mar 2010 13:42:44 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Paul Aitken <paitken@cisco.com>
References: <EDC652A26FB23C4EB6384A4584434A0401F8CBE4@307622ANEX5.global.avaya.com> <4B8B7F85.8010205@net.in.tum.de> <4B8B9829.5030505@cisco.com>
In-Reply-To: <4B8B9829.5030505@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms090008090809080408030504"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] [Sender: ipfix-bounces@ietf.org] Re: FW: [Technical Errata Reported] RFC5477 (2052)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Mar 2010 12:42:50 -0000

This is a cryptographically signed message in MIME format.

--------------ms090008090809080408030504
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Paul,

Paul Aitken wrote:
> Gerhard,
>=20
>> If we accept errata 2052 [1], we do not need erratas 2053 to 2059 [2] =

>> because the examples do not show any reduced size encoding.
>>
>> [1] http://www.rfc-editor.org/errata_search.php?rfc=3D5477&eid=3D2052
>> [2] http://www.rfc-editor.org/errata_search.php?rfc=3D5476
>=20
> Yes we do. As you can see by reading further down the mail, I had a=20
> brain fart and typed unsigned32 instead of unsigned64.

Yes, sorry, I only your mail after sending my comment to the list.

> I've asked RFC editor to fix the errata.
>=20
>=20
> The u64 plan was discussed on the mailing list last year:
>=20
> http://www.ietf.org/mail-archive/web/ipfix/current/msg05134.html
> http://www.ietf.org/mail-archive/web/ipfix/current/msg05135.html
> http://www.ietf.org/mail-archive/web/ipfix/current/msg05136.html

I know that it was discussed, there was no final conclusion.

Maybe you want to respond to my recent mail:
http://www.ietf.org/mail-archive/web/ipfix/current/msg05311.html

Regards,
Gerhard


--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms090008090809080408030504
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms090008090809080408030504--

From trammell@tik.ee.ethz.ch  Mon Mar  1 04:46:42 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id D4D9628C2E8 for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 04:46:42 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.599
X-Spam-Level: 
X-Spam-Status: No, score=-6.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 9yz6aO7rX9Ip for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 04:46:41 -0800 (PST)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id AEF963A6922 for <ipfix@ietf.org>; Mon,  1 Mar 2010 04:46:41 -0800 (PST)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id 7C85BD9376 for <ipfix@ietf.org>; Mon,  1 Mar 2010 13:46:40 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id vHpM60tNqSka for <ipfix@ietf.org>; Mon,  1 Mar 2010 13:46:40 +0100 (MET)
Received: from [10.0.1.11] (84-75-167-74.dclient.hispeed.ch [84.75.167.74]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id F1A87D934A for <ipfix@ietf.org>; Mon,  1 Mar 2010 13:46:39 +0100 (MET)
From: Brian Trammell <trammell@tik.ee.ethz.ch>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: quoted-printable
Date: Mon, 1 Mar 2010 13:46:39 +0100
Message-Id: <3146E35B-9877-44AE-B884-17541E6ED5B8@tik.ee.ethz.ch>
To: ipfix@ietf.org
Mime-Version: 1.0 (Apple Message framework v1077)
X-Mailer: Apple Mail (2.1077)
Subject: [IPFIX] WGLC comments on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Mar 2010 12:46:42 -0000

Greetings, all,

I have reviewed draft-ietf-ipfix-configuration-model-04 in response to =
the WGLC, and have the following comments, ranked in order of impact; =
most of these deal with the applicability of the configuration data =
model to implementations other than the "exemplar" implementation(s) the =
authors had in mind while while writing.  I will say that overall, the =
draft does an more than reasonable job of being generally applicable. I =
apologize for starting a new thread, but I have not yet had time to =
properly review other comments, but plan on being in Anaheim prepared to =
discuss those too. I have also explicitly not reviewed the YANG =
definition; I leave that to people with actual YANG expertise.

First and foremost is the issue of caches, as it is in cache =
implementation that IPFIX packet-to-flow MPs will differentiate the most =
from each other, and indeed most vendors should be expected to subclass =
this cache class for their own configuration parameters; therefore =
getting this right is probably the most crucial. cacheMode immediate =
seems like a bit of a hack, as it invalidates basically every other =
cache field. However, I suppose I can see it as a way to keep the class =
design clean (otherwise, there would be nothing to connect OP to =
layout/EP in a PSAMP implementation...).=20

1. I'm concerned about the definition of cacheMode timeout (4.3, p. 24), =
as it seems to make it impossible to configure a cache to allow cache =
exit on TCP state tracking. Perhaps this is merely an editorial =
oversight?

2. Destination (4.4.1, pp .27-29) appears not to match the properties of =
the particular transports:

  a. can't specify a local IP address for TCP or UDP. This seems =
necessary on devices with multiple management interfaces, which I don't =
think is all that weird a deployment plan.

  b. can't specify 1..n remote IP addresses for SCTP. This breaks one of =
the advantages of SCTP.

  c. can't configure SCTP reliability per-template, as opposed to per =
destination. see 2.b.

3. Options (4.4.3, p. 30) configuration is rather odd and highly =
restrictive, even to the point of being unusable. The biggest problem is =
that RFC5473 common properties are something completely different from =
the options specified in 5101 and 5476. "Enable common properties", =
without some definition of which fields to use as common properties and =
under what circumstances, is not a useful directive. Secondarily, =
options specified in 5610 and 5655 are not covered; do we really want to =
amend this document every time we specify another WG-blessed options =
template? I'd suggest using an OptionsTemplate subclass of Template, =
instead, and providing the definition of common (non-5473) options =
templates in an appendix to the configuration data model.

4. There does not seem to be any room for logical configurable processes =
between selection and cache, or cache and EP. The example that comes to =
mind is perhaps implementation-specific [I'm mentioning this because yaf =
does it]: fragmentation behavior (reassemble, ignore, drop). How could I =
allow then for the configuration of the fragment table (given that a =
selection process can only reference a cache)?=20

5. CacheLayout (4.3.1, p. 25) does not appear to be able to support =
adaptive reduced-length export (exporting, say, counters with unsigned32 =
unless one of them actually requires unsigned64) [yes, yaf does this =
too]. Though this could probably be solved with a deviation, and an =
additional vendor-specific CacheField attribute...

6. CacheLayout (4.3.1, p. 25) could do with an explicit example to =
illustrate the "unless the corresponding IE is not applicable or cannot =
be derived" case: I presume this is intended for dual-stack IPv4/IPv6 =
measurement, RFC5103 biflows, and the like? It does appear than this =
clause is sufficient for such cases, but I had to read this a couple of =
times before I understood that's what was meant.

7. Real File Readers and File Writers will probably need more than a =
single filename endpoint on the file system (however, most file readers =
and file writers in document-based workflows won't be configured using =
this information model...) How to handle this is probably =
implementation-specific; in any case, I don't have a good suggestion =
here, yet; perhaps we can discuss in Anaheim.

8. Is there any benefit to cascading caches as with selection processes, =
or rather, are the benefits of cascading caches considered in scope for =
this application? The simplest example would be simultaneous export of =
flow data as well as flow aggregate data (e.g. traffic counters per AS =
pair), perhaps as part of a cascading-aggregation mediator. I ask only =
because it seems a rather small and easily supportable change to greatly =
increase flexibility.

9. Destination exportMemberType appears to reference a renamed MIB =
object as of MIB-10; this is called ipfixExportMemberType now. I would =
suggest copying the complete description from MIB, because the semantics =
are kind of confusing otherwise (I had to read MIB, which I probably =
should have done by now but oh well, in order to figure out what the =
difference between parallel and primary was). A tertiary problem: these =
exportMemberTypes seem a bit restrictive (e.g. what if I want to do =
parallel export and also have secondaries?), but I suppose that's a MIB =
comment.=20

10. draft-ietf-ipfix-file is now RFC5655 (the editor will catch this); =
since this is now an RFC, references to things "proposed" in this draft =
should probably be changed to things "specified" in RFC 5655.

That's about it...

Best regards,

Brian=

From paitken@cisco.com  Mon Mar  1 05:47:52 2010
Return-Path: <paitken@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 6C18228C2EF for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 05:47:52 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id UCoMs5olkuMc for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 05:47:51 -0800 (PST)
Received: from ams-iport-2.cisco.com (ams-iport-2.cisco.com [144.254.224.141]) by core3.amsl.com (Postfix) with ESMTP id 8F31B28C307 for <ipfix@ietf.org>; Mon,  1 Mar 2010 05:47:50 -0800 (PST)
Authentication-Results: ams-iport-2.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AtQAAF5Ui0uQ/uCWe2dsb2JhbACbDxUBARYkBhyhRJdOAoJTgiYE
X-IronPort-AV: E=Sophos;i="4.49,560,1262563200";  d="scan'208";a="3884207"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-2.cisco.com with ESMTP; 01 Mar 2010 13:15:19 +0000
Received: from [144.254.153.37] (dhcp-144-254-153-37.cisco.com [144.254.153.37]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o21Dlnen005822; Mon, 1 Mar 2010 13:47:49 GMT
Message-ID: <4B8BC585.3030503@cisco.com>
Date: Mon, 01 Mar 2010 13:47:49 +0000
From: Paul Aitken <paitken@cisco.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-GB; rv:1.8.1.22) Gecko/20090605 SeaMonkey/1.1.17 (Ubuntu-1.1.17+nobinonly-0ubuntu0.9.04.1)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <4AD6E120.50002@cisco.com> <4AD6EAF4.9040303@net.in.tum.de>	<FB57D5CC-A231-48DD-9935-CF21754F9E26@tik.ee.ethz.ch> <EDC652A26FB23C4EB6384A4584434A0401B179BA@307622ANEX5.global.avaya.com> <4B870A49.6000608@cisco.com> <4B879C38.4080100@net.in.tum.de>
In-Reply-To: <4B879C38.4080100@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 8bit
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Mistake in PSAMP RFCs: Selector ID length
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Mar 2010 13:47:52 -0000

Gerhard,

Clearly changing all the identifiers to 64 bits is a large exercise.

With the benefit of hindsight, the data model should only specify the 
data types (signed, unsigned) and leave field lengths to be defined in 
the templates.

But now that we have a data model, I'm not convinced that changing all 
these sizes is beneficial or even necessary.

I'd be interested to hear what others think.

Cheers,
P.

> Dear all,
> 
> IMO, the distinction between "protocol-level" and other identifiers is 
> quite arbitrary.
> 
> Most identifiers are currently unsigned32:
> http://www.iana.org/assignments/ipfix/ipfix.xhtml
> 
> If we change selectorId to unsigned32, it would be consistent other 
> identifiers and with IPFIX-MIB, where we have:
> ipfixSelectionProcessSelectorIndex    Unsigned32
> 
> However, if you want to change it to unsigned64, why shouldn't we change 
> observationPointId etc. as well?
> And what about ipfixSelectionProcessSelectorIndex?
> The IPFIX-MIB does not say that
> ipfixSelectionProcessSelectorIndex == selectorId,
> yet this would be an obvious implementation.
> 
> These are just some thoughts about possible consequences, which does not 
> mean that I do not like unsigned64 in general.
> 
> Regards,
> Gerhard
> 
> 
> Paul Aitken wrote:
>> Dan,
>>
>> As you will have seen, I just opened errata 2052 - 2059 for this.
>>
>> Thanks,
>> P.
>>
>>
>>> It looks like two errata on RFC 5478 and 5477 should be issues.
>>> I suggest that one of you takes the pen to enter the errata texts,
>>> and when they are ready and have reached consensus I can approve them.
>>> Dan
>>>  
>>>
>>>> -----Original Message-----
>>>> From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] On 
>>>> Behalf Of Brian Trammell
>>>> Sent: Thursday, October 15, 2009 11:41 AM
>>>> To: Gerhard Muenz
>>>> Cc: psamp; ipfix@ietf.org
>>>> Subject: Re: [IPFIX] Mistake in PSAMP RFCs: Selector ID length
>>>>
>>>> hi Benoit, Gerhard,
>>>>
>>>> I agree with the change to 64 bit, to keep it in line with every 
>>>> other protocol-level identifier (save odID, bound by the header 
>>>> format to 32 bits), with the understanding that RLE will often be 
>>>> used to use it as
>>>> 32 bit.
>>>>
>>>> However, in this case we might want as well to update 5476 with a 
>>>> line noting that all the examples use RLE.
>>>>
>>>> Best regards,
>>>>
>>>> Brian
>>>>
>>>> On Oct 15, 2009, at 11:27 AM, Gerhard Muenz wrote:
>>>>
>>>>> Benoit,
>>>>>
>>>>> Benoit Claise wrote:
>>>>>> Dear all,
>>>>>> We found a mistake in the PSAMP RFCs.
>>>>>> RFC5477
>>>>>>            8.1.2. selectorId
>>>>>>       Description:
>>>>>>          The Selector ID is the unique ID identifying a Primitive 
>>>>>> Selector.
>>>>>>          Each Primitive Selector must have a unique ID in the 
>>>>>> Observation
>>>>>>          Domain.
>>>>>>       Abstract Data Type:  unsigned16                   
>>>>>> <------------------
>>>>>>       Data Type Semantics:  identifier
>>>>>>       ElementId:  302
>>>>>>       Status:  current
>>>>>>      RFC5476, the selectorID length is 4 bytes in ALL examples 
>>>>>> (and there are many).
>>>>>>
>>>>>>   IPFIX Options Template Record:
>>>>>>       0                   1                   2                   3
>>>>>>       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
>>>>>>      
>>>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>>>      |          Set ID = 3           |          Length = 
>>>>>> 26          |
>>>>>>      
>>>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>>>      |        Template ID = 262      |         Field Count = 
>>>>>> 4       |
>>>>>>      
>>>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>>>      |     Scope Field Count =  1    |0|  selectionSequenceId = 
>>>>>> 301  |
>>>>>>      
>>>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>>>      |       Scope 1 Length = 4      |0|     ingressInterface = 
>>>>>> 10   |
>>>>>>      
>>>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>>>      |        Field Length = 4       |0|      selectorId = 
>>>>>> 302       |
>>>>>>      
>>>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>>>      |        Field Length = 4       |0|      selectorId = 
>>>>>> 302       |
>>>>>>      
>>>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>>>      |        Field Length = 4       |
>>>>>>      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>>>
>>>>>> The proposed solution is to increase the Selector ID length and 
>>>>>> put an ERRATA on RFC5477.
>>>>>> 16 bits is definitely not enough and clearly not was intended.  
>>>>>> Since we can do reduced length encoding with larger identifiers, 
>>>>>> we don't see any harm in making all identifiers 64 bit Feedback?
>>>>> 4 Bytes = 32 Bit
>>>>>
>>>>> You propose to increase it to 64 (=8 Byte), which means that both 
>>>>> RFCs need to be corrected. Or is 64 a mistake?
>>>>>
>>>>> Regards,
>>>>> Gerhard
>>>>>
>>>>>> Regards, Benoit.
>>>>>> ------------------------------------------------------------------------ 
>>>>>>
>>>>>> _______________________________________________
>>>>>> IPFIX mailing list
>>>>>> IPFIX@ietf.org
>>>>>> https://www.ietf.org/mailman/listinfo/ipfix
>>>>> -- 
>>>>> Dipl.-Ing. Gerhard Münz
>>>>> Chair for Network Architectures and Services (I8) Technische 
>>>>> Universität München - Department of Informatics Boltzmannstr. 3, 
>>>>> 85748 Garching bei München, Germany
>>>>> Phone:  +49 89 289-18008       Fax: +49 89 289-18033
>>>>> E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz
>>>>>
>>>>> _______________________________________________
>>>>> IPFIX mailing list
>>>>> IPFIX@ietf.org
>>>>> https://www.ietf.org/mailman/listinfo/ipfix
>>>> _______________________________________________
>>>> IPFIX mailing list
>>>> IPFIX@ietf.org
>>>> https://www.ietf.org/mailman/listinfo/ipfix
>>>>
>>> _______________________________________________
>>> IPFIX mailing list
>>> IPFIX@ietf.org
>>> https://www.ietf.org/mailman/listinfo/ipfix
>>
>>
> 


-- 
Paul Aitken
Cisco Systems Ltd, Edinburgh, Scotland.

From paitken@cisco.com  Mon Mar  1 05:50:22 2010
Return-Path: <paitken@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id A6B4628C2EF for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 05:50:22 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.599
X-Spam-Level: 
X-Spam-Status: No, score=-6.599 tagged_above=-999 required=5 tests=[AWL=4.000,  BAYES_00=-2.599, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id BlPuD2wh0DGE for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 05:50:21 -0800 (PST)
Received: from ams-iport-1.cisco.com (ams-iport-1.cisco.com [144.254.224.140]) by core3.amsl.com (Postfix) with ESMTP id 8BED428C321 for <ipfix@ietf.org>; Mon,  1 Mar 2010 05:50:18 -0800 (PST)
Authentication-Results: ams-iport-1.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AtQAAJtUi0uQ/uCWe2dsb2JhbACbDxUBARYkBhyIG5ktl06EewQ
X-IronPort-AV: E=Sophos;i="4.49,560,1262563200"; d="scan'208";a="57606333"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 01 Mar 2010 13:50:17 +0000
Received: from [144.254.153.37] (dhcp-144-254-153-37.cisco.com [144.254.153.37]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o21DoHqV006537; Mon, 1 Mar 2010 13:50:17 GMT
Message-ID: <4B8BC619.8060806@cisco.com>
Date: Mon, 01 Mar 2010 13:50:17 +0000
From: Paul Aitken <paitken@cisco.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-GB; rv:1.8.1.22) Gecko/20090605 SeaMonkey/1.1.17 (Ubuntu-1.1.17+nobinonly-0ubuntu0.9.04.1)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <EDC652A26FB23C4EB6384A4584434A0401F8CBE4@307622ANEX5.global.avaya.com>	<4B8B7F85.8010205@net.in.tum.de> <4B8B9829.5030505@cisco.com>
In-Reply-To: <4B8B9829.5030505@cisco.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] [Sender: ipfix-bounces@ietf.org] Re: [Sender: ipfix-bounces@ietf.org] Re: FW: [Technical Errata Reported] RFC5477 (2052)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Mar 2010 13:50:22 -0000

Gerhard,

> As you can see by reading further down the mail, I had a 
> brain fart and typed unsigned32 instead of unsigned64.
> 
> I've asked RFC editor to fix the errata.

RFC editor has now fixed the errata:

http://www.rfc-editor.org/errata_search.php?rfc=5477&eid=2052


Section 8.1.2 says:

   Abstract Data Type:  unsigned16

It should say:

   Abstract Data Type:  unsigned64


Cheers,
P.

-- 
Paul Aitken
Cisco Systems Ltd, Edinburgh, Scotland.

From muenz@net.in.tum.de  Mon Mar  1 06:38:44 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 844EE28C375 for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 06:38:44 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id qTO7blufAgt5 for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 06:38:42 -0800 (PST)
Received: from mail-out2.informatik.tu-muenchen.de (mail-out2.informatik.tu-muenchen.de [131.159.0.36]) by core3.amsl.com (Postfix) with ESMTP id C913028C35E for <ipfix@ietf.org>; Mon,  1 Mar 2010 06:38:41 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id A50352243A84; Mon,  1 Mar 2010 15:38:39 +0100 (CET)
Message-ID: <4B8BD178.8030802@net.in.tum.de>
Date: Mon, 01 Mar 2010 15:38:48 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Andrew Johnson <andrjohn@cisco.com>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com>
In-Reply-To: <4B7EB360.8060902@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms020901030401040107070305"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Mar 2010 14:38:44 -0000

This is a cryptographically signed message in MIME format.

--------------ms020901030401040107070305
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Andrew,

Thanks a lot for your thorough review.

Please find answers and comments inline.

Regards,
Gerhard


Andrew Johnson wrote:
> One general comment:
>=20
>   Why do only some of the UML fields have settings given>
>   e.g. "direction =3D both"

The meaning of "=3D X" is explained in Section 3.1:

    A default value is indicated after the attribute behind the equality
    symbol ("=3D").  It means that this default value MUST be used by the=

    Monitoring Device if the parameter is not configured by the user.

However, nothing is written about what happens if "=3D X" is not present.=


So, I have rewritten the first paragraphs of the section to clarify the=20
meaning of default values, multiplicity indicators, and attribute=20
properties, as well as their mapping to YANG

Please let me know if the text is ok and answers your question.

NEW:

3.1.  UML Representation

    We use UML class diagrams [UML] to explain the structure of the
    configuration data model.  The attributes of the classes are the
    configuration or state parameters of the Monitoring Device.


     +------------------------------------------------+
     | Destination                                    |
     +------------------------------------------------+
     | name                                           |<>-----+
     | exportMemberType =3D parallel                    |       | 0..1
     | ipfixVersion =3D 10                              |       |
     | transportProtocol                              |   +------------+
     | sourceIpAddress[0..*] {[0..1] for UDP and TCP} |   | Transport- |
     | destinationIpAddress                           |   | Layer-     |
     | destinationPort =3D 4739|4740                    |   | Security   =
|
     | ifName/ifIndex[0..1]                           |   +------------+
     | sendBufferSize {opt.}                          |
     | rateLimit[0..1]                                |
     | timedReliability =3D 0 {SCTP only}               |
     | numberOfStreams {opt.} {SCTP only}             |
     | templateRefreshTimeout =3D 600 {UDP only}        |
     | optionsTemplateRefreshTimeout =3D 600 {UDP only} |
     | templateRefreshPacket[0..1] {UDP only}         |
     | optionsTemplateRefreshPacket[0..1] {UDP only}  |
     +------------------------------------------------+

                  Figure 2: UML example: Destination class

    As an example, Figure 2 shows the UML diagram of the Destination
    class, which is specified in more detail in Section 4.4.1.  The upper=

    box contains the name of the class.  The lower box lists the
    attributes of the class.  Each attribute corresponds to a parameter
    of the configuration data model.

    Behind an attribute's name, there may appear a multiplicity indicator=

    in brackets (i.e., between "[" and "]").  An attribute with a
    multiplicity indicator of "[0..1]" represents an OPTIONAL
    configuration parameter which is only included in the configuration
    data if the user configures it.  Typically, the absence of an
    OPTIONAL parameter has a specific meaning.  For example, not
    configuring rateLimit in the Destination class means that no rate
    limiting will be applied to the exported data.  In YANG, an OPTIONAL
    parameter is specified as a "leaf" without "mandatory true"
    substatement.  The "description" substatement specifies the behavior
    for the case that the parameter is not configured.

    A multiplicity indicator of "[0..*]" means that this parameter MAY be=

    configured multiple times with different values.  In the example,
    multiple source IP addresses may be configured for a multi-homed
    Exporting Process if SCTP is transport protocol.  In YANG, an
    attribute with multiplicity indicator "[0..*]" corresponds to a
    "leaf-list".

    Attributes without multiplicity indicator may be endued with a
    default value which is indicated behind the equality symbol ("=3D").
    If a default value exists, the parameter does not have to be
    explicitly configured by the user.  If the parameter is not
    configured by the user, the Monitoring Device MUST use the given
    default value for the given parameter.  In the example, IPFIX version=

    10 must be used unless a different value is configured for
    ipfixVersion.  In YANG, an attribute with default value corresponds
    to a "leaf" with "default" substatement.

    In the example, there exist two default values for the destination
    port, namely the registered ports for IPFIX with and without
    transport layer security (i.e., DTLS or TLS), which are 4740 and
    4739, respectively.  In the UML diagram, the two default values are
    separated by a vertical bar ("|").  In YANG, such alternative default=

    values cannot be specified formally.  Instead, they are defined in
    the "description" substatement of the "leaf".

    Further attribute properties are denoted in braces (i.e., between "{"=

    and "}").  An attribute with property "{opt.}", such as
    sendBufferSize in the Destination class, represents a parameter that
    MAY be configured by the user.  If not configured by the user, the
    Monitoring Device MUST assign an appropriate value for this
    parameter.  As a result, the parameter will always exist in the
    configuration data, yet it is not mandatory for the user to configure=

    it.  In YANG, this behavior cannot be specified formally.  Therefore,=

    it is specified in the "description" substatement of the "leaf".

    The availability of a parameter may depend on another parameter
    value.  In the UML diagram, such restrictions are indicated as
    attribute properties.  In the example, the parameter timedReliability=

    can only be configured if the transport protocol is SCTP ("{SCTP
    only}").  In YANG, such restrictions are formalized in the "when"
    substatement of the "leaf".

    If the multiplicity of a parameter changes depending on another
    parameter value, this is also denoted as an attribute property.  In
    the example, the parameter sourceIpAddress may be configured only
    once if the transport protocol is UDP or TCP ("{[0..1] for UDP and
    TCP}").  In YANG, this is formalized in the "must" substatement of
    the "leaf".

    Another attribute property not shown in the example is "{readOnly}"
    specifying state parameters which cannot be configured.  In YANG,
    this corresponds to the "config false" substatement.

    Attributes without multiplicity indicator, without default value, and=

    without "{readOnly}" property are mandatory configuration parameters.=

    These parameters MUST be configured by the user unless an attribute
    property determines that the parameter is not available.  In YANG, a
    mandatory parameter corresponds to a "leaf" with "mandatory true"
    substatement.  In the example, the user MUST configure the parameters=

    name, transportProtocol, and destinationIpAddress.  The remaining
    parameters have either a default value, the "{opt.}" property, or a
    multiplicity indicator starting with zero.


> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> Section 3:
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> If a Selection Process is part of a Metering Process and a single
> Selection Process can feed into more than one Cache, then it is
> implied that a single Metering Process can have more than once cache.

Note: The text has slightly changed after Atsushi's review. Now, it=20
explicitly says that a Metering Process may contain multiple Selection=20
Processes and multiple Caches.

Furthermore, it says that only packets observed at a single Observation=20
Point may enter a Selection Process. IMO, this results from the PSAMP=20
definitions of Observed Packet Stream and Selection Process. Please tell =

me if this conclusion is wrong.

This is how the text reads now:


                +-----------------------------------+
                | Metering Process                  |
                | +-----------+ Selected            |
       Observed | | Selection | Packet    +-------+ |  Stream of
       Packet  -->| Process   |---------->| Cache |--> Flow Records or
       Stream   | +-----------+ Stream    +-------+ |  Packet Reports
                +-----------------------------------+

      Figure 1: Selection Process and Cache forming a Metering Process

    The configuration data model adopts the separation of Selection
    Processes and Caches in order to support the flexible configuration
    and combination of these functional blocks.  As defined in [RFC5476],=

    the Selection Process takes the Observed Packet Stream as its input
    and selects a subset of that stream as its output (Selected Packet
    Stream).  The Observed Packet Stream at the input of a Selection
    Process is composed of packets from a single Observation Point.  The
    action of the Selection Process on a single packet of its input is
    defined by a Primitive Selector or a Composite Selector.  The Cache
    generates Flow Records or Packet Reports from the Selected Packet
    Stream, depending on the configured Cache Mode.

    Beyond the basic Metering Process shown in Figure 1, the
    configuration data model enables the specification of more complex
    configurations.  A Metering Process may be composed of multiple
    Selection Processes and multiple Caches.  A single Cache may process
    the output of multiple Selection Processes.  The Selected Packet
    Stream may be copied to enter several Caches, for example one Cache
    that generates Flow Records and another Cache that generates Packet
    Reports.  Selection Processes can also be cascaded, such that the
    Selected Packet Stream at the output of one Selection Process is
    passed to another Selection Process.  Cascading Selection Processes
    can be useful in specific contexts such as the example in
    Section 7.3.

> I can see why it would be convenient to configure multiple caches to
> be populated from a single Selection Process, and it could even be
> implemented that way (to some extent), but that is not the PSAMP model.=

> The configuration model doesn't have to exactly match this model, but
> you should ensure that the differences are made clear.

The PSAMP model has no notion of Caches.

Could you please show me the paragraphs in the PSAMP documents which=20
would indirectly prevent the existence of more than one Cache per=20
Metering Process?


> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> From section 3.1:
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> Why have so many SCTP or UDP only options?  Can't you use UML
> inheritance to do something like:
>=20
>     +------------------------------------------------+
>     | Destination                                    |
>     +------------------------------------------------+
>     | name                                           |
>     | exportMemberType =3D parallel                    |
>     | ipfixVersion =3D 10                              |
>     | transportProtocol                              |
>     | destinationIpAddress                           |
>     | destinationPort =3D 4739|4740                    |
>     | sendBufferSize {opt.}                          |
>     | rateLimit[0..1]                                |
>     +------------------------------------------------+
>           ^
>           | 1
>      +----+-------------------+
>      |                        |
>      |    +------------------------------------------------+
>      |    | SCTP Destination                               |
>      |    +------------------------------------------------+
>      |    | localIpAddress[0..*] {SCTP only}               |<>-----+	  =
    =20
>      |    | timedReliability =3D 0 {SCTP only}               |       | =
0..1   =20
>      |    | numberOfStreams {opt.} {SCTP only}             |       |	  =
    =20
>      |    | sourceIpAddress[0..1] {UDP only}               |   +-------=
-----+
>      |    | templateRefreshTimeout =3D 600 {UDP only}        |   | Tran=
sport- |
>      |    | optionsTemplateRefreshTimeout =3D 600 {UDP only} |   | Laye=
r-     |
>      |    | templateRefreshPacket[0..1] {UDP only}         |   | Securi=
ty   |
>      |    | optionsTemplateRefreshPacket[0..1] {UDP only}  |   +-------=
-----+
>      |    +------------------------------------------------+
>      |
>      +---------------------+
>                            |
>     +------------------------------------------------+
>     | UDP Destination                                |
>     +------------------------------------------------+
>     | sourceIpAddress[0..1] {UDP only}               |<>-----+	      =20
>     | templateRefreshTimeout =3D 600 {UDP only}        |       | 0..1  =
 =20
>     | optionsTemplateRefreshTimeout =3D 600 {UDP only} |       |	      =
=20
>     | templateRefreshPacket[0..1] {UDP only}         |   +------------+=

>     | optionsTemplateRefreshPacket[0..1] {UDP only}  |   | Transport- |=

>     +------------------------------------------------+   | Layer-     |=

> 						           | Security   |
>                                                          +------------+=


This looks nice. Nevertheless, there are some drawbacks:

Since YANG does not support inheritance, it cannot be mapped to YANG=20
directly.

Furthermore, I would have to explain even more UML symbols and stuff in=20
the document.

Note that the Destination class might be changed as proposed in
http://www.ietf.org/mail-archive/web/ipfix/current/msg05316.html

> Are there no options that are used with TCP?

You mean parameters which are available for TCP only?
I do not know any. If you have some, please let me know.

> Are the security parameters the same across DTLS and TLS?

There was some discussion on the mailing list whether we should add any=20
(D)TLS parameters to the IPFIX configuration data model.

The conclusion was that only parameters regarding the mutual=20
authentication of Exporters and Collectors should be added. Further=20
parameters are too specific.

Regarding authentication, the parameters are the same for DTLS and TLS.

> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> From section 4.1:
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> Doesn't the direction belong in the Observation Point class, not the
> Interface or LC class.

If the current position of the direction parameter bothers you, we can=20
move it.

However, my reasoning is the following:
- If it appears in the ObservationPoint class, it looks like the=20
Observation Point is doing some filtering.
- If it appears in the Interface/Linecard class, it looks like the=20
filtering is already done at the interface or linecard.

I thought that the interface/linecard is doing the filtering.

There is another small advantage: The current position of the parameter=20
makes it easier to add vendor or implementation-specific packet sources=20
for which the direction parameter has no meaning, for example:

+------------------+
| PcapFile         |
+------------------+
| filename         |
+------------------+

> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> From Section 4.2.1
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> Why packetsDropped and not packetsSelected, when PSAMP talks about
> packetsSelected in the stats it needs to export?


All state parameters are adopted from the IPFIX-MIB. There, we have:

+--ipfixSelectionProcessStatsTable(4)
    |
    +--ipfixSelectionProcessStatsEntry(1)=20
[ipfixMeteringProcessCacheId,ipfixSelectionProcessIndex,ipfixSelectionPro=
cessSelectorIndex]
       |
       +-- r-n Counter64 ipfixSelectionProcessStatsPacketsObserved(1)
       +-- r-n Counter64 ipfixSelectionProcessStatsPacketsDropped(2)
       +-- r-n TimeStamp ipfixSelectionProcessStatsDiscontinuityTime(3)


Probably, ipfixSelectionProcessStatsPacketsSelected would make more=20
sense in the MIB.

I assume that the MIB standardization cannot be stopped any more, right?

So, what do you suggest?

> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> From Section 4.3.1
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> Isn't is possible to generate a cache without any keys?
> This would generate a cache with a single flow?  Do we allow this?

A the moment, there is no restriction which says that a flow key field=20
must exist.

Do you want to add such a restriction?

> Would it be useful to have two lists, one for keys and one for
> non-keys?

There is an argument for having a single list:

A Cache implementation may generate records using the specified field=20
order. So, the fields occur in the same order as specified in the=20
configuration.

With separate lists for flow key fields and non-key fields, we cannot=20
arbitrarily mix these two types of fields.

Is there any advantage of having separate lists?

> For a permanent cache, update counters should be used and for a
> timeout cache the delta counters should be used.  Where is this
> sort of thing enforced?

This is not enforced. I would not strictly enforce it because strange=20
configurations might make sense in specific use cases.

However, we could add a note that totalCounters are typically used for=20
"permanent" and deltaCounters for "timeout".

Could you provide a sentence or two?

> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> From Section 4.4.1
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>    exportMemberType:  Configures the export member type that correspond=
s
>       to the ipfixTransportSessionGroupMemberType object in
>       [I-D.ietf-ipfix-mib].  The following parameter values are
>       specified by the configuration data model:
>       *  primary: primary target of the Exporting Process
>       *  secondary: secondary target of the Exporting Process used when=

>          the primary target is not reachable
>       *  parallel: parallel exporting to all destinations and files of
>          the Exporting Process
>       *  loadBalancing: load-balancing between different destinations
>          and files of the Exporting Process
>       "parallel" is the default value if this parameter is not
>       configured.  If one destination or file is configured as "primary=
"
>=20
> -----------------------------------------------------------------------=
-
> Comments
> -----------------------------------------------------------------------=
-
> It's not clear to me that this parameter belongs to the Destination
> class.  The number of restrictions on when you can use the various
> types suggests that we could simplify things.
>=20
> A Cache can pass flow records to multiple Exporting Processes, so we
> don't really need a parallel option.  This would mean that each
> Exporting Process is responsible for sending one complete set of
> records.  i.e. one record in, one record out.  It could do that, either=

> using a redundancy model or with a load-balancing one.
>=20
> In other words, the exportType would be an Exporting Process attribute
> which would indicate whether the list of Destinations should be load-
> balanced across or treated as Primary, Seondary, Tertiary, etc.

At the moment, the parameter is inline with the IPFIX-MIB:

ipfixExportMemberType OBJECT-TYPE
     SYNTAX      INTEGER {
                     unknown(0),
                     primary(1),
                     secondary(2),
                     parallel(3),
                     loadBalancing(4)
                 }
     MAX-ACCESS  read-only
     STATUS      current
     DESCRIPTION
         "The type of a member Transport Session in a Transport
         Session group (identified by the value of ipfixExportIndex,
         ipfixObservationDomainId and ipfixMeteringProcessCacheId).
         The following values are valid:

         unknown(0)
             This value MUST be used if the status of the group
             membership cannot be detected by the equipment. This
             value should be avoided as far as possible.

         primary(1)
             This value is used for a group member that is used as
             the primary target of an Exporter. Other group members
             (with the same ipfixExportIndex and
             ipfixMeteringProcessCacheId) MUST NOT have the value
             primary(1) but MUST have the value secondary(2).
             This value MUST also be specified if the Exporter does
             not support Transport Session grouping. In this case
             the group contains only one Transport Session.

         secondary(2)
             This value is used for a group member that is used as a
             secondary target of an Exporter. The Exporter will use
             one of the targets specified as secondary(2) within the
             same Transport Session group when the primary target is
             not reachable.

         parallel(3)
             This value is used for a group member that is used for
             duplicate exporting i.e., all group members identified
             by the ipfixExportIndex are exporting the same Records
             in parallel. This implies that all group members MUST
             have the same membertype parallel(3).

         loadBalancing(4)
             This value is used for a group member that is used
             as one target for load-balancing. This means that a
             Record is sent to one of the group members in this
             group identified by ipfixExportIndex.
             This implies that all group members MUST have the same
             membertype loadBalancing(4)."
     ::=3D { ipfixExportEntry 2 }


I'm open to better proposals.

"parallel" and "loadBalancing" can be configured once with a single=20
ExportingProcess class parameter.

How would you configure "primary" and "secondary with a single Exporting =

Process parameter?


> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> From Section 4.7: TransportSession class
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> I think you want to add a timestamp for when the session came up
> and when it went down (unless it's still active).

Currently, there is no such parameter in ipfixTransportSessionTable.

Hm, it seems that we should have asked you to review the IPFIX-MIB :)

Shall we go beyond IPFIX-MIB and add such state parameters?

> There's no association with the relevant destination.  It seems to me
> that your really want this as part of each Destination or FileReader.
> Specifically, I think you want to separate the Transport Session from
> the File Reader stats.  Potentially there could be a one to many
> relationship of Destination and Transport Sessions, each with their own=

> timestamp for coming up and going down.

Currently, TransportSession is a mere copy of ipfixTransportSessionTable =

which exists for both, Exporters and Collectors.

In the case of the _Exporting Process_ , we could move the state=20
parameters to the _Destination_ and _FileWriter_ class.

I think that there can be at most one Transport Session per Destination=20
and per FileWriter object. At least at the same point in time.
However, it may happen that no Transport Session exists, for example in=20
the case of a secondary export destination.

So, we could move a single set of Transport Session state parameters to=20
the Destination and FileWriter classes. The parameters would appear or=20
not, depending on whether the Transport Session exists.

In the case of the _Collecting Process_ , we can also move the Transport =

Session state parameters to the _Receiver_ and _FileReader_ classes. As=20
you say, there can be multiple Transport Sessions per Receiver.

Probably, it makes sense to have a container=20
<transportSessionStatistics> which groups the state parameters of a=20
Transport Session. This container may appear multiple times in the case=20
of the Receiver class, and only once in all other classes.

Would you agree with these changes?


--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms020901030401040107070305
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature

MIAGCSqGSIb3DQEHAqCAMIACAQExCzAJBgUrDgMCGgUAMIAGCSqGSIb3DQEHAQAAoIIO5jCC
BEYwggOvoAMCAQICEGb9R+PCGeToms2Z3fU6yyQwDQYJKoZIhvcNAQEFBQAwXzELMAkGA1UE
BhMCVVMxFzAVBgNVBAoTDlZlcmlTaWduLCBJbmMuMTcwNQYDVQQLEy5DbGFzcyAxIFB1Ymxp
YyBQcmltYXJ5IENlcnRpZmljYXRpb24gQXV0aG9yaXR5MB4XDTA1MTAyODAwMDAwMFoXDTE1
MTAyNzIzNTk1OVowgd0xCzAJBgNVBAYTAlVTMRcwFQYDVQQKEw5WZXJpU2lnbiwgSW5jLjEf
MB0GA1UECxMWVmVyaVNpZ24gVHJ1c3QgTmV0d29yazE7MDkGA1UECxMyVGVybXMgb2YgdXNl
IGF0IGh0dHBzOi8vd3d3LnZlcmlzaWduLmNvbS9ycGEgKGMpMDUxHjAcBgNVBAsTFVBlcnNv
bmEgTm90IFZhbGlkYXRlZDE3MDUGA1UEAxMuVmVyaVNpZ24gQ2xhc3MgMSBJbmRpdmlkdWFs
IFN1YnNjcmliZXIgQ0EgLSBHMjCCASIwDQYJKoZIhvcNAQEBBQADggEPADCCAQoCggEBAMnf
rOfq+PgDFMQAktXBfjbCPO98chXLwKuMPRyVzm8eECw/AO2XJua2x+atQx0/pIdHR0w+VPhs
+Mf8sZ69MHC8l7EDBeqV8a1AxUR6SwWi8mD81zplYu//EHuiVrvFTnAt1qIfPO2wQuhejVch
rKaZ2RHp0hoHwHRHQgv8xTTq/ea6JNEdCBU3otdzzwFBL2OyOj++pRpu9MlKWz2VphW7NQIZ
+dTvvI8OcXZZu0u2Ptb8Whb01g6J8kn+bAztFenZiHWcec5gJ925rXXOL3OVekA6hXVJsLjf
aLyrzROChRFQo+A8C67AClPN1zBvhTJGG+RJEMJs4q8fef/btLUCAwEAAaOB/zCB/DASBgNV
HRMBAf8ECDAGAQH/AgEAMEQGA1UdIAQ9MDswOQYLYIZIAYb4RQEHFwEwKjAoBggrBgEFBQcC
ARYcaHR0cHM6Ly93d3cudmVyaXNpZ24uY29tL3JwYTALBgNVHQ8EBAMCAQYwEQYJYIZIAYb4
QgEBBAQDAgEGMC4GA1UdEQQnMCWkIzAhMR8wHQYDVQQDExZQcml2YXRlTGFiZWwzLTIwNDgt
MTU1MB0GA1UdDgQWBBQRfV4ZfTwE32ps1qKKGj8x2DuUUjAxBgNVHR8EKjAoMCagJKAihiBo
dHRwOi8vY3JsLnZlcmlzaWduLmNvbS9wY2ExLmNybDANBgkqhkiG9w0BAQUFAAOBgQA8o9oC
YzrEk6qrctPcrVA4HgyeFkqIt+7r2f8PjZWg1rv6aguuYYTYaEeJ70+ssh9JQZtJM3aTi55u
uUMcYL3C3Ioth8FFwBFyBBprJCpsb+f8BxMp0Hc6I+f1wYVoGb/GAVQgGa41gsxiPGEJxvTV
67APpp8zhZrTcY5Qj5ndYjCCBUowggQyoAMCAQICEF0kYdlVNx5E1qcmxun5IrswDQYJKoZI
hvcNAQEFBQAwgd0xCzAJBgNVBAYTAlVTMRcwFQYDVQQKEw5WZXJpU2lnbiwgSW5jLjEfMB0G
A1UECxMWVmVyaVNpZ24gVHJ1c3QgTmV0d29yazE7MDkGA1UECxMyVGVybXMgb2YgdXNlIGF0
IGh0dHBzOi8vd3d3LnZlcmlzaWduLmNvbS9ycGEgKGMpMDUxHjAcBgNVBAsTFVBlcnNvbmEg
Tm90IFZhbGlkYXRlZDE3MDUGA1UEAxMuVmVyaVNpZ24gQ2xhc3MgMSBJbmRpdmlkdWFsIFN1
YnNjcmliZXIgQ0EgLSBHMjAeFw0xMDAxMDIwMDAwMDBaFw0xMTAxMDIyMzU5NTlaMIIBEzEX
MBUGA1UEChMOVmVyaVNpZ24sIEluYy4xHzAdBgNVBAsTFlZlcmlTaWduIFRydXN0IE5ldHdv
cmsxRjBEBgNVBAsTPXd3dy52ZXJpc2lnbi5jb20vcmVwb3NpdG9yeS9SUEEgSW5jb3JwLiBi
eSBSZWYuLExJQUIuTFREKGMpOTgxHjAcBgNVBAsTFVBlcnNvbmEgTm90IFZhbGlkYXRlZDEz
MDEGA1UECxMqRGlnaXRhbCBJRCBDbGFzcyAxIC0gTmV0c2NhcGUgRnVsbCBTZXJ2aWNlMRYw
FAYDVQQDFA1HZXJoYXJkIE11ZW56MSIwIAYJKoZIhvcNAQkBFhNtdWVuekBuZXQuaW4udHVt
LmRlMIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAl/YDW9cnAXJjesGhwtGwj9JN
IHWXQ7QA18s+wGZNsQY0k4eKBeLTk/RAiC1hP+uQAXLoGQHQF/ljjQCQeRmnsN6w04iAzKVo
Ru+NP7Y6ipn8uztx+z0Y3w43CO+LWUudp3PIF3+Fm2zwsAFXj/IYy/8Kq6m9+nQgYO/D4HoN
DDR3//GepXghK0x93gR0GCkm8pgKINHNBD47/IZbDe/cIpAC9HaoxCeLiB1QaCDIYPIrJ0Vx
kk0crAA0tPaMDjdSQTmNRAn2P+7Wg7K+DLvauBHfxLs8PrsolI+XdwiBu6henK9nPxvtoHmm
mfxcdslKLbS9Xz2rHxi6PVVgv1ubWwIDAQABo4HMMIHJMAkGA1UdEwQCMAAwRAYDVR0gBD0w
OzA5BgtghkgBhvhFAQcXATAqMCgGCCsGAQUFBwIBFhxodHRwczovL3d3dy52ZXJpc2lnbi5j
b20vcnBhMAsGA1UdDwQEAwIFoDAdBgNVHSUEFjAUBggrBgEFBQcDBAYIKwYBBQUHAwIwSgYD
VR0fBEMwQTA/oD2gO4Y5aHR0cDovL0luZEMxRGlnaXRhbElELWNybC52ZXJpc2lnbi5jb20v
SW5kQzFEaWdpdGFsSUQuY3JsMA0GCSqGSIb3DQEBBQUAA4IBAQBPU7OeWNo2LRzsdu/RuuAk
L5cj/RzeZ0xv9q4K/eosy9dRu45XACpPvb0fjc3u/dsY4m3jDqzUsB02IrrnF9jpEo10BJPw
b/RZ7NaU7jC/gEknL3bSXiD2PVwL4ZI5ChH5+TJUboWhVh+FpOwOfP/yKk7wQM/46iXXzRIz
enk0rcJp71k4aQgQTML3/8MDeD6TUipuQwQMY9n6T6DsNd/6JpGa+x9ZX4HIEG6+EWwQ85G5
vGfDBaq3dFrksHE75u7b7pf75wm9HIFU2SDHP2tymTHJt/XM9sqP1URIC57s+uZV0/d5KRc1
uPcJ0FAnAiC9o7ybPIh27shiq0FvOEwfMIIFSjCCBDKgAwIBAgIQXSRh2VU3HkTWpybG6fki
uzANBgkqhkiG9w0BAQUFADCB3TELMAkGA1UEBhMCVVMxFzAVBgNVBAoTDlZlcmlTaWduLCBJ
bmMuMR8wHQYDVQQLExZWZXJpU2lnbiBUcnVzdCBOZXR3b3JrMTswOQYDVQQLEzJUZXJtcyBv
ZiB1c2UgYXQgaHR0cHM6Ly93d3cudmVyaXNpZ24uY29tL3JwYSAoYykwNTEeMBwGA1UECxMV
UGVyc29uYSBOb3QgVmFsaWRhdGVkMTcwNQYDVQQDEy5WZXJpU2lnbiBDbGFzcyAxIEluZGl2
aWR1YWwgU3Vic2NyaWJlciBDQSAtIEcyMB4XDTEwMDEwMjAwMDAwMFoXDTExMDEwMjIzNTk1
OVowggETMRcwFQYDVQQKEw5WZXJpU2lnbiwgSW5jLjEfMB0GA1UECxMWVmVyaVNpZ24gVHJ1
c3QgTmV0d29yazFGMEQGA1UECxM9d3d3LnZlcmlzaWduLmNvbS9yZXBvc2l0b3J5L1JQQSBJ
bmNvcnAuIGJ5IFJlZi4sTElBQi5MVEQoYyk5ODEeMBwGA1UECxMVUGVyc29uYSBOb3QgVmFs
aWRhdGVkMTMwMQYDVQQLEypEaWdpdGFsIElEIENsYXNzIDEgLSBOZXRzY2FwZSBGdWxsIFNl
cnZpY2UxFjAUBgNVBAMUDUdlcmhhcmQgTXVlbnoxIjAgBgkqhkiG9w0BCQEWE211ZW56QG5l
dC5pbi50dW0uZGUwggEiMA0GCSqGSIb3DQEBAQUAA4IBDwAwggEKAoIBAQCX9gNb1ycBcmN6
waHC0bCP0k0gdZdDtADXyz7AZk2xBjSTh4oF4tOT9ECILWE/65ABcugZAdAX+WONAJB5Gaew
3rDTiIDMpWhG740/tjqKmfy7O3H7PRjfDjcI74tZS52nc8gXf4WbbPCwAVeP8hjL/wqrqb36
dCBg78Pgeg0MNHf/8Z6leCErTH3eBHQYKSbymAog0c0EPjv8hlsN79wikAL0dqjEJ4uIHVBo
IMhg8isnRXGSTRysADS09owON1JBOY1ECfY/7taDsr4Mu9q4Ed/Euzw+uyiUj5d3CIG7qF6c
r2c/G+2geaaZ/Fx2yUottL1fPasfGLo9VWC/W5tbAgMBAAGjgcwwgckwCQYDVR0TBAIwADBE
BgNVHSAEPTA7MDkGC2CGSAGG+EUBBxcBMCowKAYIKwYBBQUHAgEWHGh0dHBzOi8vd3d3LnZl
cmlzaWduLmNvbS9ycGEwCwYDVR0PBAQDAgWgMB0GA1UdJQQWMBQGCCsGAQUFBwMEBggrBgEF
BQcDAjBKBgNVHR8EQzBBMD+gPaA7hjlodHRwOi8vSW5kQzFEaWdpdGFsSUQtY3JsLnZlcmlz
aWduLmNvbS9JbmRDMURpZ2l0YWxJRC5jcmwwDQYJKoZIhvcNAQEFBQADggEBAE9Ts55Y2jYt
HOx279G64CQvlyP9HN5nTG/2rgr96izL11G7jlcAKk+9vR+Nze792xjibeMOrNSwHTYiuucX
2OkSjXQEk/Bv9Fns1pTuML+ASScvdtJeIPY9XAvhkjkKEfn5MlRuhaFWH4Wk7A58//IqTvBA
z/jqJdfNEjN6eTStwmnvWThpCBBMwvf/wwN4PpNSKm5DBAxj2fpPoOw13/omkZr7H1lfgcgQ
br4RbBDzkbm8Z8MFqrd0WuSwcTvm7tvul/vnCb0cgVTZIMc/a3KZMcm39cz2yo/VREgLnuz6
5lXT93kpFzW49wnQUCcCIL2jvJs8iHbuyGKrQW84TB8xggTsMIIE6AIBATCB8jCB3TELMAkG
A1UEBhMCVVMxFzAVBgNVBAoTDlZlcmlTaWduLCBJbmMuMR8wHQYDVQQLExZWZXJpU2lnbiBU
cnVzdCBOZXR3b3JrMTswOQYDVQQLEzJUZXJtcyBvZiB1c2UgYXQgaHR0cHM6Ly93d3cudmVy
aXNpZ24uY29tL3JwYSAoYykwNTEeMBwGA1UECxMVUGVyc29uYSBOb3QgVmFsaWRhdGVkMTcw
NQYDVQQDEy5WZXJpU2lnbiBDbGFzcyAxIEluZGl2aWR1YWwgU3Vic2NyaWJlciBDQSAtIEcy
AhBdJGHZVTceRNanJsbp+SK7MAkGBSsOAwIaBQCgggLOMBgGCSqGSIb3DQEJAzELBgkqhkiG
9w0BBwEwHAYJKoZIhvcNAQkFMQ8XDTEwMDMwMTE0Mzg0OFowIwYJKoZIhvcNAQkEMRYEFNkK
XEjjXfj9OF3ReMR29EzTTrWJMF8GCSqGSIb3DQEJDzFSMFAwCwYJYIZIAWUDBAECMAoGCCqG
SIb3DQMHMA4GCCqGSIb3DQMCAgIAgDANBggqhkiG9w0DAgIBQDAHBgUrDgMCBzANBggqhkiG
9w0DAgIBKDCCAQMGCSsGAQQBgjcQBDGB9TCB8jCB3TELMAkGA1UEBhMCVVMxFzAVBgNVBAoT
DlZlcmlTaWduLCBJbmMuMR8wHQYDVQQLExZWZXJpU2lnbiBUcnVzdCBOZXR3b3JrMTswOQYD
VQQLEzJUZXJtcyBvZiB1c2UgYXQgaHR0cHM6Ly93d3cudmVyaXNpZ24uY29tL3JwYSAoYykw
NTEeMBwGA1UECxMVUGVyc29uYSBOb3QgVmFsaWRhdGVkMTcwNQYDVQQDEy5WZXJpU2lnbiBD
bGFzcyAxIEluZGl2aWR1YWwgU3Vic2NyaWJlciBDQSAtIEcyAhBdJGHZVTceRNanJsbp+SK7
MIIBBQYLKoZIhvcNAQkQAgsxgfWggfIwgd0xCzAJBgNVBAYTAlVTMRcwFQYDVQQKEw5WZXJp
U2lnbiwgSW5jLjEfMB0GA1UECxMWVmVyaVNpZ24gVHJ1c3QgTmV0d29yazE7MDkGA1UECxMy
VGVybXMgb2YgdXNlIGF0IGh0dHBzOi8vd3d3LnZlcmlzaWduLmNvbS9ycGEgKGMpMDUxHjAc
BgNVBAsTFVBlcnNvbmEgTm90IFZhbGlkYXRlZDE3MDUGA1UEAxMuVmVyaVNpZ24gQ2xhc3Mg
MSBJbmRpdmlkdWFsIFN1YnNjcmliZXIgQ0EgLSBHMgIQXSRh2VU3HkTWpybG6fkiuzANBgkq
hkiG9w0BAQEFAASCAQB+4JpBTr6dIwYRvJ775kzeTRcJz6pRLb2TEg+zHIoKYUm74Ge3stwr
Py2BbtkaQqMcbtsr/RKVy3xIo+UtZJCmQW0IN/pTGeqlbrsV11Ebchq6u1aLrbHbkSLYZ3YA
NbMdswSXfmg+d7s0fFV9kblKxHh/A/JpDUDjPQWbGlOzjgqhKU55c+yqVaJHFwYOzxfMbN9+
1F1baLloT9zsCPe15IMa3u0/H/reH45L3X3rUvT+Zwepem41A3lQqmt+8cwNQJzqUJWzNsnw
xnyNzDyrqfUdh6JZ++htrX57Kg0wh19qystHS4mjywUqql2yHrT+kNITZPMZSEIwwioyyvPu
AAAAAAAA
--------------ms020901030401040107070305--

From wwwrun@core3.amsl.com  Mon Mar  1 06:39:22 2010
Return-Path: <wwwrun@core3.amsl.com>
X-Original-To: ipfix@ietf.org
Delivered-To: ipfix@core3.amsl.com
Received: by core3.amsl.com (Postfix, from userid 30) id D8D5A28C382; Mon,  1 Mar 2010 06:39:22 -0800 (PST)
X-idtracker: yes
To: IETF-Announce <ietf-announce@ietf.org> 
From: The IESG <iesg-secretary@ietf.org>
Message-Id: <20100301143922.D8D5A28C382@core3.amsl.com>
Date: Mon,  1 Mar 2010 06:39:22 -0800 (PST)
Cc: ipfix@ietf.org
Subject: [IPFIX] Last Call: draft-ietf-ipfix-mediators-problem-statement (IPFIX Mediation: Problem Statement) to Informational RFC
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
Reply-To: ietf@ietf.org
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Mar 2010 14:39:23 -0000

The IESG has received a request from the IP Flow Information Export WG 
(ipfix) to consider the following document:

- 'IPFIX Mediation: Problem Statement '
   <draft-ietf-ipfix-mediators-problem-statement-08.txt> as an Informational RFC

The IESG plans to make a decision in the next few weeks, and solicits
final comments on this action.  Please send substantive comments to the
ietf@ietf.org mailing lists by 2010-03-15. Exceptionally, 
comments may be sent to iesg@ietf.org instead. In either case, please 
retain the beginning of the Subject line to allow automated sorting.

The file can be obtained via
http://www.ietf.org/internet-drafts/draft-ietf-ipfix-mediators-problem-statement-08.txt


IESG discussion can be tracked via
https://datatracker.ietf.org/public/pidtracker.cgi?command=view_id&dTag=17243&rfc_flag=0


From muenz@net.in.tum.de  Mon Mar  1 08:58:38 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 5E3D028B23E for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 08:58:38 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id SchsOEDmqjkY for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 08:58:36 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id C907328C288 for <ipfix@ietf.org>; Mon,  1 Mar 2010 08:58:35 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 5F949200AD9A; Mon,  1 Mar 2010 17:58:34 +0100 (CET)
Message-ID: <4B8BF243.7090303@net.in.tum.de>
Date: Mon, 01 Mar 2010 17:58:43 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Lothar Braun <braun@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu>	<4AD793E9-3463-43F1-8840-13BB6FF0B1E6@net.in.tum.de> <33C32EC7-A6D6-4238-B376-3F91FA3B74D4@net.in.tum.de>
In-Reply-To: <33C32EC7-A6D6-4238-B376-3F91FA3B74D4@net.in.tum.de>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms020204080900070206080205"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] WG last call on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Mar 2010 16:58:38 -0000

This is a cryptographically signed message in MIME format.

--------------ms020204080900070206080205
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Lothar,

Thanks a lot for your review.

Please find answers inline.

Regards,
Gerhard

Lothar Braun wrote:

>> o  Definitions adopted from [RFC5101]: *  Collection Process *
>> Collector *  Data Record *  Exporter *  Flow Record *  Information
>> Element *  IPFIX Device *  IPFIX Message *  Observation Domain *
>> Observation Point *  (Options) Template
>>=20
>> o  Definitions adopted from [I-D.ietf-ipfix-file]: *  File Reader *
>> File Writer
>>=20
>> o  Definitions adopted from [RFC5476]: *  Filtering *  Observed
>> Packet Stream *  Packet Report *  PSAMP Device *  Sampling *
>> Selection Process *  Selection Sequence *  Selection Sequence
>> Report Interpretation *  Selector, Primitive Selector, Composite
>> Selector
>=20
> You might want to add the term "Flow Keys" somewhere to this lists.

Right, done.

>> Cache:  The Cache is a functional block in a Metering Process which
>>  maintains IPFIX Flow Records or PSAMP Packet Reports.  According=20
>> to [RFC5101], the maintenance of Flow Records may include creating=20
>> new records, updating existing ones, computing Flow statistics,=20
>> deriving further Flow properties, detecting Flow expiration,=20
>> passing Flow Records to the Exporting Process, and deleting Flow=20
>> Records.  The maintenance of Packet Reports covers the same set of=20
>> functions.
>>=20
>> Cache Layout:  The Cache Layout defines the superset of fields that
>>  are included in the Packet Reports or Flow Records maintained by=20
>> the Cache.  The fields are specified by the corresponding=20
>> Information Elements.  In general, the largest possible subset of=20
>> the specified fields is derived for every Packet Report or Flow=20
>> Record.  More specific rules about which fields must be included=20
>> are given in Section 4.3.1.
>>=20
>> Cache Mode:  The Cache Mode specifies whether Packet Reports or
>> Flow Records are generated by the Cache.  In the case of Flow
>> Records, it also specifies the Flow expiration policy.
>>=20
>=20
> I had some difficulties to understand the definition of the Cache. I
> could not understand what the packet cache does when generating PSAMP
> Packet Reports. It was clear after reading Section 4.3 when the Cache
> Modes are explained. Maybe there could be some text that explains
> that maintaining Packet Reports and Flow Records is the same task
> except that Packet Records expire after the first packet.

After Atsushi's review, the text of the Cache definition has been=20
changed to:

    Cache:  The Cache is a functional block in a Metering Process which
       generates IPFIX Flow Records or PSAMP Packet Reports from a
       Selected Packet Stream, in accordance with its configuration.  If
       Flow Records are generated, the Cache performs tasks like creating=

       new records, updating existing ones, computing Flow statistics,
       deriving further Flow properties, detecting Flow expiration,
       passing Flow Records to the Exporting Process, and deleting Flow
       Records.  If Packet Reports are generated, the Cache performs
       tasks like extracting packet contents and derived packet
       properties from the Selected Packet Stream, creating new records,
       and passing them as Packet Reports to the Exporting Process.


Does this text make things clearer?

 > Could there be an example for a Cache?

Not sure what you mean. What kind of example? Where should it appear in=20
the text?

>> +-----------------------------------+ | Metering Process
>> | | +-----------+ Selected            | Observed | | Selection |
>> Packet    +-------+ |  Stream of Packet  -->| Process
>> |---------->| Cache |--> Flow Records or Stream   | +-----------+
>> Stream    +-------+ |  Packet Reports=20
>> +-----------------------------------+
>>=20
>> Figure 1: Selection Process and Cache forming a Metering Process
>>=20
>> The configuration data model adopts the separation of Selection=20
>> Processes and Caches in order to support the flexible configuration
>> and combination of these functional blocks.  As defined in
>> [RFC5476], the Selection Process takes the Observed Packet Stream
>> as its input and selects a subset of that stream as its output
>> (Selected Packet Stream).  The action of the Selection Process on a
>> single packet of its input is defined by a Primitive Selector or a
>> Composite Selector. The Cache generates Flow Records or Packet
>> Reports from the Selected Packet Stream, depending on the
>> configured Cache Mode.
>>=20
>> The Observed Packet Stream at the input of a Selection Process MUST
>> only contain packets originating from a single Observation Domain.
>> Similarly, the Selected Packet Stream at the input of a Cache MUST
>> only contain packets originating from a single Observation Domain.
>> Packets from Observation Points belonging to different Observation
>> Domains MUST NOT enter the same Selection Process or the same
>> Cache.
>=20
> When I read this, I had the feeling that this text does poses some
> restriction on how I am able to configure my monitor. But it seems
> like this corresponds to the definition of an Observation Domain:
>=20
> Observation Domain
>=20
> An Observation Domain is the largest set of Observation Points for=20
> which Flow information can be aggregated by a Metering Process. ...
>=20
> Is there some further restriction that I miss?

The text has been changed because of some late changes in IPFIX-MIB and=20
because of Atsushi's review.

PSAMP defines:

    * Observed Packet Stream

       The Observed Packet Stream is the set of all packets observed at
       the Observation Point.

    * Selection Process

       A Selection Process takes the Observed Packet Stream as its input
       and selects a subset of that stream as its output.

According to my interpretation of these definitions, a Selection Process =

may only processes packets from one Observation Point. Therefore, the=20
text needed to be more restrictive.

On the other hand, there is no IPFIX/PSAMP document saying that a=20
Metering Process cannot process packets of multiple ODs. Therefore, this =

restriction was removed.

I propose to add another paragraph in response to your next comment. So=20
please find the complete new text below.


>> 3.2.  Exporter Configuration
>>=20
>> Figure 4 below shows the main classes of the configuration data
>> model which are involved in the configuration of an IPFIX or PSAMP=20
>> Exporter.  The role of the classes can be briefly summarized as=20
>> follows:
>>=20
>> o  The ObservationPoint class specifies an Observation Point (i.e.,
>>  an interface or linecard) of the Monitoring Device at which=20
>> packets are captured for traffic measurements.  An object of the=20
>> ObservationPoint class may be associated with one or more objects=20
>> of the SelectionProcess class configuring Selection Processes that=20
>> process the observed packets in parallel.  As long as an=20
>> ObservationPoint object is specified without any references to=20
>> SelectionProcess objects, the Observation Point is not deployed for
>> traffic measurements.
>>=20
>> o  The SelectionProcess class contains the configuration parameters
>>  of a Selection Process.  The Selection Process may be composed of=20
>> a single Selector or a sequence of Selectors, defining a Primitive=20
>> or Composite Selector, respectively.
>>=20
>> The Selection Process selects packets from an Observed Packet=20
>> Stream originating from one or multiple Observation Points.
>=20
> Could you please introduce the selectAll policy of the
> SelectionProcess at this point?
>=20
> When I read this, I was wondering why I cannot connect an Observation
> Point directly to a Cache, if I do not want to sample or filter
> packets.

I propose to mention this in Section 3 already:

                +-----------------------------------+
                | Metering Process                  |
                | +-----------+ Selected            |
       Observed | | Selection | Packet    +-------+ |  Stream of
       Packet  -->| Process   |---------->| Cache |--> Flow Records or
       Stream   | +-----------+ Stream    +-------+ |  Packet Reports
                +-----------------------------------+

      Figure 1: Selection Process and Cache forming a Metering Process

    The configuration data model adopts the separation of Selection
    Processes and Caches in order to support the flexible configuration
    and combination of these functional blocks.  As defined in [RFC5476],=

    the Selection Process takes the Observed Packet Stream as its input
    and selects a subset of that stream as its output (Selected Packet
    Stream).  The Observed Packet Stream at the input of a Selection
    Process is composed of packets from a single Observation Point.  The
    action of the Selection Process on a single packet of its input is
    defined by a Primitive Selector or a Composite Selector.  The Cache
    generates Flow Records or Packet Reports from the Selected Packet
    Stream, depending on the configured Cache Mode.

    Every Metering Process contains at least one Selection Process and at=

    least one Cache.  If all packets in the Observed Packet Stream are
    selected without any Filtering or Sampling, this corresponds to a
    Selection Process implementing a trivial Selector which selects all
    packets.

    Beyond the basic Metering Process shown in Figure 1, the
    configuration data model enables the specification of more complex
    configurations.  A Metering Process may be composed of multiple
    Selection Processes and multiple Caches.  A single Cache may process
    the output of multiple Selection Processes.  The Selected Packet
    Stream may be copied to enter several Caches, for example one Cache
    that generates Flow Records and another Cache that generates Packet
    Reports.  Selection Processes can also be cascaded, such that the
    Selected Packet Stream at the output of one Selection Process is
    passed to another Selection Process.  Cascading Selection Processes
    can be useful in specific contexts such as the example in
    Section 7.3.

Is this ok?

>> 4.1.  ObservationPoint Class
>>=20
>> +-------------------------------+ | ObservationPoint              |
>>  +-------------------------------+          1
>> +--------------------+ | name
>> |<>----------| Interface/Linecard | | observationPointId {readOnly}
>> |            +--------------------+ | observationDomainId
>> | |                               | 0..*  0..*
>> +--------------------+ |
>> |----------->| SelectionProcess   |=20
>> +-------------------------------+            +--------------------+
>>=20
>>=20
>> +------------------+   +----------------------------------+ |
>> Interface        |   | Linecard                         |=20
>> +------------------+   +----------------------------------+ |
>> ifIndex/ifName   |   | entPhysicalIndex/entPhysicalName | |
>> direction =3D both |   | direction =3D both                 |=20
>> +------------------+   +----------------------------------+
>>=20
>> Figure 6: ObservationPoint class
>>=20
>=20
> Both, the Interface and the Linecard, contain the "direction". Is
> there a reason why it is not a part of the Observation Point if it
> has to be specified in both cases?

Good question which has also been mentioned by Andrew.

There is a small advantage: The current position of the parameter=20
"direction" makes it easier to add vendor or implementation-specific=20
packet sources for which the direction parameter has no meaning, for=20
example:

+------------------+
| PcapFile         |
+------------------+
| filename         |
+------------------+

>>=20
>> 4.3.1.  CacheLayout Class
>>=20
>> +--------------+ | CacheLayout  | +--------------+   1..*
>> +--------------------------+ |              |<>------| CacheField
>> | |              |        +--------------------------+ |
>> |        | name                     | |              |        |
>> ieId/ieName              | |              |        | ieLength
>> {opt.}          | |              |        |
>> ieEnterpriseNumber[0..1] | |              |        |
>> isFlowKey[0..1]          | +--------------+
>> +--------------------------+
>>=20
>> Figure 12: CacheLayout class
>>=20
>> A Cache generates and maintains Packet Reports or Flow Records=20
>> containing information that has been extracted from the incoming=20
>> stream of packets.  Using the CacheField class, the CacheLayout
>> class specifies the superset of fields that are included in the
>> Packet Reports or Flow Records (see Figure 12).
>>=20
>> If Packet Reports are generated (i.e., if Cache Mode is
>> "immediate"), every field specified by the Cache Layout MUST be
>> included in the resulting Packet Report unless the corresponding
>> Information Element is not applicable or cannot be derived from the
>> content or treatment of the incoming packet.
>=20
> Could you give an example for "not applicable or cannot be derived
> from"? Maybe something like when a source port needs to be extracted
> from an ICMP packet.

Ok. What about the following new paragraph:

    For example, if a field is configured to contain the TCP source port
    (Information Element tcpSourcePort [RFC5102]), the field MUST be
    included in all Packet Reports which are related to TCP packets.
    Obviously, the field value cannot be determined for non-TCP packets.

At the moment, we leave it open how to handle the UDP case. There may be =

very simple Exporters supporting only a single Template per Cache.
In this case, the source port field can be simply set to zero.

>> If Flow Records are generated (i.e., if Cache Mode is "timeout" or=20
>> "permanent"), every Flow Key field specified by the Cache Layout
>> MUST be included as Flow Key in the resulting Flow Record unless
>> the corresponding Information Element is not applicable or cannot
>> be derived from the content or treatment of the incoming packet.
>> Two packets MUST be accounted by different Flow Records if
>> different subsets of the Flow Key fields are applicable or
>> derivable.
>=20
> I do not understand this. What do you mean with "different subsets of
> the Flow Key fields".
>=20
> Is it maybe: "Two packets MUST be accounted by different Flow Records
> if they differ in at least one Flow Key field."

I like this sentence. And it is important that this behavior is well=20
understood.

Therefore, I propose the following changes:

    If Flow Records are generated (i.e., if Cache Mode is "timeout" or
    "permanent"), every Flow Key field specified by the Cache Layout MUST=

    be included as Flow Key in the resulting Flow Record unless the
    corresponding Information Element is not applicable or cannot be
    derived from the content or treatment of the incoming packet.  Two
    packets MUST be accounted by different Flow Records if they differ in=

    at least one Flow Key field.  Hence, if a Flow Key field can be
    determined for one packet but not for the other, the two packets are
    accounted in different Flow Records.

    Every non-key field specified by the Cache Layout MUST be included in=

    the resulting Flow Record unless the corresponding Information
    Element is not applicable or cannot be derived for the given Flow.
    Two packets are accounted by the same Flow Record if they only differ=

    in non-key fields but not in any of the Flow Key fields.  The same
    applies if one or more non-key fields can be determined for one
    packet but not for the other.

    For example, if a non-key field specifies an Information Element
    whose value is determined by the first packet observed within a Flow
    (which is the default rule according to [RFC5102]), this field MUST
    be included in the resulting Flow Record if it can be determined from=

    the first packet of the Flow.

Is this ok?

>> ieId, ieName, ieEnterpriseNumber:  These parameters specify a field
>> by the combination of the Information Element identifier or name,=20
>> and the Information Element enterprise number.  Either ieId or=20
>> ieName MUST be specified. ieEnterpriseNumber MUST be used for=20
>> enterprise-specific Information Elements.  If ieEnterpriseNumber is
>> omitted or zero, this is Information Element is not enterprise-=20
>> specific but registered at IANA.
>=20
> If I want to configure a custom non-IANA field using the Enterprise
> ID 1. The field should have the ieID 1. What do I have to configure.
> Do I need to configure ieID 1, ieEnterpriseNumber 1=20

yes.

> or do I need to give ieID 32768 (the first bit if the id field) and
> ieEnterpriseNumber 1?

no.

> So my question is: Is the enterprise Bit set by the user or is it
> done by the monitor if he sees the ieEnterpriseNumber field?

RFC5102 says:

    The value of these identifiers is in the range of 1-32767.  Within
    this range, Information Element identifier values in the sub-range of=

    1-127 are compatible with field types used by NetFlow version 9
    [RFC3954].

    Enterprise-specific Information Element identifiers have the same
    range of 1-32767, but they are coupled with an additional enterprise
    identifier.  For enterprise-specific Information Elements,
    Information Element identifier 0 is also reserved.

So, 32768 is not a valid ID.

I will restrict the value range of all ieId parameters in YANG.

>> ieLength:  This parameter specifies the length of the field in=20
>> octets.  A value of 65535 means that the field is encoded as=20
>> variable-length Information Element.  For Information Elements of=20
>> integer and float type, the field length MAY be set to a smaller=20
>> value than the standard length of the abstract data type if the=20
>> rules of reduced size encoding are fulfilled (see [RFC5101],=20
>> Section 6.2).  If not configured by the user, the field length is=20
>> set by the Monitoring Device.
>=20
> The monitoring device can set the length field if it nows the ID
> (either IANA or known custom ID). But what if the monitor does not
> know a specific ID? Should there be a default value (which one?) for
> unknown IDs or should the device reject such a configuration?

If the Monitoring Device does not know the ID, the Cache cannot extract=20
or derive the field value from the packets, anyway.

Since the Monitoring Device is obliged to export these fields unless=20
they are not available, the device needs some knowledge about the IEs.=20
So, I think that the device should be able to set a default length.

Do you think that we should change the text?

> I think the device should reject the configuration.


>> 7.3.  Export of Flow Records and Packet Reports
>>=20
>> This configuration example demonstrates the combined export of Flow
>>  Records and Packet Reports for a single Observation Point.  A=20
>> Selection Process (Selection Sequence ID =3D 1) applies random
>> Sampling to the stream of observed packets.  The output is passed
>> to a Cache generating Flow Records.  In parallel, the output is
>> passed to a second Selection Process (Selection Sequence ID =3D 2)
>> which discards all non-ICMP packets.  A second Cache generates
>> Packet Reports of the retained ICMP packets.  The output of both
>> caches is exported to a single Collector using SCTP.
>>=20
>> <ipfix xmlns=3D"urn:ietf:params:xml:ns:ietf-ipfix-psamp">
>>=20
>> <observationPoint> <name>OP at linecard 3</name>=20
>> <observationDomainId>9876</observationDomainId> <interface>=20
>> <ifIndex>4</ifIndex> <direction>ingress</direction> </interface>=20
>> <selectionProcess>Sampling</selectionProcess> </observationPoint>
>>=20
>> <selectionProcess> <name>Sampling</name>=20
>> <selectionSequenceId>1</selectionSequenceId> <selector>=20
>> <name>Random sampler</name> <selectorId>1</selectorId>=20
>> <sampUniProb> <probability>0.01</probability> </sampUniProb>=20
>> </selector> <selectionProcess>ICMP</selectionProcess> <cache>Flow
>> cache</cache> </selectionProcess>
>>=20
>> <selectionProcess> <name>ICMP</name>=20
>> <selectionSequenceId>2</selectionSequenceId> <selector> <name>ICMP
>> filter</name>
>> <selectorId>2</selectorId> <filterMatch> <ieId>4</ieId>=20
>> <value>1</value> </filterMatch> </selector> <cache>Packet
>> cache</cache> </selectionProcess>
>>=20
>> <cache> <name>Flow cache</name> <cacheMode>timeout</cacheMode>=20
>> <maxRecords>4096</maxRecords> <activeTimeout>5</activeTimeout>=20
>> <inactiveTimeout>10</inactiveTimeout> <cacheLayout> <cacheField>=20
>> <name>Field 1</name> <ieName>sourceIPv4Address</ieName>=20
>> <isFlowKey/> </cacheField> <cacheField> <name>Field 2</name>=20
>> <ieName>destinationIPv4Address</ieName> <isFlowKey/> </cacheField>=20
>> <cacheField> <name>Field 6</name>=20
>> <ieName>flowStartMilliSeconds</ieName> </cacheField> <cacheField>=20
>> <name>Field 7</name> <ieName>flowEndSeconds</ieName> </cacheField>=20
>> <cacheField> <name>Field 8</name> <ieName>octetDeltaCount</ieName>=20
>> </cacheField> <cacheField> <name>Field 9</name>=20
>> <ieName>packetDeltaCount</ieName> </cacheField> </cacheLayout>=20
>> <exportingProcess>Export</exportingProcess> </cache>
>>=20
>> <cache> <name>Packet cache</name>
>> <cacheMode>immediate</cacheMode> <maxRecords>512</maxRecords>=20
>> <cacheLayout> <cacheField> <name>Field 1</name> <ieId>313</ieId>=20
>> <ieLength>64</ieLength> </cacheField> <cacheField> <name>Field
>> 2</name> <ieId>154</ieId> </cacheField> </cacheLayout>=20
>> <exportingProcess>Export</exportingProcess> </cache>
>>=20
>> <exportingProcess> <name>Export</name> <destination> <name>SCTP
>> collector</name> <transportProtocol>sctp</transportProtocol>=20
>> <destinationIpAddress>192.0.2.1</destinationIpAddress>=20
>> <timedReliability>0</timedReliability>=20
>> <numberOfStreams>2</numberOfStreams> </destination>=20
>> </exportingProcess>
>>=20
>> </ipfix>
>>=20
>> The Observed Packet Stream at the input of the Selection Process=20
>> "ICMP" originates from the Selection Process "Sampling", which thus
>>  constitutes a pseudo Observation Point.  In order to inform the=20
>> Collector about the cascaded Selection Processes, the Exporting=20
>> Process exports two Selection Sequence Report Interpretations as=20
>> defined in [RFC5476], section 6.5.1, including the following
>> fields:
>=20
> Shouldn't there be something like
>=20
> <option> <selectionSequence/> </option>
>=20
> in the exportingProcess configuration?

Yes, that would be useful to configure a PSAMP compliant Monitoring Devic=
e.

I added

    <options>
       <name>Options 1</name>
       <optionsType>selectionSequence</optionsType>
       <optionsTimeout>0</optionsTimeout>
     </options>

to examples 7.1 and 7.3.


--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms020204080900070206080205
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms020204080900070206080205--

From root@core3.amsl.com  Mon Mar  1 09:45:02 2010
Return-Path: <root@core3.amsl.com>
X-Original-To: ipfix@ietf.org
Delivered-To: ipfix@core3.amsl.com
Received: by core3.amsl.com (Postfix, from userid 0) id 2304028C42A; Mon,  1 Mar 2010 09:45:01 -0800 (PST)
From: Internet-Drafts@ietf.org
To: i-d-announce@ietf.org
Content-Type: Multipart/Mixed; Boundary="NextPart"
Mime-Version: 1.0
Message-Id: <20100301174502.2304028C42A@core3.amsl.com>
Date: Mon,  1 Mar 2010 09:45:02 -0800 (PST)
Cc: ipfix@ietf.org
Subject: [IPFIX] I-D ACTION:draft-ietf-ipfix-psamp-mib-00.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Mar 2010 17:45:02 -0000

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts 
directories.
This draft is a work item of the IP Flow Information Export Working Group of the IETF.

	Title		: Definitions of Managed Objects for Packet Sampling
	Author(s)	: T. Dietz, B. Claise, J. Quittek
	Filename	: draft-ietf-ipfix-psamp-mib-00.txt
	Pages		: 34
	Date		: 2010-3-1
	
   This memo defines a portion of the Management Information Base (MIB)
   for use with network management protocols in the Internet community.
   In particular, it describes extensions to the IPFIX MIB module
   [I-D.ietf-ipfix-mib].  For IPFIX implementations that use packet
   sampling (PSAMP) techniques as described in [RFC5475], this memo
   defines the PSAMP MIB module containing managed objects for providing
   information on applied packet selection functions and their
   parameters.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-ipfix-psamp-mib-00.txt

Internet-Drafts are also available by anonymous FTP at:
ftp://ftp.ietf.org/internet-drafts/

Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

--NextPart
Content-Type: Message/External-body;
	name="draft-ietf-ipfix-psamp-mib-00.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID: <2010-3-1094204.I-D@ietf.org>


--NextPart--


From muenz@net.in.tum.de  Mon Mar  1 10:36:49 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 7745F28C50F for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 10:36:49 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id W+rqWfEtBJ6e for <ipfix@core3.amsl.com>; Mon,  1 Mar 2010 10:36:47 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 71D2B28C4B8 for <ipfix@ietf.org>; Mon,  1 Mar 2010 10:36:47 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 08CE021102B4; Mon,  1 Mar 2010 19:36:42 +0100 (CET)
Message-ID: <4B8C0943.6070309@net.in.tum.de>
Date: Mon, 01 Mar 2010 19:36:51 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Brian Trammell <trammell@tik.ee.ethz.ch>
References: <3146E35B-9877-44AE-B884-17541E6ED5B8@tik.ee.ethz.ch>
In-Reply-To: <3146E35B-9877-44AE-B884-17541E6ED5B8@tik.ee.ethz.ch>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms010608020808080409000607"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] WGLC comments on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Mar 2010 18:36:49 -0000

This is a cryptographically signed message in MIME format.

--------------ms010608020808080409000607
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Brian,

Thanks a lot for your review.

Data models in network management will never satisfy all people. So,=20
please keep in mind that the goal of this document is to standardize a=20
configuration data model that can be easily implemented by many vendors. =

There will always be additional vendor-specific parameters, and there=20
will always be vendor-specific restrictions of the model. Thanks to=20
YANG, there are possibilities to make restrictions and extension.

So, ask yourself if the parameters which are contained in the=20
configuration data model can be easily mapped to one of the YAF command=20
line options. Not the other way round.

More comments inline.

Gerhard


Brian Trammell wrote:
> Greetings, all,
>=20
> I have reviewed draft-ietf-ipfix-configuration-model-04 in response
> to the WGLC, and have the following comments, ranked in order of
> impact; most of these deal with the applicability of the
> configuration data model to implementations other than the "exemplar"
> implementation(s) the authors had in mind while while writing.  I
> will say that overall, the draft does an more than reasonable job of
> being generally applicable. I apologize for starting a new thread,
> but I have not yet had time to properly review other comments, but
> plan on being in Anaheim prepared to discuss those too. I have also
> explicitly not reviewed the YANG definition; I leave that to people
> with actual YANG expertise.
>=20
> First and foremost is the issue of caches, as it is in cache
> implementation that IPFIX packet-to-flow MPs will differentiate the
> most from each other, and indeed most vendors should be expected to
> subclass this cache class for their own configuration parameters;
> therefore getting this right is probably the most crucial. cacheMode
> immediate seems like a bit of a hack, as it invalidates basically
> every other cache field. However, I suppose I can see it as a way to
> keep the class design clean (otherwise, there would be nothing to
> connect OP to layout/EP in a PSAMP implementation...).

Right.

> 1. I'm concerned about the definition of cacheMode timeout (4.3, p.
> 24), as it seems to make it impossible to configure a cache to allow
> cache exit on TCP state tracking. Perhaps this is merely an editorial
> oversight?

We can allow a timeout Cache to expire a flow after a certain flow=20
termination event. Is this what you propose?

Could you please provide some text how to change the following parameter =

description:

    cacheMode:  Configures the Cache Mode.  The following parameter
       values are specified by the configuration data model:
       *  immediate: Data Records expire after the first packet
       *  timeout: Data Records expire after active or inactive timeout
       *  permanent: Data Records never expire, but are periodically
          exported with interval set by exportInterval
       In the case of "immediate", PSAMP Packet Reports are generated.
       Otherwise, IPFIX Flow Records are generated.

> 2. Destination (4.4.1, pp .27-29) appears not to match the properties
> of the particular transports:
 >
> a. can't specify a local IP address for TCP or UDP. This seems
> necessary on devices with multiple management interfaces, which I
> don't think is all that weird a deployment plan.

This will be fixed as proposed in this mail unless anybody objects:
http://www.ietf.org/mail-archive/web/ipfix/current/msg05316.html

> b. can't specify 1..n remote IP addresses for SCTP. This breaks one
> of the advantages of SCTP.

I see send_connectx() in the SCTP API:
http://tools.ietf.org/html/draft-ietf-tsvwg-sctpsocket-21#section-8.9

However, what happens if you configure multiple remote IP addresses and=20
they do not belong to the same SCTP endpoint?
What shall the Exporting Process do in this situation?

I'm not sure if it makes sense to configure multiple remote addresses=20
for a single destination. It seems to be safer to configure multiple=20
export destinations and to use export member type primary and secondary.

What do you think?

> c. can't configure SCTP reliability per-template, as opposed to per
> destination. see 2.b.

As discussed long ago, Templates are not configured. The are created by=20
the Exporting Process depending on the Data Records to be exported.

Therefore, there is no good place for a per-Template reliability=20
parameter in the configuration data model. The only possibility would be =

to configure the reliability in the Cache, and this would be a very=20
strange place to do it.

I had discussed this issue with Benoit. The conclusion was that we have=20
a per-Destination parameter. If people want to configure more, they can=20
come up with an extension.

If you have another convincing solution, please let me know.

> 3. Options (4.4.3, p. 30) configuration is rather odd and highly
> restrictive, even to the point of being unusable. The biggest problem
> is that RFC5473 common properties are something completely different
> from the options specified in 5101 and 5476. "Enable common
> properties", without some definition of which fields to use as common
> properties and under what circumstances, is not a useful directive.

The idea is that the Exporting Process defines common properties=20
depending on the Data Records to be exported. This can be done=20
dynamically (depending on the value distribution) or using knowledge=20
about the configuration (e.g., if a filter ensures that the accounted=20
packets are from/to a specific endpoint, this can be encoded as common=20
property).

If you think that reducingRedundancy is not useful, I will remove from=20
the list.

> Secondarily, options specified in 5610 and 5655 are not covered; do
> we really want to amend this document every time we specify another
> WG-blessed options template?=20

Apart from reducingRedundancy, all mentioned Options type appear in the=20
protocol RFCs (RFC5101, RFC5476).

Adding more parameter values for optionsType can be easily achieved with =

YANG extensions because they are specified as identities (it would note=20
be possible with an enum type).

 > I'd suggest using an OptionsTemplate
> subclass of Template, instead, and providing the definition of common
> (non-5473) options templates in an appendix to the configuration data
> model.

I do not understand your proposal. There is no Template class for=20
configuration.

> 4. There does not seem to be any room for logical configurable
> processes between selection and cache, or cache and EP. The example
> that comes to mind is perhaps implementation-specific [I'm mentioning
> this because yaf does it]: fragmentation behavior (reassemble,
> ignore, drop). How could I allow then for the configuration of the
> fragment table (given that a selection process can only reference a
> cache)?

You can always extend the model. For example, if fragmentation=20
reassembly can be regarded as part of the Cache, you can easily define=20
your own new cacheType (again, identities and identityref in YANG).

We also have implemented a lot of cool additional functionalities which=20
we are going to configure by extending the model. It is not the purpose=20
of this document to cover all of these.

> 5. CacheLayout (4.3.1, p. 25) does not appear to be able to support
> adaptive reduced-length export (exporting, say, counters with
> unsigned32 unless one of them actually requires unsigned64) [yes, yaf
> does this too]. Though this could probably be solved with a
> deviation, and an additional vendor-specific CacheField attribute...

Is reduced size encoding performed by the Cache or by the Exporting Proce=
ss?

If you implement a Cache in hardware, it is probably easier to always=20
use the maximum length and to reduce it later in the Exporting Process.
We assume that the Cache reserves a certain field length for all Data=20
Records in the Cache.  (Variable-length fields are treated differently)

The ieLength specifies the field length in the Cache. The length in the=20
exported Data Record can be different. Note again: we are not=20
configuring a Template, here. The exported Data Record may even contain=20
a completely different set of fields, for example in the case of=20
reducing redundancy.

What about removing the following sentence to avoid any confusion:

	For Information Elements of
       integer and float type, the field length MAY be set to a smaller
       value than the standard length of the abstract data type if the
       rules of reduced size encoding are fulfilled (see [RFC5101],
       Section 6.2).

> 6. CacheLayout (4.3.1, p. 25) could do with an explicit example to
> illustrate the "unless the corresponding IE is not applicable or
> cannot be derived" case: I presume this is intended for dual-stack
> IPv4/IPv6 measurement, RFC5103 biflows, and the like? It does appear
> than this clause is sufficient for such cases, but I had to read this
> a couple of times before I understood that's what was meant.

Right. I tried to improve this. Please look into my reply to Lothar's=20
review:
http://www.ietf.org/mail-archive/web/ipfix/current/msg05325.html


> 7. Real File Readers and File Writers will probably need more than a
> single filename endpoint on the file system (however, most file
> readers and file writers in document-based workflows won't be
> configured using this information model...) How to handle this is
> probably implementation-specific; in any case, I don't have a good
> suggestion here, yet; perhaps we can discuss in Anaheim.

I assume that the URI filename parameter is useful for many=20
FileWriter/Reader implementations.  All additional parameters are=20
probably very specific to the implementation.

> 8. Is there any benefit to cascading caches as with selection
> processes, or rather, are the benefits of cascading caches considered
> in scope for this application? The simplest example would be
> simultaneous export of flow data as well as flow aggregate data (e.g.
> traffic counters per AS pair), perhaps as part of a
> cascading-aggregation mediator. I ask only because it seems a rather
> small and easily supportable change to greatly increase flexibility.

Uff. I already have problems to convince people that there may be more=20
than one parallel Caches in a Metering Process. Now, you even propose=20
multiple Caches in a series :)

IMO, aggregation is done in an Intermediate Process, not in a Metering=20
Process and therefore not in the Cache of a Metering Process.

The configuration of Mediators and Intermediate Process is out of scope=20
of this document. Otherwise, we will never finish...

> 9. Destination exportMemberType appears to reference a renamed MIB
> object as of MIB-10; this is called ipfixExportMemberType now. I
> would suggest copying the complete description from MIB, because the
> semantics are kind of confusing otherwise (I had to read MIB, which I
> probably should have done by now but oh well, in order to figure out
> what the difference between parallel and primary was). A tertiary
> problem: these exportMemberTypes seem a bit restrictive (e.g. what if
> I want to do parallel export and also have secondaries?), but I
> suppose that's a MIB comment.

Andrew suggested to improve this as well.  See:
http://www.ietf.org/mail-archive/web/ipfix/current/msg05323.html
I'm open to concrete suggestions.

> 10. draft-ietf-ipfix-file is now RFC5655 (the editor will catch
> this); since this is now an RFC, references to things "proposed" in
> this draft should probably be changed to things "specified" in RFC
> 5655.

Done.


--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms010608020808080409000607
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms010608020808080409000607--

From akoba@nttv6.net  Tue Mar  2 04:23:53 2010
Return-Path: <akoba@nttv6.net>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 4D26E3A7B89 for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 04:23:53 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.611
X-Spam-Level: 
X-Spam-Status: No, score=0.611 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HOST_MISMATCH_COM=0.311, RDNS_DYNAMIC=0.1, SUBJ_RE_NUM=2.799]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id dmDmsfaEnWKY for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 04:23:52 -0800 (PST)
Received: from mail.nttv6.net (mail.nttv6.net [IPv6:2001:fa8::25]) by core3.amsl.com (Postfix) with ESMTP id 547E83A8191 for <ipfix@ietf.org>; Tue,  2 Mar 2010 04:23:51 -0800 (PST)
Received: from [127.0.0.1] (dhcp-3-152.nttv6.com [192.47.163.152]) by mail.nttv6.net (8.14.3/8.14.3) with ESMTP id o22CNn4k059232; Tue, 2 Mar 2010 21:23:50 +0900 (JST) (envelope-from akoba@nttv6.net)
Date: Tue, 02 Mar 2010 21:19:05 +0900
From: Atsushi Kobayashi <akoba@nttv6.net>
To: Gerhard Muenz <muenz@net.in.tum.de>
In-Reply-To: <4B8B7E20.1020504@net.in.tum.de>
References: <20100209010851.F028.17391CF2@nttv6.net> <4B8B7E20.1020504@net.in.tum.de>
Message-Id: <20100302203953.40CF.17391CF2@nttv6.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
X-Mailer: Becky! ver. 2.50.05 [ja] (Unregistered)
X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.2.3 (mail.nttv6.net [192.16.178.5]); Tue, 02 Mar 2010 21:23:50 +0900 (JST)
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] WGLC comments on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 02 Mar 2010 12:23:53 -0000

Hi Gerhard,

Thank you for your reply. Please see in-line.

On Mon, 01 Mar 2010 09:43:12 +0100
Gerhard Muenz <muenz@net.in.tum.de> wrote:

> 
> Hi Atsushi,
> 
> Atsushi Kobayashi wrote:
> > Hi Gerhard, and all,
> > 
> > I find my following answer could not have been shared in list due to our
> > network trouble.
> 
> Thank you for you comments. As far as I can see, there are only two 
> issues left:
> 
> 1) Match Filters
> 
> >>>>>>> 4.2.3.  Filter Classes
> >>>>>>>
> >>>>>>>            +--------------------------+
> >>>>>>>            | FilterMatch              |
> >>>>>>>            +--------------------------+
> >>>>>>>            | ieId/ieName              |
> >>>>>>>            | ieEnterpriseNumber[0..1] |
> >>>>>>>            | value                    |
> >>>>>>>            +--------------------------+
> >>>>>>>                          Figure 10: Filter classes
> >>>>>>>
> >>>>>>>    The Filter classes in Figure 10 contain the configuration parameters
> >>>>>>>    of specific Filtering methods.  For property match filtering, the
> >>>>>>>    configuration parameters are:
> >>>>>>>
> >>>>>>>    ieId, ieName, ieEnterpriseNumber:  The property to be matched is
> >>>>>>>       specified by either ieId or ieName, specifying the ID or name of
> >>>>>>>       the Information Element, respectively. ieEnterpriseNumber MUST be
> >>>>>>>       used for enterprise-specific Information Elements.  If
> >>>>>>>       ieEnterpriseNumber is omitted or zero, this is Information Element
> >>>>>>>       is not enterprise-specific but registered at IANA.
> >>>>>>>
> >>>>>>>    value:  Matching value.
> >>>>>>>
> >>>>>> Can the ranges of value be defined like hash value?
> >>>>> No.  The reason is that min/max fields are not specified in the Selector 
> >>>>> Report Interpretation:
> >>>>> http://tools.ietf.org/html/rfc5476#section-6.5.2.5
> >>>>>
> >>>>> In the case of FilterHash, we do min/max fields in the Selector Report 
> >>>>> Interpretation:
> >>>>> http://tools.ietf.org/html/rfc5476#section-6.5.2.6
> >>>>>
> >>>>>> And, I would like to specify its instruction, "match" or "unmatch".
> >>>>> This is not specified in the Selector Report Interpretation. Therefore, 
> >>>>> we cannot allow configuring this.
> >>>>>
> >>>> Above all, they apparently would be required from viewpoint of usability.
> >>> I agree. But this is not the document to solve this problem.
> >>>
> >>>> Please see http://tools.ietf.org/html/rfc5475#section-6.1 that described
> >>>> it as an extension of the basic model.
> >>>>
> >>>> Implementor will add above functions because only exact match is too
> >>>> poor. In that case, can IPFIX-CONFIG expand to add these functions? 
> >>>> How does it expand?
> >>> The model can be extensible by vendors or in later revisions of the model.
> >>>
> >> Ok, I understood what you would like to say. I am concerned that the
> >> extensions are built indiscriminately.
> 
> To summarize this issue: PSAMP protocol only specifies simplistic match 
> filters, and the current configuration data model supports the 
> configuration of such filters.
> 
> If the IPFIX WG wants the configuration data model to offer more 
> configuration possibilities than simple match filters, the WG must be 
> aware of the fact that this Selector configuration cannot be reported 
> within a Selector Report Interpretation.
> 
> Therfore, I will not extend the configuration data model unless we have 
> reached consensus within the IPFIX WG that this restriction is not a 
> problem.
> 

To keep consistency for these documents, it seems to be better to remain
this restriction. I hope that "match", "unmatch", and range filter can
be applied to it in near future. A document for these expansions would
cover the CONFIG-MODEL and selector representation using structured data.

I would like to know other IPFIX member's comments, could you please
bring up this topics in next meeting.

> 
> 2) sourceIpAddress of an Exporting Process
> 
> >>>>>>>    sourceIpAddress:  This parameter sets the source IP address.  If this
> >>>>>>>       parameter is omitted, the address assigned to the outgoing
> >>>>>>>       interface is used.
> >>>>>>>
> >>>>>> I would like to make available in the case of TCP.
> >>>>> We adopted this parameter because the Cisco CLI has it for UDP.  As far 
> >>>>> as I understand, it is not used to specify the outgoing interface but 
> >>>>> the (forged) source address in the IP header only.
> >>>>>
> >>>> No. In current network operation, source IP address is important
> >>>> parameter as a indicator of exporting device(router). Operators are
> >>>> often confused that a source IP address is changed when route changes
> >>>> happen. Exporting device may select source address as a optimal exit points, or
> >>>> arbitrarily.  Then, some collectors may fail to recognize identical
> >>>> exporting device before and after route changes.
> >>>  >
> >>>> This function is necessary to avoid the confusion.
> >>> I will ask Benoit how this parameter is used.
> >>>
> >>>>> How do you want to forge the IP address of a TCP endpoint?  It must 
> >>>>> belong to the interface.
> >>>>>
> >>>>> The parameter description remains unclear in the case of multiple 
> >>>>> addresses assigned to a single interface.
> >>>>>
> >>>>> I'm not sure what to do.
> >>>>> We can remove the parameter.
> >>>>> We can change to "...one of the addresses assigned to the outgoing 
> >>>>> interface is used."
> >>>>> We can try to find a generic solution that defines the outgoing interface.
> >>>> I propose the following.
> >>>>
> >>>> The following parameters are applicable if TCP or UDP is transport
> >>>> protocol:
> >>>>
> >>>> sourceIpAddress:  This parameter sets the source IP address. 
> >>>> If this parameter is omitted, Exporter selects outgoing interface
> >>>> according to routing policy, and then arbitrarily assigns the address.
> >>> You cannot use an arbitrary source IP address with TCP.
> >>>
> >> I mean exporting device may select appropriate source ip
> >> address as TCP client along with routing policy if the parameter is
> >> omitted. The routing policy may change between before and after its
> >> shutdown or link-failure, etc. This may cause in the case of TCP.
> >>
> >> In simple example, if you can use source IP address on loopback, it
> >> avoid this problem.
> 
> After discussing this issue with you and Paul, I've learned that the 
> source IP address parameter can be configured independently from the 
> interface.
> 
> A note about how this works on *ix systems:
> The IP address is configured with bind() while the interface is 
> configured with socket option SO_BINDTODEVICE. In the case of SCTP, 
> multiple source IP can be bound using sctp_bindx().
> 
> So, I propose to have the parameter sourceIpAddress for all transport 
> protocols. In the case of SCTP, it replaces localIPAddress and thus may 
> appear more than once.
> In addition, I propose to add parameters ifIndex/ifName to optionally 
> specify the interface.
> 
> NEW:
> 
> 4.4.1.  Destination Class
> 
>     +------------------------------------------------+
>     | Destination                                    |
>     +------------------------------------------------+
>     | name                                           |<>-----+
>     | exportMemberType = parallel                    |       | 0..1
>     | ipfixVersion = 10                              |       |
>     | transportProtocol                              |   +------------+
> => | sourceIpAddress[0..*] {[0..1] for UDP and TCP} |   | Transport- |
>     | destinationIpAddress                           |   | Layer-     |
>     | destinationPort = 4739|4740                    |   | Security   |
> => | ifName/ifIndex[0..1]                           |   +------------+
>     | sendBufferSize {opt.}                          |
>     | rateLimit[0..1]                                |
>     | timedReliability = 0 {SCTP only}               |
>     | numberOfStreams {opt.} {SCTP only}             |
>     | templateRefreshTimeout = 600 {UDP only}        |
>     | optionsTemplateRefreshTimeout = 600 {UDP only} |
>     | templateRefreshPacket[0..1] {UDP only}         |
>     | optionsTemplateRefreshPacket[0..1] {UDP only}  |
>     +------------------------------------------------+
>                         Figure 14: Destination class
> 
> 
>     sourceIpAddress:  This parameter specifies the source IP address(es)
>        used by the Exporting Process.
>        If the transport protocol is SCTP, this parameter MAY appear
>        multiple times to specify the list of eligible local IP addresses
>        of a multi-homed Exporting Process [RFC4960].  If this parameter
>        is omitted and the transport protocol is SCTP, all locally
>        assigned IP addresses are eligible local IP addresses.
>        If the transport protocol is UDP or TCP, this parameter MAY appear
>        only once. If this parameter is omitted and the transport protocol
>        is UDP or TCP, the IP address assigned to the outgoing interface
>        is used as source IP address.
> 
>     ifIndex/ifName:  Either the index or the name of the interface used
>        by the Exporting Process to export IPFIX messages to the given
>        destination MAY be specified according to corresponding objects in
>        the IF-MIB <xref target="RFC2863"/>.  If omitted, the Exporting
>        Process selects the outgoing interface based on local routing
>        decision and accepts return traffic, such as transport layer
>        acknowledgments, on all available interfaces.
> 
> Do you agree with these changes?

Almost OK, but there are two questions.
I wonder why you need ifName? I think only sourceIpAddress is enough.
And also, when ifName,ifIndex, and sourceIpAddr are set in CONFIG, which
one is applied to source ip address?

Regards,
Atsushi
> 
> Regards,
> Gerhard
> 
> 



From muenz@net.in.tum.de  Tue Mar  2 05:56:49 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 95EB33A8A08 for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 05:56:49 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[AWL=0.000,  BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id rcaaWk6eBNE6 for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 05:56:48 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id A1ABF3A8963 for <ipfix@ietf.org>; Tue,  2 Mar 2010 05:56:47 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 48C35227BF90; Tue,  2 Mar 2010 14:56:44 +0100 (CET)
Message-ID: <4B8D1926.8030905@net.in.tum.de>
Date: Tue, 02 Mar 2010 14:56:54 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Atsushi Kobayashi <akoba@nttv6.net>
References: <20100209010851.F028.17391CF2@nttv6.net> <4B8B7E20.1020504@net.in.tum.de> <20100302203953.40CF.17391CF2@nttv6.net>
In-Reply-To: <20100302203953.40CF.17391CF2@nttv6.net>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms070308000401030202010008"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] WGLC comments on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 02 Mar 2010 13:56:49 -0000

This is a cryptographically signed message in MIME format.

--------------ms070308000401030202010008
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Atsushi,

Please see inline.

Atsushi Kobayashi wrote:
>> 1) Match Filters
>>
>>>>>>>>> 4.2.3.  Filter Classes
>>>>>>>>>
>>>>>>>>>            +--------------------------+
>>>>>>>>>            | FilterMatch              |
>>>>>>>>>            +--------------------------+
>>>>>>>>>            | ieId/ieName              |
>>>>>>>>>            | ieEnterpriseNumber[0..1] |
>>>>>>>>>            | value                    |
>>>>>>>>>            +--------------------------+
>>>>>>>>>                          Figure 10: Filter classes
>>>>>>>>>
>>>>>>>>>    The Filter classes in Figure 10 contain the configuration pa=
rameters
>>>>>>>>>    of specific Filtering methods.  For property match filtering=
, the
>>>>>>>>>    configuration parameters are:
>>>>>>>>>
>>>>>>>>>    ieId, ieName, ieEnterpriseNumber:  The property to be matche=
d is
>>>>>>>>>       specified by either ieId or ieName, specifying the ID or =
name of
>>>>>>>>>       the Information Element, respectively. ieEnterpriseNumber=
 MUST be
>>>>>>>>>       used for enterprise-specific Information Elements.  If
>>>>>>>>>       ieEnterpriseNumber is omitted or zero, this is Informatio=
n Element
>>>>>>>>>       is not enterprise-specific but registered at IANA.
>>>>>>>>>
>>>>>>>>>    value:  Matching value.
>>>>>>>>>
>>>>>>>> Can the ranges of value be defined like hash value?
>>>>>>> No.  The reason is that min/max fields are not specified in the S=
elector=20
>>>>>>> Report Interpretation:
>>>>>>> http://tools.ietf.org/html/rfc5476#section-6.5.2.5
>>>>>>>
>>>>>>> In the case of FilterHash, we do min/max fields in the Selector R=
eport=20
>>>>>>> Interpretation:
>>>>>>> http://tools.ietf.org/html/rfc5476#section-6.5.2.6
>>>>>>>
>>>>>>>> And, I would like to specify its instruction, "match" or "unmatc=
h".
>>>>>>> This is not specified in the Selector Report Interpretation. Ther=
efore,=20
>>>>>>> we cannot allow configuring this.
>>>>>>>
>>>>>> Above all, they apparently would be required from viewpoint of usa=
bility.
>>>>> I agree. But this is not the document to solve this problem.
>>>>>
>>>>>> Please see http://tools.ietf.org/html/rfc5475#section-6.1 that des=
cribed
>>>>>> it as an extension of the basic model.
>>>>>>
>>>>>> Implementor will add above functions because only exact match is t=
oo
>>>>>> poor. In that case, can IPFIX-CONFIG expand to add these functions=
?=20
>>>>>> How does it expand?
>>>>> The model can be extensible by vendors or in later revisions of the=
 model.
>>>>>
>>>> Ok, I understood what you would like to say. I am concerned that the=

>>>> extensions are built indiscriminately.
>> To summarize this issue: PSAMP protocol only specifies simplistic matc=
h=20
>> filters, and the current configuration data model supports the=20
>> configuration of such filters.
>>
>> If the IPFIX WG wants the configuration data model to offer more=20
>> configuration possibilities than simple match filters, the WG must be =

>> aware of the fact that this Selector configuration cannot be reported =

>> within a Selector Report Interpretation.
>>
>> Therfore, I will not extend the configuration data model unless we hav=
e=20
>> reached consensus within the IPFIX WG that this restriction is not a=20
>> problem.
>>
>=20
> To keep consistency for these documents, it seems to be better to remai=
n
> this restriction. I hope that "match", "unmatch", and range filter can
> be applied to it in near future. A document for these expansions would
> cover the CONFIG-MODEL and selector representation using structured dat=
a.
>=20
> I would like to know other IPFIX member's comments, could you please
> bring up this topics in next meeting.

Yes, I can put this on the slides.

>> 2) sourceIpAddress of an Exporting Process
>>
>>>>>>>>>    sourceIpAddress:  This parameter sets the source IP address.=
  If this
>>>>>>>>>       parameter is omitted, the address assigned to the outgoin=
g
>>>>>>>>>       interface is used.
>>>>>>>>>
>>>>>>>> I would like to make available in the case of TCP.
>>>>>>> We adopted this parameter because the Cisco CLI has it for UDP.  =
As far=20
>>>>>>> as I understand, it is not used to specify the outgoing interface=
 but=20
>>>>>>> the (forged) source address in the IP header only.
>>>>>>>
>>>>>> No. In current network operation, source IP address is important
>>>>>> parameter as a indicator of exporting device(router). Operators ar=
e
>>>>>> often confused that a source IP address is changed when route chan=
ges
>>>>>> happen. Exporting device may select source address as a optimal ex=
it points, or
>>>>>> arbitrarily.  Then, some collectors may fail to recognize identica=
l
>>>>>> exporting device before and after route changes.
>>>>>  >
>>>>>> This function is necessary to avoid the confusion.
>>>>> I will ask Benoit how this parameter is used.
>>>>>
>>>>>>> How do you want to forge the IP address of a TCP endpoint?  It mu=
st=20
>>>>>>> belong to the interface.
>>>>>>>
>>>>>>> The parameter description remains unclear in the case of multiple=
=20
>>>>>>> addresses assigned to a single interface.
>>>>>>>
>>>>>>> I'm not sure what to do.
>>>>>>> We can remove the parameter.
>>>>>>> We can change to "...one of the addresses assigned to the outgoin=
g=20
>>>>>>> interface is used."
>>>>>>> We can try to find a generic solution that defines the outgoing i=
nterface.
>>>>>> I propose the following.
>>>>>>
>>>>>> The following parameters are applicable if TCP or UDP is transport=

>>>>>> protocol:
>>>>>>
>>>>>> sourceIpAddress:  This parameter sets the source IP address.=20
>>>>>> If this parameter is omitted, Exporter selects outgoing interface
>>>>>> according to routing policy, and then arbitrarily assigns the addr=
ess.
>>>>> You cannot use an arbitrary source IP address with TCP.
>>>>>
>>>> I mean exporting device may select appropriate source ip
>>>> address as TCP client along with routing policy if the parameter is
>>>> omitted. The routing policy may change between before and after its
>>>> shutdown or link-failure, etc. This may cause in the case of TCP.
>>>>
>>>> In simple example, if you can use source IP address on loopback, it
>>>> avoid this problem.
>> After discussing this issue with you and Paul, I've learned that the=20
>> source IP address parameter can be configured independently from the=20
>> interface.
>>
>> A note about how this works on *ix systems:
>> The IP address is configured with bind() while the interface is=20
>> configured with socket option SO_BINDTODEVICE. In the case of SCTP,=20
>> multiple source IP can be bound using sctp_bindx().
>>
>> So, I propose to have the parameter sourceIpAddress for all transport =

>> protocols. In the case of SCTP, it replaces localIPAddress and thus ma=
y=20
>> appear more than once.
>> In addition, I propose to add parameters ifIndex/ifName to optionally =

>> specify the interface.
>>
>> NEW:
>>
>> 4.4.1.  Destination Class
>>
>>     +------------------------------------------------+
>>     | Destination                                    |
>>     +------------------------------------------------+
>>     | name                                           |<>-----+
>>     | exportMemberType =3D parallel                    |       | 0..1
>>     | ipfixVersion =3D 10                              |       |
>>     | transportProtocol                              |   +------------=
+
>> =3D> | sourceIpAddress[0..*] {[0..1] for UDP and TCP} |   | Transport-=
 |
>>     | destinationIpAddress                           |   | Layer-     =
|
>>     | destinationPort =3D 4739|4740                    |   | Security =
  |
>> =3D> | ifName/ifIndex[0..1]                           |   +-----------=
-+
>>     | sendBufferSize {opt.}                          |
>>     | rateLimit[0..1]                                |
>>     | timedReliability =3D 0 {SCTP only}               |
>>     | numberOfStreams {opt.} {SCTP only}             |
>>     | templateRefreshTimeout =3D 600 {UDP only}        |
>>     | optionsTemplateRefreshTimeout =3D 600 {UDP only} |
>>     | templateRefreshPacket[0..1] {UDP only}         |
>>     | optionsTemplateRefreshPacket[0..1] {UDP only}  |
>>     +------------------------------------------------+
>>                         Figure 14: Destination class
>>
>>
>>     sourceIpAddress:  This parameter specifies the source IP address(e=
s)
>>        used by the Exporting Process.
>>        If the transport protocol is SCTP, this parameter MAY appear
>>        multiple times to specify the list of eligible local IP address=
es
>>        of a multi-homed Exporting Process [RFC4960].  If this paramete=
r
>>        is omitted and the transport protocol is SCTP, all locally
>>        assigned IP addresses are eligible local IP addresses.
>>        If the transport protocol is UDP or TCP, this parameter MAY app=
ear
>>        only once. If this parameter is omitted and the transport proto=
col
>>        is UDP or TCP, the IP address assigned to the outgoing interfac=
e
>>        is used as source IP address.
>>
>>     ifIndex/ifName:  Either the index or the name of the interface use=
d
>>        by the Exporting Process to export IPFIX messages to the given
>>        destination MAY be specified according to corresponding objects=
 in
>>        the IF-MIB <xref target=3D"RFC2863"/>.  If omitted, the Exporti=
ng
>>        Process selects the outgoing interface based on local routing
>>        decision and accepts return traffic, such as transport layer
>>        acknowledgments, on all available interfaces.
>>
>> Do you agree with these changes?
>=20
> Almost OK, but there are two questions.
> I wonder why you need ifName? I think only sourceIpAddress is enough.
> And also, when ifName,ifIndex, and sourceIpAddr are set in CONFIG, whic=
h
> one is applied to source ip address?

This is something which confused my as well, but Paul put me on the=20
right track.

I try to explain:

If you do not bind the socket to a specific (link-layer) interface, the=20
OS will look into the routing table and choose the outgoing interface=20
with the lowest routing costs to the destination. The source IP address=20
does not matter at all in this decision. Even in the case of TCP,=20
asymmetric routing may lead to the effect that outgoing packets are sent =

on one router interface and incoming packets are received on another=20
interface. This is a valid TCP connection.

As administrator, you want to enforce that IPFIX messages leave the=20
router on the management interface, independently of the routing table.=20
This is done with ifName/ifIndex.

Is it clear now?

Regards,
Gerhard

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms070308000401030202010008
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms070308000401030202010008--

From Thomas.Dietz@nw.neclab.eu  Tue Mar  2 07:16:38 2010
Return-Path: <Thomas.Dietz@nw.neclab.eu>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id AB2C03A8B3A for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 07:16:38 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id hG2G+nMnbYLG for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 07:16:37 -0800 (PST)
Received: from smtp0.neclab.eu (smtp0.neclab.eu [195.37.70.41]) by core3.amsl.com (Postfix) with ESMTP id E61C53A8B34 for <ipfix@ietf.org>; Tue,  2 Mar 2010 07:16:36 -0800 (PST)
Received: from localhost (localhost.localdomain [127.0.0.1]) by smtp0.neclab.eu (Postfix) with ESMTP id A73A32C00C9A7 for <ipfix@ietf.org>; Tue,  2 Mar 2010 16:16:37 +0100 (CET)
X-Virus-Scanned: Amavisd on Debian GNU/Linux (atlas2.office)
Received: from smtp0.neclab.eu ([127.0.0.1]) by localhost (atlas2.office [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id NKQmgEVLNzAe for <ipfix@ietf.org>; Tue,  2 Mar 2010 16:16:37 +0100 (CET)
Received: from VENUS.office (mx1.office [192.168.24.3]) by smtp0.neclab.eu (Postfix) with ESMTP id 8C7BC2C000350 for <ipfix@ietf.org>; Tue,  2 Mar 2010 16:16:32 +0100 (CET)
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Date: Tue, 2 Mar 2010 16:16:31 +0100
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=SHA1; boundary="----=_NextPart_000_014B_01CABA23.B87AFCD0"
Message-ID: <547F018265F92642B577B986577D671C01235205@VENUS.office>
In-Reply-To: <20100301174502.2304028C42A@core3.amsl.com>
X-MS-Has-Attach: yes
X-MS-TNEF-Correlator: 
Thread-Topic: [IPFIX] I-D ACTION:draft-ietf-ipfix-psamp-mib-00.txt
Thread-Index: Acq5ZwW7tf3Jt9nnQ72/4A2S5hbjSgAs18gg
References: <20100301174502.2304028C42A@core3.amsl.com>
From: "Thomas Dietz" <Thomas.Dietz@nw.neclab.eu>
To: <ipfix@ietf.org>
Subject: Re: [IPFIX] I-D ACTION:draft-ietf-ipfix-psamp-mib-00.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 02 Mar 2010 15:16:38 -0000

This is a multi-part message in MIME format.

------=_NextPart_000_014B_01CABA23.B87AFCD0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Dear all,

Yesterday I posted a new draft draft-ietf-ipfix-psamp-mib-00.txt. Well, =
this
draft is not really new. It is the successor of =
draft-ietf-psamp-mib-06.txt
which has expired and since the PSAMP WG was closed had to move to the =
IPFIX
WG now.

The draft has changed considerably, especially since the IPFIX MIB draft =
is
now ready and in the RFC editor queue. The new PSAMP MIB draft reflects =
the
decisions taken in the IPFIX MIB draft. All configuration objects have =
gone.
The Selection and Filtering functions have moved (OID wise) into the =
IPFIX
SELECTOR MIB OID space. The objects were aligned to the RFC5477 and =
RFC5474.

Of course there is still some work to do. So all comments and =
suggestions
are appreciated.

Best Regards,

Thomas

--=20
Thomas Dietz                 E-mail: Thomas.Dietz@nw.neclab.eu
NEC Europe Ltd.              Phone:  +49 6221 4342-128
NEC Laboratories Europe      Fax:    +49 6221 4342-155
Network Research Division
Kurfuersten-Anlage 36
69115 Heidelberg, Germany    http://www.nw.neclab.eu

NEC Europe Limited           Registered in England 2832014
Registered Office: NEC House, 1 Victoria Road, London W3 6BL

> -----Original Message-----
> From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] On Behalf
> Of Internet-Drafts@ietf.org
> Sent: Montag, 1. M=E4rz 2010 18:45
> To: i-d-announce@ietf.org
> Cc: ipfix@ietf.org
> Subject: [IPFIX] I-D ACTION:draft-ietf-ipfix-psamp-mib-00.txt
>=20
> A New Internet-Draft is available from the on-line Internet-Drafts
> directories.
> This draft is a work item of the IP Flow Information Export Working
> Group of the IETF.
>=20
> 	Title		: Definitions of Managed Objects for Packet Sampling
> 	Author(s)	: T. Dietz, B. Claise, J. Quittek
> 	Filename	: draft-ietf-ipfix-psamp-mib-00.txt
> 	Pages		: 34
> 	Date		: 2010-3-1
>=20
>    This memo defines a portion of the Management Information Base =
(MIB)
>    for use with network management protocols in the Internet =
community.
>    In particular, it describes extensions to the IPFIX MIB module
>    [I-D.ietf-ipfix-mib].  For IPFIX implementations that use packet
>    sampling (PSAMP) techniques as described in [RFC5475], this memo
>    defines the PSAMP MIB module containing managed objects for
> providing
>    information on applied packet selection functions and their
>    parameters.
>=20
> A URL for this Internet-Draft is:
> http://www.ietf.org/internet-drafts/draft-ietf-ipfix-psamp-mib-00.txt
>=20
> Internet-Drafts are also available by anonymous FTP at:
> ftp://ftp.ietf.org/internet-drafts/
>=20
> Below is the data which will enable a MIME compliant mail reader
> implementation to automatically retrieve the ASCII version of the
> Internet-Draft.

------=_NextPart_000_014B_01CABA23.B87AFCD0
Content-Type: application/x-pkcs7-signature;
	name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment;
	filename="smime.p7s"
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------=_NextPart_000_014B_01CABA23.B87AFCD0--

From n.brownlee@auckland.ac.nz  Tue Mar  2 17:56:59 2010
Return-Path: <n.brownlee@auckland.ac.nz>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 7BD0C3A8BFA for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 17:56:59 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.599
X-Spam-Level: 
X-Spam-Status: No, score=-6.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id JoIvxVg9JAmM for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 17:56:58 -0800 (PST)
Received: from mailhost.auckland.ac.nz (larry.its.auckland.ac.nz [130.216.12.34]) by core3.amsl.com (Postfix) with ESMTP id 7F6633A8BEF for <ipfix@ietf.org>; Tue,  2 Mar 2010 17:56:57 -0800 (PST)
Received: from localhost (localhost.localdomain [127.0.0.1]) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 438361BE64; Wed,  3 Mar 2010 14:56:57 +1300 (NZDT)
X-Virus-Scanned: by amavisd-new at mailhost.auckland.ac.nz
Received: from mailhost.auckland.ac.nz ([127.0.0.1]) by localhost (larry.its.auckland.ac.nz [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 5lSfUMlTunQ3; Wed,  3 Mar 2010 14:56:57 +1300 (NZDT)
Received: from nevil-laptop.sfac.auckland.ac.nz (nevil-laptop.sfac.auckland.ac.nz [130.216.38.130]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 38D761BE60; Wed,  3 Mar 2010 14:56:48 +1300 (NZDT)
Message-ID: <4B8DC1E0.4070909@auckland.ac.nz>
Date: Wed, 03 Mar 2010 14:56:48 +1300
From: Nevil Brownlee <n.brownlee@auckland.ac.nz>
User-Agent: Thunderbird 2.0.0.12 (Macintosh/20080213)
MIME-Version: 1.0
To: Atsushi Kobayashi <akoba@nttv6.net>
References: <4B874A68.2040606@auckland.ac.nz> <20100226132551.405B.17391CF2@nttv6.net>
In-Reply-To: <20100226132551.405B.17391CF2@nttv6.net>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] DRAFT agenda for IETF in Anaheim
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 03 Mar 2010 01:56:59 -0000

Hi Atsushi:

1) OK, I've added this to the Anaheim agenda.

2) Your draft says 'Standards Track,' but independent submissions can
    only be Informational or Experimental.  Standards Track drafts need
    to become a WG item ... let's discuss it further in Anaheim.

3) As you've reminded me below, I'll post some comments real soon now!

Cheers, Nevil


Atsushi Kobayashi wrote:
> Hi Nevil, and all,
> 
> We are preparing the next version draft of new information element for
> link layer monitoring. 
> http://tools.ietf.org/id/draft-kashima-ipfix-data-link-layer-monitoring-01.txt
> 
> In last meeting, we discussed how to progress it furthermore.
> Please see minute of last meeting.
> 
>           New Drafts
>              Data-link Layer Monitoring, Shingo Kashima.
>                Proposes to make 3 new IEs?
>                dataLinkFrameType - why not use ifType (from RMON) for values.
>                Juergen says "since Ethernet is so common, this should be
>                documented in an RFC."  Could be an Independent Submission.
>                Dan agrees we need a more detailed draft, discussed on WG.
>                Nevil will send some comments to the authors.
> 
> I and Kashima-san would like to have 5 or 10 minutes to present the
> update in IPFIX WG. But, I am not sure how to progress it. Should it be
> presented at another working group, or independent submission? 
> Will you give me any suggestion about it?
> 
> Regards,
> Atsushi

-- 
---------------------------------------------------------------------
  Nevil Brownlee                    Computer Science Department | ITS
  Phone: +64 9 373 7599 x88941             The University of Auckland
  FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand

From n.brownlee@auckland.ac.nz  Tue Mar  2 18:39:19 2010
Return-Path: <n.brownlee@auckland.ac.nz>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 0A8B23A8CAB for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 18:39:19 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.599
X-Spam-Level: 
X-Spam-Status: No, score=-6.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id qjMRmZjqPclL for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 18:39:18 -0800 (PST)
Received: from mailhost.auckland.ac.nz (larry.its.auckland.ac.nz [130.216.12.34]) by core3.amsl.com (Postfix) with ESMTP id 28F3E3A89FA for <ipfix@ietf.org>; Tue,  2 Mar 2010 18:39:17 -0800 (PST)
Received: from localhost (localhost.localdomain [127.0.0.1]) by mailhost.auckland.ac.nz (Postfix) with ESMTP id D09E41BC08 for <ipfix@ietf.org>; Wed,  3 Mar 2010 15:39:08 +1300 (NZDT)
X-Virus-Scanned: by amavisd-new at mailhost.auckland.ac.nz
Received: from mailhost.auckland.ac.nz ([127.0.0.1]) by localhost (larry.its.auckland.ac.nz [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id UAQUU1yw0vjp for <ipfix@ietf.org>; Wed,  3 Mar 2010 15:39:08 +1300 (NZDT)
Received: from nevil-laptop.sfac.auckland.ac.nz (nevil-laptop.sfac.auckland.ac.nz [130.216.38.130]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mailhost.auckland.ac.nz (Postfix) with ESMTP id AD8181BC04 for <ipfix@ietf.org>; Wed,  3 Mar 2010 15:39:08 +1300 (NZDT)
Message-ID: <4B8DCBCC.5060900@auckland.ac.nz>
Date: Wed, 03 Mar 2010 15:39:08 +1300
From: Nevil Brownlee <n.brownlee@auckland.ac.nz>
User-Agent: Thunderbird 2.0.0.12 (Macintosh/20080213)
MIME-Version: 1.0
To: IPFIX Working Group <ipfix@ietf.org>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Subject: [IPFIX] Anaheim Agenda - second draft
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 03 Mar 2010 02:39:19 -0000

Hi all:

Thanks for all the feedback on my DRAFT agenda.
The next version is now available at

     http://www.ietf.org/proceedings/10mar/agenda/ipfix.txt

Please send me any further comments/suggestions/requests for
time slots.

Cheers, Nevil


-- 
---------------------------------------------------------------------
  Nevil Brownlee                    Computer Science Department | ITS
  Phone: +64 9 373 7599 x88941             The University of Auckland
  FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand

From n.brownlee@auckland.ac.nz  Tue Mar  2 18:43:14 2010
Return-Path: <n.brownlee@auckland.ac.nz>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 0C06D3A8CAB for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 18:43:14 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.599
X-Spam-Level: 
X-Spam-Status: No, score=-6.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Rike97mOTVq5 for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 18:43:12 -0800 (PST)
Received: from mailhost.auckland.ac.nz (larry.its.auckland.ac.nz [130.216.12.34]) by core3.amsl.com (Postfix) with ESMTP id A1E373A8C8E for <ipfix@ietf.org>; Tue,  2 Mar 2010 18:43:12 -0800 (PST)
Received: from localhost (localhost.localdomain [127.0.0.1]) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 236701BC08 for <ipfix@ietf.org>; Wed,  3 Mar 2010 15:43:13 +1300 (NZDT)
X-Virus-Scanned: by amavisd-new at mailhost.auckland.ac.nz
Received: from mailhost.auckland.ac.nz ([127.0.0.1]) by localhost (larry.its.auckland.ac.nz [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id JXVwiIDOmZuf for <ipfix@ietf.org>; Wed,  3 Mar 2010 15:43:13 +1300 (NZDT)
Received: from nevil-laptop.sfac.auckland.ac.nz (nevil-laptop.sfac.auckland.ac.nz [130.216.38.130]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 048E01BC04 for <ipfix@ietf.org>; Wed,  3 Mar 2010 15:43:13 +1300 (NZDT)
Message-ID: <4B8DCCC1.6000906@auckland.ac.nz>
Date: Wed, 03 Mar 2010 15:43:13 +1300
From: Nevil Brownlee <n.brownlee@auckland.ac.nz>
User-Agent: Thunderbird 2.0.0.12 (Macintosh/20080213)
MIME-Version: 1.0
To: IPFIX Working Group <ipfix@ietf.org>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Subject: [IPFIX] WG LC starting for draft-ietf-ipfix-anon-02
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 03 Mar 2010 02:43:14 -0000

Hi all:

The latest revision of the IPFIX Anonymisation draft was published
on 15 Feb 10; it addresses all the issues raised to date, it's now
ready for WG Last Call.

The Last Call for this draft starts today (3 March), it will end
on 20 March.  Please take a look at the draft and post your comments
to this list.

Cheers Nevil
IPFIX co-chair

-- 
---------------------------------------------------------------------
  Nevil Brownlee                    Computer Science Department | ITS
  Phone: +64 9 373 7599 x88941             The University of Auckland
  FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand

From akoba@nttv6.net  Tue Mar  2 20:12:07 2010
Return-Path: <akoba@nttv6.net>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id A695E28C199 for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 20:12:06 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.611
X-Spam-Level: 
X-Spam-Status: No, score=0.611 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HOST_MISMATCH_COM=0.311, RDNS_DYNAMIC=0.1, SUBJ_RE_NUM=2.799]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id NjGUkOxDcd83 for <ipfix@core3.amsl.com>; Tue,  2 Mar 2010 20:12:05 -0800 (PST)
Received: from mail.nttv6.net (mail.nttv6.net [IPv6:2001:fa8::25]) by core3.amsl.com (Postfix) with ESMTP id 6F7663A8A35 for <ipfix@ietf.org>; Tue,  2 Mar 2010 20:12:04 -0800 (PST)
Received: from [127.0.0.1] (dhcp-3-152.nttv6.com [192.47.163.152]) by mail.nttv6.net (8.14.3/8.14.3) with ESMTP id o234Buxv064658; Wed, 3 Mar 2010 13:11:57 +0900 (JST) (envelope-from akoba@nttv6.net)
Date: Wed, 03 Mar 2010 13:07:11 +0900
From: Atsushi Kobayashi <akoba@nttv6.net>
To: Nevil Brownlee <n.brownlee@auckland.ac.nz>
In-Reply-To: <4B8DC1E0.4070909@auckland.ac.nz>
References: <20100226132551.405B.17391CF2@nttv6.net> <4B8DC1E0.4070909@auckland.ac.nz>
Message-Id: <20100303125436.40F3.17391CF2@nttv6.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
X-Mailer: Becky! ver. 2.50.05 [ja] (Unregistered)
X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.2.3 (mail.nttv6.net [192.16.178.5]); Wed, 03 Mar 2010 13:11:57 +0900 (JST)
Cc: IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] DRAFT agenda for IETF in Anaheim
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 03 Mar 2010 04:12:07 -0000

Hi Nevil,

On Wed, 03 Mar 2010 14:56:48 +1300
Nevil Brownlee <n.brownlee@auckland.ac.nz> wrote:

> 
> Hi Atsushi:
> 
> 1) OK, I've added this to the Anaheim agenda.
> 

Thank you! Kashima-san will present it.

> 2) Your draft says 'Standards Track,' but independent submissions can
>     only be Informational or Experimental.  Standards Track drafts need
>     to become a WG item ... let's discuss it further in Anaheim.
> 

Yes, let's discuss also how to progress it.
I do not adhere to "Standards Track". Our goal is to register new
Information Elements for data link frame into IANA. If we achieve it, it
does not matter.

> 3) As you've reminded me below, I'll post some comments real soon now!

I am looking forward to your comments.

Regards,
Atsushi

> 
> Cheers, Nevil
> 
> 
> Atsushi Kobayashi wrote:
> > Hi Nevil, and all,
> > 
> > We are preparing the next version draft of new information element for
> > link layer monitoring. 
> > http://tools.ietf.org/id/draft-kashima-ipfix-data-link-layer-monitoring-01.txt
> > 
> > In last meeting, we discussed how to progress it furthermore.
> > Please see minute of last meeting.
> > 
> >           New Drafts
> >              Data-link Layer Monitoring, Shingo Kashima.
> >                Proposes to make 3 new IEs?
> >                dataLinkFrameType - why not use ifType (from RMON) for values.
> >                Juergen says "since Ethernet is so common, this should be
> >                documented in an RFC."  Could be an Independent Submission.
> >                Dan agrees we need a more detailed draft, discussed on WG.
> >                Nevil will send some comments to the authors.
> > 
> > I and Kashima-san would like to have 5 or 10 minutes to present the
> > update in IPFIX WG. But, I am not sure how to progress it. Should it be
> > presented at another working group, or independent submission? 
> > Will you give me any suggestion about it?
> > 
> > Regards,
> > Atsushi
> 
> -- 
> ---------------------------------------------------------------------
>   Nevil Brownlee                    Computer Science Department | ITS
>   Phone: +64 9 373 7599 x88941             The University of Auckland
>   FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand
> 

--- 
Atsushi KOBAYASHI  <akoba@nttv6.net>
NTT Information Sharing Platform Lab.
tel:+81-(0)422-59-3978 fax:+81-(0)422-59-5637



From braun@net.in.tum.de  Wed Mar  3 00:51:20 2010
Return-Path: <braun@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id CCC523A8D1D for <ipfix@core3.amsl.com>; Wed,  3 Mar 2010 00:51:20 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id jdmZGYqbn2zh for <ipfix@core3.amsl.com>; Wed,  3 Mar 2010 00:51:18 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 90C063A8914 for <ipfix@ietf.org>; Wed,  3 Mar 2010 00:51:18 -0800 (PST)
Received: from honshu.net.in.tum.de (honshu.net.in.tum.de [131.159.20.100]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 793332083FF6; Wed,  3 Mar 2010 09:51:00 +0100 (CET)
Mime-Version: 1.0 (Apple Message framework v1077)
Content-Type: text/plain; charset=iso-8859-1
From: Lothar Braun <braun@net.in.tum.de>
In-Reply-To: <4B8BF243.7090303@net.in.tum.de>
Date: Wed, 3 Mar 2010 09:51:00 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <D71955E5-D24F-4859-B4F0-1C080B8CF192@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu>	<4AD793E9-3463-43F1-8840-13BB6FF0B1E6@net.in.tum.de> <33C32EC7-A6D6-4238-B376-3F91FA3B74D4@net.in.tum.de> <4B8BF243.7090303@net.in.tum.de>
To: Gerhard Muenz <muenz@net.in.tum.de>
X-Mailer: Apple Mail (2.1077)
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] WG last call on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 03 Mar 2010 08:51:20 -0000

Hi,

please find nmy answers inline.

Best regards,
  Lothar

On Mar 1, 2010, at 5:58 PM, Gerhard Muenz wrote:
>>> Cache:  The Cache is a functional block in a Metering Process which
>>> maintains IPFIX Flow Records or PSAMP Packet Reports.  According to =
[RFC5101], the maintenance of Flow Records may include creating new =
records, updating existing ones, computing Flow statistics, deriving =
further Flow properties, detecting Flow expiration, passing Flow Records =
to the Exporting Process, and deleting Flow Records.  The maintenance of =
Packet Reports covers the same set of functions.
>>> Cache Layout:  The Cache Layout defines the superset of fields that
>>> are included in the Packet Reports or Flow Records maintained by the =
Cache.  The fields are specified by the corresponding Information =
Elements.  In general, the largest possible subset of the specified =
fields is derived for every Packet Report or Flow Record.  More specific =
rules about which fields must be included are given in Section 4.3.1.
>>> Cache Mode:  The Cache Mode specifies whether Packet Reports or
>>> Flow Records are generated by the Cache.  In the case of Flow
>>> Records, it also specifies the Flow expiration policy.
>> I had some difficulties to understand the definition of the Cache. I
>> could not understand what the packet cache does when generating PSAMP
>> Packet Reports. It was clear after reading Section 4.3 when the Cache
>> Modes are explained. Maybe there could be some text that explains
>> that maintaining Packet Reports and Flow Records is the same task
>> except that Packet Records expire after the first packet.
>=20
> After Atsushi's review, the text of the Cache definition has been =
changed to:
>=20
>   Cache:  The Cache is a functional block in a Metering Process which
>      generates IPFIX Flow Records or PSAMP Packet Reports from a
>      Selected Packet Stream, in accordance with its configuration.  If
>      Flow Records are generated, the Cache performs tasks like =
creating
>      new records, updating existing ones, computing Flow statistics,
>      deriving further Flow properties, detecting Flow expiration,
>      passing Flow Records to the Exporting Process, and deleting Flow
>      Records.  If Packet Reports are generated, the Cache performs
>      tasks like extracting packet contents and derived packet
>      properties from the Selected Packet Stream, creating new records,
>      and passing them as Packet Reports to the Exporting Process.
>=20
>=20
> Does this text make things clearer?

Yes, this definition is clearer.

> > Could there be an example for a Cache?
>=20
> Not sure what you mean. What kind of example? Where should it appear =
in the text?

I think the new definition is good enough in explaining the functions of =
the cache, so that it should be clear without an example.

>>> +-----------------------------------+ | Metering Process
>>> | | +-----------+ Selected            | Observed | | Selection |
>>> Packet    +-------+ |  Stream of Packet  -->| Process
>>> |---------->| Cache |--> Flow Records or Stream   | +-----------+
>>> Stream    +-------+ |  Packet Reports =
+-----------------------------------+
>>> Figure 1: Selection Process and Cache forming a Metering Process
>>> The configuration data model adopts the separation of Selection =
Processes and Caches in order to support the flexible configuration
>>> and combination of these functional blocks.  As defined in
>>> [RFC5476], the Selection Process takes the Observed Packet Stream
>>> as its input and selects a subset of that stream as its output
>>> (Selected Packet Stream).  The action of the Selection Process on a
>>> single packet of its input is defined by a Primitive Selector or a
>>> Composite Selector. The Cache generates Flow Records or Packet
>>> Reports from the Selected Packet Stream, depending on the
>>> configured Cache Mode.
>>> The Observed Packet Stream at the input of a Selection Process MUST
>>> only contain packets originating from a single Observation Domain.
>>> Similarly, the Selected Packet Stream at the input of a Cache MUST
>>> only contain packets originating from a single Observation Domain.
>>> Packets from Observation Points belonging to different Observation
>>> Domains MUST NOT enter the same Selection Process or the same
>>> Cache.
>> When I read this, I had the feeling that this text does poses some
>> restriction on how I am able to configure my monitor. But it seems
>> like this corresponds to the definition of an Observation Domain:
>> Observation Domain
>> An Observation Domain is the largest set of Observation Points for =
which Flow information can be aggregated by a Metering Process. ...
>> Is there some further restriction that I miss?
>=20
> The text has been changed because of some late changes in IPFIX-MIB =
and because of Atsushi's review.
>=20
> PSAMP defines:
>=20
>   * Observed Packet Stream
>=20
>      The Observed Packet Stream is the set of all packets observed at
>      the Observation Point.
>=20
>   * Selection Process
>=20
>      A Selection Process takes the Observed Packet Stream as its input
>      and selects a subset of that stream as its output.
>=20
> According to my interpretation of these definitions, a Selection =
Process may only processes packets from one Observation Point. =
Therefore, the text needed to be more restrictive.
>=20
> On the other hand, there is no IPFIX/PSAMP document saying that a =
Metering Process cannot process packets of multiple ODs. Therefore, this =
restriction was removed.
>=20
> I propose to add another paragraph in response to your next comment. =
So please find the complete new text below.
>=20
>=20
>>> 3.2.  Exporter Configuration
>>> Figure 4 below shows the main classes of the configuration data
>>> model which are involved in the configuration of an IPFIX or PSAMP =
Exporter.  The role of the classes can be briefly summarized as follows:
>>> o  The ObservationPoint class specifies an Observation Point (i.e.,
>>> an interface or linecard) of the Monitoring Device at which packets =
are captured for traffic measurements.  An object of the =
ObservationPoint class may be associated with one or more objects of the =
SelectionProcess class configuring Selection Processes that process the =
observed packets in parallel.  As long as an ObservationPoint object is =
specified without any references to SelectionProcess objects, the =
Observation Point is not deployed for
>>> traffic measurements.
>>> o  The SelectionProcess class contains the configuration parameters
>>> of a Selection Process.  The Selection Process may be composed of a =
single Selector or a sequence of Selectors, defining a Primitive or =
Composite Selector, respectively.
>>> The Selection Process selects packets from an Observed Packet Stream =
originating from one or multiple Observation Points.
>> Could you please introduce the selectAll policy of the
>> SelectionProcess at this point?
>> When I read this, I was wondering why I cannot connect an Observation
>> Point directly to a Cache, if I do not want to sample or filter
>> packets.
>=20
> I propose to mention this in Section 3 already:
>=20
>               +-----------------------------------+
>               | Metering Process                  |
>               | +-----------+ Selected            |
>      Observed | | Selection | Packet    +-------+ |  Stream of
>      Packet  -->| Process   |---------->| Cache |--> Flow Records or
>      Stream   | +-----------+ Stream    +-------+ |  Packet Reports
>               +-----------------------------------+
>=20
>     Figure 1: Selection Process and Cache forming a Metering Process
>=20
>   The configuration data model adopts the separation of Selection
>   Processes and Caches in order to support the flexible configuration
>   and combination of these functional blocks.  As defined in =
[RFC5476],
>   the Selection Process takes the Observed Packet Stream as its input
>   and selects a subset of that stream as its output (Selected Packet
>   Stream).  The Observed Packet Stream at the input of a Selection
>   Process is composed of packets from a single Observation Point.  The
>   action of the Selection Process on a single packet of its input is
>   defined by a Primitive Selector or a Composite Selector.  The Cache
>   generates Flow Records or Packet Reports from the Selected Packet
>   Stream, depending on the configured Cache Mode.
>=20
>   Every Metering Process contains at least one Selection Process and =
at
>   least one Cache.  If all packets in the Observed Packet Stream are
>   selected without any Filtering or Sampling, this corresponds to a
>   Selection Process implementing a trivial Selector which selects all
>   packets.
>=20
>   Beyond the basic Metering Process shown in Figure 1, the
>   configuration data model enables the specification of more complex
>   configurations.  A Metering Process may be composed of multiple
>   Selection Processes and multiple Caches.  A single Cache may process
>   the output of multiple Selection Processes.  The Selected Packet
>   Stream may be copied to enter several Caches, for example one Cache
>   that generates Flow Records and another Cache that generates Packet
>   Reports.  Selection Processes can also be cascaded, such that the
>   Selected Packet Stream at the output of one Selection Process is
>   passed to another Selection Process.  Cascading Selection Processes
>   can be useful in specific contexts such as the example in
>   Section 7.3.
>=20
> Is this ok?

Yes, I'm fine with this text.

>>> 4.1.  ObservationPoint Class
>>> +-------------------------------+ | ObservationPoint              |
>>> +-------------------------------+          1
>>> +--------------------+ | name
>>> |<>----------| Interface/Linecard | | observationPointId {readOnly}
>>> |            +--------------------+ | observationDomainId
>>> | |                               | 0..*  0..*
>>> +--------------------+ |
>>> |----------->| SelectionProcess   | =
+-------------------------------+            +--------------------+
>>> +------------------+   +----------------------------------+ |
>>> Interface        |   | Linecard                         | =
+------------------+   +----------------------------------+ |
>>> ifIndex/ifName   |   | entPhysicalIndex/entPhysicalName | |
>>> direction =3D both |   | direction =3D both                 | =
+------------------+   +----------------------------------+
>>> Figure 6: ObservationPoint class
>> Both, the Interface and the Linecard, contain the "direction". Is
>> there a reason why it is not a part of the Observation Point if it
>> has to be specified in both cases?
>=20
> Good question which has also been mentioned by Andrew.
>=20
> There is a small advantage: The current position of the parameter =
"direction" makes it easier to add vendor or implementation-specific =
packet sources for which the direction parameter has no meaning, for =
example:
>=20
> +------------------+
> | PcapFile         |
> +------------------+
> | filename         |
> +------------------+

Ok, I can see your point (and agree that it is a good idea to have the =
direction within the linecard/interface class).

>>> 4.3.1.  CacheLayout Class
>>> +--------------+ | CacheLayout  | +--------------+   1..*
>>> +--------------------------+ |              |<>------| CacheField
>>> | |              |        +--------------------------+ |
>>> |        | name                     | |              |        |
>>> ieId/ieName              | |              |        | ieLength
>>> {opt.}          | |              |        |
>>> ieEnterpriseNumber[0..1] | |              |        |
>>> isFlowKey[0..1]          | +--------------+
>>> +--------------------------+
>>> Figure 12: CacheLayout class
>>> A Cache generates and maintains Packet Reports or Flow Records =
containing information that has been extracted from the incoming stream =
of packets.  Using the CacheField class, the CacheLayout
>>> class specifies the superset of fields that are included in the
>>> Packet Reports or Flow Records (see Figure 12).
>>> If Packet Reports are generated (i.e., if Cache Mode is
>>> "immediate"), every field specified by the Cache Layout MUST be
>>> included in the resulting Packet Report unless the corresponding
>>> Information Element is not applicable or cannot be derived from the
>>> content or treatment of the incoming packet.
>> Could you give an example for "not applicable or cannot be derived
>> from"? Maybe something like when a source port needs to be extracted
>> from an ICMP packet.
>=20
> Ok. What about the following new paragraph:
>=20
>   For example, if a field is configured to contain the TCP source port
>   (Information Element tcpSourcePort [RFC5102]), the field MUST be
>   included in all Packet Reports which are related to TCP packets.
>   Obviously, the field value cannot be determined for non-TCP packets.

Yes, this is what I had in mind.

> At the moment, we leave it open how to handle the UDP case. There may =
be very simple Exporters supporting only a single Template per Cache.
> In this case, the source port field can be simply set to zero.

Could it be a good idea to add an optional "default if key does not =
apply to packet" field to the CacheLayout class?=20
This default value could also be taken if the device does not know how =
to retrieve a given IE (e.g. if it is a custom vendor specific ID).

The user who configures the device can then decide which value he want's =
to receive in this case.

Thinking about this a little more I would also add something like this:
=09
	If the Cache does not know how to derive a field and does not =
have=20
	a default value for this field, it MUST reject the =
configuration.

What do you think?

>>> If Flow Records are generated (i.e., if Cache Mode is "timeout" or =
"permanent"), every Flow Key field specified by the Cache Layout
>>> MUST be included as Flow Key in the resulting Flow Record unless
>>> the corresponding Information Element is not applicable or cannot
>>> be derived from the content or treatment of the incoming packet.
>>> Two packets MUST be accounted by different Flow Records if
>>> different subsets of the Flow Key fields are applicable or
>>> derivable.
>> I do not understand this. What do you mean with "different subsets of
>> the Flow Key fields".
>> Is it maybe: "Two packets MUST be accounted by different Flow Records
>> if they differ in at least one Flow Key field."
>=20
> I like this sentence. And it is important that this behavior is well =
understood.
>=20
> Therefore, I propose the following changes:
>=20
>   If Flow Records are generated (i.e., if Cache Mode is "timeout" or
>   "permanent"), every Flow Key field specified by the Cache Layout =
MUST
>   be included as Flow Key in the resulting Flow Record unless the
>   corresponding Information Element is not applicable or cannot be
>   derived from the content or treatment of the incoming packet.  Two
>   packets MUST be accounted by different Flow Records if they differ =
in
>   at least one Flow Key field.  Hence, if a Flow Key field can be
>   determined for one packet but not for the other, the two packets are
>   accounted in different Flow Records.
>=20
>   Every non-key field specified by the Cache Layout MUST be included =
in
>   the resulting Flow Record unless the corresponding Information
>   Element is not applicable or cannot be derived for the given Flow.
>   Two packets are accounted by the same Flow Record if they only =
differ
>   in non-key fields but not in any of the Flow Key fields.  The same
>   applies if one or more non-key fields can be determined for one
>   packet but not for the other.
>=20
>   For example, if a non-key field specifies an Information Element
>   whose value is determined by the first packet observed within a Flow
>   (which is the default rule according to [RFC5102]), this field MUST
>   be included in the resulting Flow Record if it can be determined =
from
>   the first packet of the Flow.
>=20
> Is this ok?

Fine.

>>> ieId, ieName, ieEnterpriseNumber:  These parameters specify a field
>>> by the combination of the Information Element identifier or name, =
and the Information Element enterprise number.  Either ieId or ieName =
MUST be specified. ieEnterpriseNumber MUST be used for =
enterprise-specific Information Elements.  If ieEnterpriseNumber is
>>> omitted or zero, this is Information Element is not enterprise- =
specific but registered at IANA.
>> If I want to configure a custom non-IANA field using the Enterprise
>> ID 1. The field should have the ieID 1. What do I have to configure.
>> Do I need to configure ieID 1, ieEnterpriseNumber 1=20
>=20
> yes.
>=20
>> or do I need to give ieID 32768 (the first bit if the id field) and
>> ieEnterpriseNumber 1?
>=20
> no.
>=20
>> So my question is: Is the enterprise Bit set by the user or is it
>> done by the monitor if he sees the ieEnterpriseNumber field?
>=20
> RFC5102 says:
>=20
>   The value of these identifiers is in the range of 1-32767.  Within
>   this range, Information Element identifier values in the sub-range =
of
>   1-127 are compatible with field types used by NetFlow version 9
>   [RFC3954].
>=20
>   Enterprise-specific Information Element identifiers have the same
>   range of 1-32767, but they are coupled with an additional enterprise
>   identifier.  For enterprise-specific Information Elements,
>   Information Element identifier 0 is also reserved.
>=20
> So, 32768 is not a valid ID.
>=20
> I will restrict the value range of all ieId parameters in YANG

Ok.

>>> ieLength:  This parameter specifies the length of the field in =
octets.  A value of 65535 means that the field is encoded as =
variable-length Information Element.  For Information Elements of =
integer and float type, the field length MAY be set to a smaller value =
than the standard length of the abstract data type if the rules of =
reduced size encoding are fulfilled (see [RFC5101], Section 6.2).  If =
not configured by the user, the field length is set by the Monitoring =
Device.
>> The monitoring device can set the length field if it nows the ID
>> (either IANA or known custom ID). But what if the monitor does not
>> know a specific ID? Should there be a default value (which one?) for
>> unknown IDs or should the device reject such a configuration?
>=20
> If the Monitoring Device does not know the ID, the Cache cannot =
extract or derive the field value from the packets, anyway.

If it cannot derive the field value, it may set the value to 0 or some =
completely random value. This is the same situation as if you try to =
extract a source port from ICMP packets.

> Since the Monitoring Device is obliged to export these fields unless =
they are not available, the device needs some knowledge about the IEs. =
So, I think that the device should be able to set a default length.

It can always set a default length to an arbitrary value (e.g. every =
unknown field is 4 bytes). This could result in the device sending =
arbitrary sized fields with some random value for fields that are =
unknown to the device.

As a user, I would like to have the device reject such a configuration.

> Do you think that we should change the text?

I think there should be something like:=20

	If the IE is unknown to the device and has now length field =
given, it MUST
	reject the configuration.

>> I think the device should reject the configuration.
>=20
>=20
>>> 7.3.  Export of Flow Records and Packet Reports
>>> This configuration example demonstrates the combined export of Flow
>>> Records and Packet Reports for a single Observation Point.  A =
Selection Process (Selection Sequence ID =3D 1) applies random
>>> Sampling to the stream of observed packets.  The output is passed
>>> to a Cache generating Flow Records.  In parallel, the output is
>>> passed to a second Selection Process (Selection Sequence ID =3D 2)
>>> which discards all non-ICMP packets.  A second Cache generates
>>> Packet Reports of the retained ICMP packets.  The output of both
>>> caches is exported to a single Collector using SCTP.
>>> <ipfix xmlns=3D"urn:ietf:params:xml:ns:ietf-ipfix-psamp">
>>> <observationPoint> <name>OP at linecard 3</name> =
<observationDomainId>9876</observationDomainId> <interface> =
<ifIndex>4</ifIndex> <direction>ingress</direction> </interface> =
<selectionProcess>Sampling</selectionProcess> </observationPoint>
>>> <selectionProcess> <name>Sampling</name> =
<selectionSequenceId>1</selectionSequenceId> <selector> <name>Random =
sampler</name> <selectorId>1</selectorId> <sampUniProb> =
<probability>0.01</probability> </sampUniProb> </selector> =
<selectionProcess>ICMP</selectionProcess> <cache>Flow
>>> cache</cache> </selectionProcess>
>>> <selectionProcess> <name>ICMP</name> =
<selectionSequenceId>2</selectionSequenceId> <selector> <name>ICMP
>>> filter</name>
>>> <selectorId>2</selectorId> <filterMatch> <ieId>4</ieId> =
<value>1</value> </filterMatch> </selector> <cache>Packet
>>> cache</cache> </selectionProcess>
>>> <cache> <name>Flow cache</name> <cacheMode>timeout</cacheMode> =
<maxRecords>4096</maxRecords> <activeTimeout>5</activeTimeout> =
<inactiveTimeout>10</inactiveTimeout> <cacheLayout> <cacheField> =
<name>Field 1</name> <ieName>sourceIPv4Address</ieName> <isFlowKey/> =
</cacheField> <cacheField> <name>Field 2</name> =
<ieName>destinationIPv4Address</ieName> <isFlowKey/> </cacheField> =
<cacheField> <name>Field 6</name> <ieName>flowStartMilliSeconds</ieName> =
</cacheField> <cacheField> <name>Field 7</name> =
<ieName>flowEndSeconds</ieName> </cacheField> <cacheField> <name>Field =
8</name> <ieName>octetDeltaCount</ieName> </cacheField> <cacheField> =
<name>Field 9</name> <ieName>packetDeltaCount</ieName> </cacheField> =
</cacheLayout> <exportingProcess>Export</exportingProcess> </cache>
>>> <cache> <name>Packet cache</name>
>>> <cacheMode>immediate</cacheMode> <maxRecords>512</maxRecords> =
<cacheLayout> <cacheField> <name>Field 1</name> <ieId>313</ieId> =
<ieLength>64</ieLength> </cacheField> <cacheField> <name>Field
>>> 2</name> <ieId>154</ieId> </cacheField> </cacheLayout> =
<exportingProcess>Export</exportingProcess> </cache>
>>> <exportingProcess> <name>Export</name> <destination> <name>SCTP
>>> collector</name> <transportProtocol>sctp</transportProtocol> =
<destinationIpAddress>192.0.2.1</destinationIpAddress> =
<timedReliability>0</timedReliability> =
<numberOfStreams>2</numberOfStreams> </destination> </exportingProcess>
>>> </ipfix>
>>> The Observed Packet Stream at the input of the Selection Process =
"ICMP" originates from the Selection Process "Sampling", which thus
>>> constitutes a pseudo Observation Point.  In order to inform the =
Collector about the cascaded Selection Processes, the Exporting Process =
exports two Selection Sequence Report Interpretations as defined in =
[RFC5476], section 6.5.1, including the following
>>> fields:
>> Shouldn't there be something like
>> <option> <selectionSequence/> </option>
>> in the exportingProcess configuration?
>=20
> Yes, that would be useful to configure a PSAMP compliant Monitoring =
Device.
>=20
> I added
>=20
>   <options>
>      <name>Options 1</name>
>      <optionsType>selectionSequence</optionsType>
>      <optionsTimeout>0</optionsTimeout>
>    </options>
>=20
> to examples 7.1 and 7.3.

Fine, thank you.

Best regards,
  Lothar

--
Lothar Braun
Chair for Network Architectures and Services (I8)
Department of Informatics
Technische Universit=E4t M=FCnchen
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18010       Fax: +49 89 289-18033
E-mail: braun@net.in.tum.de=20







From braun@net.in.tum.de  Wed Mar  3 01:33:12 2010
Return-Path: <braun@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 7ABF028C2F3 for <ipfix@core3.amsl.com>; Wed,  3 Mar 2010 01:33:12 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id vJb8UvykINaI for <ipfix@core3.amsl.com>; Wed,  3 Mar 2010 01:33:11 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id A03E428C2C1 for <ipfix@ietf.org>; Wed,  3 Mar 2010 01:33:11 -0800 (PST)
Received: from honshu.net.in.tum.de (honshu.net.in.tum.de [131.159.20.100]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 8D73D21102B4 for <ipfix@ietf.org>; Wed,  3 Mar 2010 10:33:11 +0100 (CET)
From: Lothar Braun <braun@net.in.tum.de>
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable
Date: Wed, 3 Mar 2010 10:33:11 +0100
References: <20100301094119.0F8423A8B2F@core3.amsl.com>
To: IETF IPFIX Working Group <ipfix@ietf.org>
Message-Id: <930BF37B-7E33-440F-88CC-4771705A5C26@net.in.tum.de>
Mime-Version: 1.0 (Apple Message framework v1077)
X-Mailer: Apple Mail (2.1077)
X-Virus-Scanned: ClamAV using ClamSMTP
Subject: [IPFIX] Fwd: New Version Notification for draft-braun-core-compressed-ipfix-00
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 03 Mar 2010 09:33:12 -0000

Hi all,
=20
as you might have seen on the agenda, there is a draft Compressed IPFIX =
for smart meters in constrained networks:

http://tools.ietf.org/id/draft-braun-core-compressed-ipfix-00.txt

This draft replaces draft-schmitt-6lowapp-ipfix-ws and defines a =
modified version of IPFIX that can be used to export sensor data from =
constrained smart meters in 6lowpan networks. We are working on a =
01-version for the meeting (there should only be rather small fixes) and =
we would like to hear your comments on the document.

Best regards,
 Lothar

Begin forwarded message:

> From: IETF I-D Submission Tool <idsubmission@ietf.org>
> Date: March 1, 2010 10:41:19 AM GMT+01:00
> To: braun@net.in.tum.de
> Cc: schmitt@net.in.tum.de,bclaise@cisco.com,carle@net.in.tum.de
> Subject: New Version Notification for =
draft-braun-core-compressed-ipfix-00=20
>=20
>=20
> A new version of I-D, draft-braun-core-compressed-ipfix-00.txt has =
been successfuly submitted by Lothar Braun and posted to the IETF =
repository.
>=20
> Filename:	 draft-braun-core-compressed-ipfix
> Revision:	 00
> Title:		 Compressed IPFIX for smart meters in =
constrained networks
> Creation_date:	 2010-03-01
> WG ID:		 Independent Submission
> Number_of_pages: 25
>=20
> Abstract:
> This document specifies the Compressed IPFIX protocol that serves for
> transmitting measurement data in 6LoWPAN networks [RFC4944].
> Compressed IPFIX is derived from IPFIX [RFC5101] and adopted to the
> needs of constrained networks.  This documents defines how the
> Compressed IPFIX Data and Template Records are transmitted in 6LoWPAN
> networks and how Compressed IPFIX data can be converted into
> uncompressed IPFIX data.
>=20
>=20
>=20
> The IETF Secretariat.
>=20
>=20

--
Lothar Braun
Chair for Network Architectures and Services (I8)
Department of Informatics
Technische Universit=E4t M=FCnchen
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18010       Fax: +49 89 289-18033
E-mail: braun@net.in.tum.de=20







From muenz@net.in.tum.de  Wed Mar  3 01:36:37 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id A3ABA3A8D31 for <ipfix@core3.amsl.com>; Wed,  3 Mar 2010 01:36:37 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id v31GDG9Jr41f for <ipfix@core3.amsl.com>; Wed,  3 Mar 2010 01:36:36 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 2A72D3A8D2B for <ipfix@ietf.org>; Wed,  3 Mar 2010 01:36:36 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 4D9EC21102B4; Wed,  3 Mar 2010 10:36:35 +0100 (CET)
Message-ID: <4B8E2DAD.2010308@net.in.tum.de>
Date: Wed, 03 Mar 2010 10:36:45 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Lothar Braun <braun@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu>	<4AD793E9-3463-43F1-8840-13BB6FF0B1E6@net.in.tum.de> <33C32EC7-A6D6-4238-B376-3F91FA3B74D4@net.in.tum.de> <4B8BF243.7090303@net.in.tum.de> <D71955E5-D24F-4859-B4F0-1C080B8CF192@net.in.tum.de>
In-Reply-To: <D71955E5-D24F-4859-B4F0-1C080B8CF192@net.in.tum.de>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms080408050708000604010006"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] WG last call on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 03 Mar 2010 09:36:37 -0000

This is a cryptographically signed message in MIME format.

--------------ms080408050708000604010006
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Lothar,

Ok, we have an open issue regarding the Cache Layout, which is important.=


So, other IPFIX people are encouraged to also read this e-mail and=20
comment :-)

It's about section 4.3.1, which looks like this after some revisions:

> 4.3.1.  CacheLayout Class
>=20
>             +--------------+
>             | CacheLayout  |
>             +--------------+   1..* +--------------------------+
>             |              |<>------| CacheField               |
>             |              |        +--------------------------+
>             |              |        | name                     |
>             |              |        | ieId/ieName              |
>             |              |        | ieLength {opt.}          |
>             |              |        | ieEnterpriseNumber[0..1] |
>             |              |        | isFlowKey[0..1]          |
>             +--------------+        +--------------------------+
>=20
>                        Figure 12: CacheLayout class
>=20
>    A Cache generates and maintains Packet Reports or Flow Records
>    containing information that has been extracted from the incoming
>    stream of packets.  Using the CacheField class, the CacheLayout clas=
s
>    specifies the superset of fields that are included in the Packet
>    Reports or Flow Records (see Figure 12).
>=20
>    If Packet Reports are generated (i.e., if Cache Mode is "immediate")=
,
>    every field specified by the Cache Layout MUST be included in the
>    resulting Packet Report unless the corresponding Information Element=

>    is not applicable or cannot be derived from the content or treatment=

>    of the incoming packet.
>=20
>    For example, if a field is configured to contain the TCP source port=

>    (Information Element tcpSourcePort [RFC5102]), the field MUST be
>    included in all Packet Reports which are related to TCP packets.
>    Obviously, the field value cannot be determined for non-TCP packets.=

>=20
>    If Flow Records are generated (i.e., if Cache Mode is "timeout" or
>    "permanent"), every Flow Key field specified by the Cache Layout MUS=
T
>    be included as Flow Key in the resulting Flow Record unless the
>    corresponding Information Element is not applicable or cannot be
>    derived from the content or treatment of the incoming packet.  Two
>    packets MUST be accounted by different Flow Records if they differ i=
n
>    at least one Flow Key field.  Hence, if a Flow Key field can be
>    determined for one packet but not for the other, the two packets are=

>    accounted in different Flow Records.
>=20
>    Every non-key field specified by the Cache Layout MUST be included i=
n
>    the resulting Flow Record unless the corresponding Information
>    Element is not applicable or cannot be derived for the given Flow.
>    Two packets are accounted by the same Flow Record if they only diffe=
r
>    in non-key fields but not in any of the Flow Key fields.  The same
>    applies if one or more non-key fields can be determined for one
>    packet but not for the other.
>=20
>    For example, if a non-key field specifies an Information Element
>    whose value is determined by the first packet observed within a Flow=

>    (which is the default rule according to [RFC5102]), this field MUST
>    be included in the resulting Flow Record if it can be determined fro=
m
>    the first packet of the Flow.

First question:

What happens if the Cache Layout contains an Information Element the=20
Monitoring Device cannot handle?

In my opinion, the only possible answer is that the Monitoring Device=20
returns an error. In general, the Monitoring Device MUST notify the user =

about any part of the configuration which is not supported. It MUST NOT=20
silently accept a configuration which cannot be completely implemented.

I propose to add these two sentences to the introduction. This will=20
implicitly answer the question.


Second question:

What happens to a field in the Cache Layout which cannot be determined=20
from an incoming packet?

Possible answers:

1) It MUST NOT be present in the resulting Packet Report or Flow Record.

2) It MAY still be present in the resulting Packet Report or Flow
    Record, but it MUST be obvious from the field value that the field
    was not be determined from the packet.

3) It MUST still be present in the resulting Packet Report or Flow
    Record and contain a default value which is configured by the user.

4) We use a bit vector field which marks the fields with valid data.


Currently, we do not specify this case, having in mind that 2) is=20
allowed. Example: transport port 0 is invalid and may be used for ICMP.

1) is a clean solution, but it may results in 2^n different Templates=20
per Cache, n being the number of Flow Key fields. Implementing this is=20
more complicated.

3) does not work because the Collector cannot distinguish between=20
default values and real values.

4) is disliked by many people, e.g.:
http://www.ietf.org/mail-archive/web/ipfix/current/msg05275.html


So, I see 1) and 2) as the only feasible solutions.
Depending on the decision, I will add some text to the document.

Regards,
Gerhard


--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms080408050708000604010006
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms080408050708000604010006--

From trammell@tik.ee.ethz.ch  Wed Mar  3 01:39:49 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id F172028C2FC for <ipfix@core3.amsl.com>; Wed,  3 Mar 2010 01:39:49 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.599
X-Spam-Level: 
X-Spam-Status: No, score=-6.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id lEc3bx3RsMl4 for <ipfix@core3.amsl.com>; Wed,  3 Mar 2010 01:39:49 -0800 (PST)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id C45373A8A4F for <ipfix@ietf.org>; Wed,  3 Mar 2010 01:39:48 -0800 (PST)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id 785ADD9331; Wed,  3 Mar 2010 10:39:48 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id R8zhySqmdyuO; Wed,  3 Mar 2010 10:39:48 +0100 (MET)
Received: from pb-10072.ethz.ch (pb-10072.ethz.ch [82.130.103.195]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id 202ABD936C; Wed,  3 Mar 2010 10:39:48 +0100 (MET)
Mime-Version: 1.0 (Apple Message framework v1077)
Content-Type: text/plain; charset=us-ascii
From: Brian Trammell <trammell@tik.ee.ethz.ch>
In-Reply-To: <4B8E2DAD.2010308@net.in.tum.de>
Date: Wed, 3 Mar 2010 10:39:47 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <A7039173-D573-42F5-90C0-75A43A9F204B@tik.ee.ethz.ch>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu>	<4AD793E9-3463-43F1-8840-13BB6FF0B1E6@net.in.tum.de> <33C32EC7-A6D6-4238-B376-3F91FA3B74D4@net.in.tum.de> <4B8BF243.7090303@net.in.tum.de> <D71955E5-D24F-4859-B4F0-1C080B8CF192@net.in.tum.de> <4B8E2DAD.2010308@net.in.tum.de>
To: Gerhard Muenz <muenz@net.in.tum.de>
X-Mailer: Apple Mail (2.1077)
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] WG last call on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 03 Mar 2010 09:39:50 -0000

hi Gerhard, all,

Short, short answers inline below. A much longer message follows.

Cheers,

Brian

On Mar 3, 2010, at 10:36 AM, Gerhard Muenz wrote:

>=20
> Hi Lothar,
>=20
> Ok, we have an open issue regarding the Cache Layout, which is =
important.
>=20
> So, other IPFIX people are encouraged to also read this e-mail and =
comment :-)
>=20
> It's about section 4.3.1, which looks like this after some revisions:
>=20
>> 4.3.1.  CacheLayout Class
>>            +--------------+
>>            | CacheLayout  |
>>            +--------------+   1..* +--------------------------+
>>            |              |<>------| CacheField               |
>>            |              |        +--------------------------+
>>            |              |        | name                     |
>>            |              |        | ieId/ieName              |
>>            |              |        | ieLength {opt.}          |
>>            |              |        | ieEnterpriseNumber[0..1] |
>>            |              |        | isFlowKey[0..1]          |
>>            +--------------+        +--------------------------+
>>                       Figure 12: CacheLayout class
>>   A Cache generates and maintains Packet Reports or Flow Records
>>   containing information that has been extracted from the incoming
>>   stream of packets.  Using the CacheField class, the CacheLayout =
class
>>   specifies the superset of fields that are included in the Packet
>>   Reports or Flow Records (see Figure 12).
>>   If Packet Reports are generated (i.e., if Cache Mode is =
"immediate"),
>>   every field specified by the Cache Layout MUST be included in the
>>   resulting Packet Report unless the corresponding Information =
Element
>>   is not applicable or cannot be derived from the content or =
treatment
>>   of the incoming packet.
>>   For example, if a field is configured to contain the TCP source =
port
>>   (Information Element tcpSourcePort [RFC5102]), the field MUST be
>>   included in all Packet Reports which are related to TCP packets.
>>   Obviously, the field value cannot be determined for non-TCP =
packets.
>>   If Flow Records are generated (i.e., if Cache Mode is "timeout" or
>>   "permanent"), every Flow Key field specified by the Cache Layout =
MUST
>>   be included as Flow Key in the resulting Flow Record unless the
>>   corresponding Information Element is not applicable or cannot be
>>   derived from the content or treatment of the incoming packet.  Two
>>   packets MUST be accounted by different Flow Records if they differ =
in
>>   at least one Flow Key field.  Hence, if a Flow Key field can be
>>   determined for one packet but not for the other, the two packets =
are
>>   accounted in different Flow Records.
>>   Every non-key field specified by the Cache Layout MUST be included =
in
>>   the resulting Flow Record unless the corresponding Information
>>   Element is not applicable or cannot be derived for the given Flow.
>>   Two packets are accounted by the same Flow Record if they only =
differ
>>   in non-key fields but not in any of the Flow Key fields.  The same
>>   applies if one or more non-key fields can be determined for one
>>   packet but not for the other.
>>   For example, if a non-key field specifies an Information Element
>>   whose value is determined by the first packet observed within a =
Flow
>>   (which is the default rule according to [RFC5102]), this field MUST
>>   be included in the resulting Flow Record if it can be determined =
from
>>   the first packet of the Flow.
>=20
> First question:
>=20
> What happens if the Cache Layout contains an Information Element the =
Monitoring Device cannot handle?
>=20
> In my opinion, the only possible answer is that the Monitoring Device =
returns an error. In general, the Monitoring Device MUST notify the user =
about any part of the configuration which is not supported. It MUST NOT =
silently accept a configuration which cannot be completely implemented.
>=20
> I propose to add these two sentences to the introduction. This will =
implicitly answer the question.

Yes.

> Second question:
>=20
> What happens to a field in the Cache Layout which cannot be determined =
from an incoming packet?
>=20
> Possible answers:
>=20
> 1) It MUST NOT be present in the resulting Packet Report or Flow =
Record.
>=20
> 2) It MAY still be present in the resulting Packet Report or Flow
>   Record, but it MUST be obvious from the field value that the field
>   was not be determined from the packet.
>=20
> 3) It MUST still be present in the resulting Packet Report or Flow
>   Record and contain a default value which is configured by the user.
>=20
> 4) We use a bit vector field which marks the fields with valid data.
>=20
>=20
> Currently, we do not specify this case, having in mind that 2) is =
allowed. Example: transport port 0 is invalid and may be used for ICMP.
>=20
> 1) is a clean solution, but it may results in 2^n different Templates =
per Cache, n being the number of Flow Key fields. Implementing this is =
more complicated.
>=20
> 3) does not work because the Collector cannot distinguish between =
default values and real values.
>=20
> 4) is disliked by many people, e.g.:
> http://www.ietf.org/mail-archive/web/ipfix/current/msg05275.html
>=20
>=20
> So, I see 1) and 2) as the only feasible solutions.
> Depending on the decision, I will add some text to the document.

(2).


From trammell@tik.ee.ethz.ch  Wed Mar  3 01:44:13 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id C47913A8A5C for <ipfix@core3.amsl.com>; Wed,  3 Mar 2010 01:44:13 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.599
X-Spam-Level: 
X-Spam-Status: No, score=-6.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id JuoA8Y2ewrr1 for <ipfix@core3.amsl.com>; Wed,  3 Mar 2010 01:44:11 -0800 (PST)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id 133313A8ADF for <ipfix@ietf.org>; Wed,  3 Mar 2010 01:44:10 -0800 (PST)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id 0C5E9D9328; Wed,  3 Mar 2010 10:44:11 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 9Lyb1-PS-i-R; Wed,  3 Mar 2010 10:44:10 +0100 (MET)
Received: from pb-10072.ethz.ch (pb-10072.ethz.ch [82.130.103.195]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id BA6A2D931D; Wed,  3 Mar 2010 10:44:10 +0100 (MET)
Mime-Version: 1.0 (Apple Message framework v1077)
Content-Type: text/plain; charset=us-ascii
From: Brian Trammell <trammell@tik.ee.ethz.ch>
In-Reply-To: <4B8C0943.6070309@net.in.tum.de>
Date: Wed, 3 Mar 2010 10:44:10 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <A89AA32C-086C-46B9-8B6F-66BBEA645602@tik.ee.ethz.ch>
References: <3146E35B-9877-44AE-B884-17541E6ED5B8@tik.ee.ethz.ch> <4B8C0943.6070309@net.in.tum.de>
To: Gerhard Muenz <muenz@net.in.tum.de>
X-Mailer: Apple Mail (2.1077)
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] WGLC comments on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 03 Mar 2010 09:44:14 -0000

Hi, Gerhard, all,

On Mar 1, 2010, at 7:36 PM, Gerhard Muenz wrote:

>=20
> Hi Brian,
>=20
> Thanks a lot for your review.
>=20
> Data models in network management will never satisfy all people. So, =
please keep in mind that the goal of this document is to standardize a =
configuration data model that can be easily implemented by many vendors. =
There will always be additional vendor-specific parameters, and there =
will always be vendor-specific restrictions of the model. Thanks to =
YANG, there are possibilities to make restrictions and extension.
>=20
> So, ask yourself if the parameters which are contained in the =
configuration data model can be easily mapped to one of the YAF command =
line options. Not the other way round.

Yep, I understand the point of the data model here... and YAF's =
configuration by its nature will most probably remain command-line =
(indeed I was thinking of implementing this in a Ruby script that kicked =
off YAF); the references to YAF herein were merely an attempt to use one =
specific implementation (with which I am familiar) as a tool for =
evaluating the flexibility and general applicability of the =
configuration data model. "This is out of scope" is a perfectly =
reasonable response. bue I wanted to make sure in reviewing that the =
flexibility we can get out of yang is presented in the right way... =
Indeed, in the selector->cache->ep chain, you can probably just model =
intermediate processes (like YAF's partial reassembly) as being part of =
the Cache, and that's just fine.

>=20
> More comments inline.
>=20
> Gerhard
>=20
>=20
> Brian Trammell wrote:
>> Greetings, all,
>> I have reviewed draft-ietf-ipfix-configuration-model-04 in response
>> to the WGLC, and have the following comments, ranked in order of
>> impact; most of these deal with the applicability of the
>> configuration data model to implementations other than the "exemplar"
>> implementation(s) the authors had in mind while while writing.  I
>> will say that overall, the draft does an more than reasonable job of
>> being generally applicable. I apologize for starting a new thread,
>> but I have not yet had time to properly review other comments, but
>> plan on being in Anaheim prepared to discuss those too. I have also
>> explicitly not reviewed the YANG definition; I leave that to people
>> with actual YANG expertise.
>> First and foremost is the issue of caches, as it is in cache
>> implementation that IPFIX packet-to-flow MPs will differentiate the
>> most from each other, and indeed most vendors should be expected to
>> subclass this cache class for their own configuration parameters;
>> therefore getting this right is probably the most crucial. cacheMode
>> immediate seems like a bit of a hack, as it invalidates basically
>> every other cache field. However, I suppose I can see it as a way to
>> keep the class design clean (otherwise, there would be nothing to
>> connect OP to layout/EP in a PSAMP implementation...).
>=20
> Right.
>=20
>> 1. I'm concerned about the definition of cacheMode timeout (4.3, p.
>> 24), as it seems to make it impossible to configure a cache to allow
>> cache exit on TCP state tracking. Perhaps this is merely an editorial
>> oversight?
>=20
> We can allow a timeout Cache to expire a flow after a certain flow =
termination event. Is this what you propose?

Basically, yes. Most flow generators will expire a flow after a TCP FIN =
or RST (or some number of these), so there is a difference between =
expiring flows only on timeout and expiring on timeout as well as =
"natural" end of flow (see the definition of flowEndReason, IE 136). So =
I would propose the following:

  cacheMode:  Configures the Cache Mode.  The following parameter
     values are specified by the configuration data model:
     *  immediate: Data Records expire after the first packet
     *  timeout: Data Records expire after active or inactive timeout
|    *  natural: Data Records expire after active or inactive timeout, =
or
|       on natural termination (e.g. TCP FIN, or TCP RST) as determined
|       by the Metering Process.
     *  permanent: Data Records never expire, but are periodically
        exported with interval set by exportInterval
     In the case of "immediate", PSAMP Packet Reports are generated.
     Otherwise, IPFIX Flow Records are generated.

Alternately, if "timeout" meant "timeout + natural" in this terminology, =
we could change it as follows:

  cacheMode:  Configures the Cache Mode.  The following parameter
     values are specified by the configuration data model:
     *  immediate: Data Records expire after the first packet
|    *  timeout: Data Records expire after active or inactive timeout, =
or
|       on natural termination (e.g. TCP FIN, or TCP RST) as determined
|       by the Metering Process.
     *  permanent: Data Records never expire, but are periodically
        exported with interval set by exportInterval
     In the case of "immediate", PSAMP Packet Reports are generated.
     Otherwise, IPFIX Flow Records are generated.

I'm inclined toward the first alternative, with 'natural' as default and =
'timeout' provided for MPs that don't want to be bothered with TCP =
state.


>=20
>> 2. Destination (4.4.1, pp .27-29) appears not to match the properties
>> of the particular transports:
> >
>> a. can't specify a local IP address for TCP or UDP. This seems
>> necessary on devices with multiple management interfaces, which I
>> don't think is all that weird a deployment plan.
>=20
> This will be fixed as proposed in this mail unless anybody objects:
> http://www.ietf.org/mail-archive/web/ipfix/current/msg05316.html

Perfect.

>> b. can't specify 1..n remote IP addresses for SCTP. This breaks one
>> of the advantages of SCTP.
>=20
> I see send_connectx() in the SCTP API:
> http://tools.ietf.org/html/draft-ietf-tsvwg-sctpsocket-21#section-8.9
>=20
> However, what happens if you configure multiple remote IP addresses =
and they do not belong to the same SCTP endpoint?
> What shall the Exporting Process do in this situation?

What sctp_connectx() does:=20

"The way the SCTP stack uses the list of addresses to set up the =
association is implementation dependent.  This function only specifies =
that the stack will try to make use of all the addresses in the list =
when needed." The assumptions are 1. that the list of addresses all =
correspond to a logically unitary CP and 2. that the admin knows what =
he's doing.=20

> I'm not sure if it makes sense to configure multiple remote addresses =
for a single destination. It seems to be safer to configure multiple =
export destinations and to use export member type primary and secondary.
>=20
> What do you think?

These are logically separate concepts, and should not be conflated by =
the data model. Multiple Destinations means multiple CPs: endpoint =
redundancy, and the assumption is that each Destination is its own CP =
with separate app-layer state. Multiple addresses means multiple =
endpoints on a single CP for session setup: _path_ redundancy, and the =
assumption is that every given address leads to the same CP, with the =
same app-layer state. Different tools for different situations. Of =
course you can configure to export to two addresses that aren't the same =
CP, just like you can configure an EP to export to a destination that =
_isn't_ a CP. In both cases this is a configuration error.

>> c. can't configure SCTP reliability per-template, as opposed to per
>> destination. see 2.b.
>=20
> As discussed long ago, Templates are not configured. The are created =
by the Exporting Process depending on the Data Records to be exported.

I'm not convinced this choice is the right one (both the Cache and the =
EP are valid places to make choices about what, precisely, will be =
exported: see below the discussion on=20

> Therefore, there is no good place for a per-Template reliability =
parameter in the configuration data model. The only possibility would be =
to configure the reliability in the Cache, and this would be a very =
strange place to do it.
>=20
> I had discussed this issue with Benoit. The conclusion was that we =
have a per-Destination parameter. If people want to configure more, they =
can come up with an extension.
>=20
> If you have another convincing solution, please let me know.

Not really. But now we as a WG are saying logically inconsistent things: =
I can't reconcile the "templates are applications and the reliability =
settings are per-application; here is a way to estimate loss across =
multiple templates to the same destination" view of sctp per stream =
export and the "destination-level reliability" view presented here. It =
is an inelegant contradiction.

My personal opinion would be to provide template-level configuration as =
an EP option in an extension, and to simply ignore this Destination =
option (much the same way as I work as hard as I can to ignore UDP ;) )

>> 3. Options (4.4.3, p. 30) configuration is rather odd and highly
>> restrictive, even to the point of being unusable. The biggest problem
>> is that RFC5473 common properties are something completely different
>> from the options specified in 5101 and 5476. "Enable common
>> properties", without some definition of which fields to use as common
>> properties and under what circumstances, is not a useful directive.
>=20
> The idea is that the Exporting Process defines common properties =
depending on the Data Records to be exported. This can be done =
dynamically (depending on the value distribution) or using knowledge =
about the configuration (e.g., if a filter ensures that the accounted =
packets are from/to a specific endpoint, this can be encoded as common =
property).
>=20
> If you think that reducingRedundancy is not useful, I will remove from =
the list.

It's certainly useful to have a switch, but we must then be very =
explicit about the fact that when you turn this switch on, it's up to =
the EP to determine precisely what redundancy to reduce and how, that =
this is implementation dependent, and that implementations are =
encouraged to provide implementation-specific configuration extensions =
to cover this case.

>> Secondarily, options specified in 5610 and 5655 are not covered; do
>> we really want to amend this document every time we specify another
>> WG-blessed options template?=20
>=20
> Apart from reducingRedundancy, all mentioned Options type appear in =
the protocol RFCs (RFC5101, RFC5476).

5655 options don't make much sense to configure, as runtime =
determination and implementation dependency probably dominate there. =
However, type export as in 5610 seems like it would be a useful =
configuration switch.

Although this discussion reminds me of another issue I missed, =
elaborated below.

> Adding more parameter values for optionsType can be easily achieved =
with YANG extensions because they are specified as identities (it would =
note be possible with an enum type).

Hm. Okay. Will we create then a "defined Options Template" registry =
somewhere?


> > I'd suggest using an OptionsTemplate
>> subclass of Template, instead, and providing the definition of common
>> (non-5473) options templates in an appendix to the configuration data
>> model.
>=20
> I do not understand your proposal. There is no Template class for =
configuration.

I missed this point somehow (perhaps in my assumption that Templates =
indeed should be configurable, but I did not show up to that discussion =
so I accept the decision of the rest of the WG)... but yes, indeed, so, =
withdrawn.

>> 4. There does not seem to be any room for logical configurable
>> processes between selection and cache, or cache and EP. The example
>> that comes to mind is perhaps implementation-specific [I'm mentioning
>> this because yaf does it]: fragmentation behavior (reassemble,
>> ignore, drop). How could I allow then for the configuration of the
>> fragment table (given that a selection process can only reference a
>> cache)?
>=20
> You can always extend the model. For example, if fragmentation =
reassembly can be regarded as part of the Cache, you can easily define =
your own new cacheType (again, identities and identityref in YANG).
>=20
> We also have implemented a lot of cool additional functionalities =
which we are going to configure by extending the model. It is not the =
purpose of this document to cover all of these.

Indeed. As long as you can collapse these into Cache (which as I said =
above it seems you can), that's more than sufficient.

>=20
>> 5. CacheLayout (4.3.1, p. 25) does not appear to be able to support
>> adaptive reduced-length export (exporting, say, counters with
>> unsigned32 unless one of them actually requires unsigned64) [yes, yaf
>> does this too]. Though this could probably be solved with a
>> deviation, and an additional vendor-specific CacheField attribute...
>=20
> Is reduced size encoding performed by the Cache or by the Exporting =
Process?

In YAF, there's no process separation, but from a functional block =
separation this is done in the EP.=20

> If you implement a Cache in hardware, it is probably easier to always =
use the maximum length and to reduce it later in the Exporting Process.
> We assume that the Cache reserves a certain field length for all Data =
Records in the Cache.  (Variable-length fields are treated differently)

Indeed, I would expect that only caches written in dynamically typed =
languages, where the language itself autopromotes storage on overflow, =
would do variable-sized caching of fixed length elements, if only =
because doing it in C is rather time expensive and difficult for =
relatively little benefit. And in this case the variable-sized caching =
is invisible to the programmer.

> The ieLength specifies the field length in the Cache. The length in =
the exported Data Record can be different. Note again: we are not =
configuring a Template, here. The exported Data Record may even contain =
a completely different set of fields, for example in the case of =
reducing redundancy.

Ah, okay, so it supports it, it just doesn't support configuring it =
(which can be an implementation dependent extension in either the cache, =
the EP, or both, which makes sense depending on how it's implemented). =
This is okay.

> What about removing the following sentence to avoid any confusion:
>=20
> 	For Information Elements of
>      integer and float type, the field length MAY be set to a smaller
>      value than the standard length of the abstract data type if the
>      rules of reduced size encoding are fulfilled (see [RFC5101],
>      Section 6.2).

No reason to cut that, as it wasn't the point of confusion...

>=20
>> 6. CacheLayout (4.3.1, p. 25) could do with an explicit example to
>> illustrate the "unless the corresponding IE is not applicable or
>> cannot be derived" case: I presume this is intended for dual-stack
>> IPv4/IPv6 measurement, RFC5103 biflows, and the like? It does appear
>> than this clause is sufficient for such cases, but I had to read this
>> a couple of times before I understood that's what was meant.
>=20
> Right. I tried to improve this. Please look into my reply to Lothar's =
review:
> http://www.ietf.org/mail-archive/web/ipfix/current/msg05325.html

Perfect.

>> 7. Real File Readers and File Writers will probably need more than a
>> single filename endpoint on the file system (however, most file
>> readers and file writers in document-based workflows won't be
>> configured using this information model...) How to handle this is
>> probably implementation-specific; in any case, I don't have a good
>> suggestion here, yet; perhaps we can discuss in Anaheim.
>=20
> I assume that the URI filename parameter is useful for many =
FileWriter/Reader implementations.  All additional parameters are =
probably very specific to the implementation.

Indeed. Okay, makes sense to leave it then, maybe with a statement that =
the handling of the URI is implementation dependent.

>=20
>> 8. Is there any benefit to cascading caches as with selection
>> processes, or rather, are the benefits of cascading caches considered
>> in scope for this application? The simplest example would be
>> simultaneous export of flow data as well as flow aggregate data (e.g.
>> traffic counters per AS pair), perhaps as part of a
>> cascading-aggregation mediator. I ask only because it seems a rather
>> small and easily supportable change to greatly increase flexibility.
>=20
> Uff. I already have problems to convince people that there may be more =
than one parallel Caches in a Metering Process. Now, you even propose =
multiple Caches in a series :)
>=20
> IMO, aggregation is done in an Intermediate Process, not in a Metering =
Process and therefore not in the Cache of a Metering Process.
>=20
> The configuration of Mediators and Intermediate Process is out of =
scope of this document. Otherwise, we will never finish...

Accepted. But now I have an idea of at least one subsection heading in =
the Mediator Configuration Data Model draft :)

>> 9. Destination exportMemberType appears to reference a renamed MIB
>> object as of MIB-10; this is called ipfixExportMemberType now. I
>> would suggest copying the complete description from MIB, because the
>> semantics are kind of confusing otherwise (I had to read MIB, which I
>> probably should have done by now but oh well, in order to figure out
>> what the difference between parallel and primary was). A tertiary
>> problem: these exportMemberTypes seem a bit restrictive (e.g. what if
>> I want to do parallel export and also have secondaries?), but I
>> suppose that's a MIB comment.
>=20
> Andrew suggested to improve this as well.  See:
> http://www.ietf.org/mail-archive/web/ipfix/current/msg05323.html
> I'm open to concrete suggestions.

Update the text in the configuration model to match that in the MIB, =
copy the definitions in full from the MIB so you don't have to go read =
the MIB to figure out what these mean, and call it finished. I expect =
most people will either find the MIB approach sufficient, or simply =
leave these parallel and implement their own redundancy.

>> 10. draft-ietf-ipfix-file is now RFC5655 (the editor will catch
>> this); since this is now an RFC, references to things "proposed" in
>> this draft should probably be changed to things "specified" in RFC
>> 5655.
>=20
> Done.

And, as promised, a new 11. Cache should have a way to configure biflow =
assembly and export. The simplest one is simply to state that the =
presence of reverse Information Elements in a CacheLayout implies that =
the Cache will assemble biflows suitable for export as described in RFC =
5103.

This opens a slightly larger question, though: is this a configuration =
data model for 5101 and 5102, for all the RFCs published by the WG to =
date (i.e., adding 5103, 5473, 5610, and maybe 5655), for all the work =
completed by the WG (adding per-stream), or for all the work accepted =
aside from mediation (adding flow selection, and possibly anonymisation =
in a non-mediator context)? I think the last is almost certainly out of =
scope (otherwise we're never done). Thoughts?

Cheers,

Brian=

From muenz@net.in.tum.de  Thu Mar  4 06:42:52 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 88DB328C154 for <ipfix@core3.amsl.com>; Thu,  4 Mar 2010 06:42:52 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id hXtQb7dKaI6f for <ipfix@core3.amsl.com>; Thu,  4 Mar 2010 06:42:51 -0800 (PST)
Received: from mail-out2.informatik.tu-muenchen.de (mail-out2.informatik.tu-muenchen.de [131.159.0.36]) by core3.amsl.com (Postfix) with ESMTP id C5CA628C142 for <ipfix@ietf.org>; Thu,  4 Mar 2010 06:42:50 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 82CEB200A791; Thu,  4 Mar 2010 15:42:48 +0100 (CET)
Message-ID: <4B8FC6F3.9040004@net.in.tum.de>
Date: Thu, 04 Mar 2010 15:42:59 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Brian Trammell <trammell@tik.ee.ethz.ch>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu>	<4AD793E9-3463-43F1-8840-13BB6FF0B1E6@net.in.tum.de> <33C32EC7-A6D6-4238-B376-3F91FA3B74D4@net.in.tum.de> <4B8BF243.7090303@net.in.tum.de> <D71955E5-D24F-4859-B4F0-1C080B8CF192@net.in.tum.de> <4B8E2DAD.2010308@net.in.tum.de> <A7039173-D573-42F5-90C0-75A43A9F204B@tik.ee.ethz.ch>
In-Reply-To: <A7039173-D573-42F5-90C0-75A43A9F204B@tik.ee.ethz.ch>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms070803040109010306040407"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] WG last call on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 04 Mar 2010 14:42:52 -0000

This is a cryptographically signed message in MIME format.

--------------ms070803040109010306040407
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Brian, all,

Brian Trammell wrote:
> hi Gerhard, all,
>=20
> Short, short answers inline below. A much longer message follows.
>=20
> Cheers,
>=20
> Brian
>=20
> On Mar 3, 2010, at 10:36 AM, Gerhard Muenz wrote:
>=20
>> Hi Lothar,
>>
>> Ok, we have an open issue regarding the Cache Layout, which is importa=
nt.
>>
>> So, other IPFIX people are encouraged to also read this e-mail and com=
ment :-)
>>
>> It's about section 4.3.1, which looks like this after some revisions:
>>
>>> 4.3.1.  CacheLayout Class
>>>            +--------------+
>>>            | CacheLayout  |
>>>            +--------------+   1..* +--------------------------+
>>>            |              |<>------| CacheField               |
>>>            |              |        +--------------------------+
>>>            |              |        | name                     |
>>>            |              |        | ieId/ieName              |
>>>            |              |        | ieLength {opt.}          |
>>>            |              |        | ieEnterpriseNumber[0..1] |
>>>            |              |        | isFlowKey[0..1]          |
>>>            +--------------+        +--------------------------+
>>>                       Figure 12: CacheLayout class
>>>   A Cache generates and maintains Packet Reports or Flow Records
>>>   containing information that has been extracted from the incoming
>>>   stream of packets.  Using the CacheField class, the CacheLayout cla=
ss
>>>   specifies the superset of fields that are included in the Packet
>>>   Reports or Flow Records (see Figure 12).
>>>   If Packet Reports are generated (i.e., if Cache Mode is "immediate"=
),
>>>   every field specified by the Cache Layout MUST be included in the
>>>   resulting Packet Report unless the corresponding Information Elemen=
t
>>>   is not applicable or cannot be derived from the content or treatmen=
t
>>>   of the incoming packet.
>>>   For example, if a field is configured to contain the TCP source por=
t
>>>   (Information Element tcpSourcePort [RFC5102]), the field MUST be
>>>   included in all Packet Reports which are related to TCP packets.
>>>   Obviously, the field value cannot be determined for non-TCP packets=
=2E
>>>   If Flow Records are generated (i.e., if Cache Mode is "timeout" or
>>>   "permanent"), every Flow Key field specified by the Cache Layout MU=
ST
>>>   be included as Flow Key in the resulting Flow Record unless the
>>>   corresponding Information Element is not applicable or cannot be
>>>   derived from the content or treatment of the incoming packet.  Two
>>>   packets MUST be accounted by different Flow Records if they differ =
in
>>>   at least one Flow Key field.  Hence, if a Flow Key field can be
>>>   determined for one packet but not for the other, the two packets ar=
e
>>>   accounted in different Flow Records.
>>>   Every non-key field specified by the Cache Layout MUST be included =
in
>>>   the resulting Flow Record unless the corresponding Information
>>>   Element is not applicable or cannot be derived for the given Flow.
>>>   Two packets are accounted by the same Flow Record if they only diff=
er
>>>   in non-key fields but not in any of the Flow Key fields.  The same
>>>   applies if one or more non-key fields can be determined for one
>>>   packet but not for the other.
>>>   For example, if a non-key field specifies an Information Element
>>>   whose value is determined by the first packet observed within a Flo=
w
>>>   (which is the default rule according to [RFC5102]), this field MUST=

>>>   be included in the resulting Flow Record if it can be determined fr=
om
>>>   the first packet of the Flow.
>> First question:
>>
>> What happens if the Cache Layout contains an Information Element the M=
onitoring Device cannot handle?
>>
>> In my opinion, the only possible answer is that the Monitoring Device =
returns an error. In general, the Monitoring Device MUST notify the user =
about any part of the configuration which is not supported. It MUST NOT s=
ilently accept a configuration which cannot be completely implemented.
>>
>> I propose to add these two sentences to the introduction. This will im=
plicitly answer the question.
>=20
> Yes.

This text will appear in Section 5 which is about "Adaptation to Device=20
Capabilities" and therefore should be the right place in the document:

    The deviation concept enables a device to announce deviations from
    the standard model using the "deviation" statement.  For example, it
    is possible to restrict the value range of a specific parameter or to=

    define that the configuration of a certain parameter is not supported=

    at all.  Hence, deviations are typically used to specify limitations
    due to resource constraints or functional restrictions.  Deviations
    concern existing parameters of the original configuration data model
    and must not be confused with model extensions.  Model extension are
    specified with the "augment" statement and allow adding new
    parameters to the original configuration data model.

    If certain device-specific constraints cannot be formally specified
    with YANG, they MUST be expressed with human-readable text using the
    "description" statement.  The provided information MUST enable the
    user to define a configuration which is entirely supported by the
    Monitoring Device.  On the other hand, if a Monitoring Device is
    configured, it MUST notify the user about any part of the
    configuration which is not supported.  The Monitoring Device MUST NOT=

    silently accept configuration data which cannot be completely
    enforced.

    Just like features, deviations and model extensions are announced in
    NETCONF's <hello> message.  A usage example of deviations is given in=

    Section 7.5.


>> Second question:
>>
>> What happens to a field in the Cache Layout which cannot be determined=
 from an incoming packet?
>>
>> Possible answers:
>>
>> 1) It MUST NOT be present in the resulting Packet Report or Flow Recor=
d.
>>
>> 2) It MAY still be present in the resulting Packet Report or Flow
>>   Record, but it MUST be obvious from the field value that the field
>>   was not be determined from the packet.
>>
>> 3) It MUST still be present in the resulting Packet Report or Flow
>>   Record and contain a default value which is configured by the user.
>>
>> 4) We use a bit vector field which marks the fields with valid data.
>>
>>
>> Currently, we do not specify this case, having in mind that 2) is allo=
wed. Example: transport port 0 is invalid and may be used for ICMP.
>>
>> 1) is a clean solution, but it may results in 2^n different Templates =
per Cache, n being the number of Flow Key fields. Implementing this is mo=
re complicated.
>>
>> 3) does not work because the Collector cannot distinguish between defa=
ult values and real values.
>>
>> 4) is disliked by many people, e.g.:
>> http://www.ietf.org/mail-archive/web/ipfix/current/msg05275.html
>>
>>
>> So, I see 1) and 2) as the only feasible solutions.
>> Depending on the decision, I will add some text to the document.
>=20
> (2).

Fine.

I'll wait for more feedback before writing more text.

Regards,
Gerhard



--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms070803040109010306040407
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms070803040109010306040407--

From braun@net.in.tum.de  Thu Mar  4 07:43:53 2010
Return-Path: <braun@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 9386F3A8D69 for <ipfix@core3.amsl.com>; Thu,  4 Mar 2010 07:43:53 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id kQrbe4tUb5td for <ipfix@core3.amsl.com>; Thu,  4 Mar 2010 07:43:52 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 34EB228C1ED for <ipfix@ietf.org>; Thu,  4 Mar 2010 07:43:35 -0800 (PST)
Received: from honshu.net.in.tum.de (honshu.net.in.tum.de [131.159.20.100]) by mail.net.in.tum.de (Postfix) with ESMTPSA id AF7F8200AD9A; Thu,  4 Mar 2010 16:43:35 +0100 (CET)
Mime-Version: 1.0 (Apple Message framework v1077)
Content-Type: text/plain; charset=iso-8859-1
From: Lothar Braun <braun@net.in.tum.de>
In-Reply-To: <4B8FC6F3.9040004@net.in.tum.de>
Date: Thu, 4 Mar 2010 16:43:35 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <1CEC4FEC-8C83-4A43-B7BA-F3C7F50A9CBC@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu>	<4AD793E9-3463-43F1-8840-13BB6FF0B1E6@net.in.tum.de> <33C32EC7-A6D6-4238-B376-3F91FA3B74D4@net.in.tum.de> <4B8BF243.7090303@net.in.tum.de> <D71955E5-D24F-4859-B4F0-1C080B8CF192@net.in.tum.de> <4B8E2DAD.2010308@net.in.tum.de> <A7039173-D573-42F5-90C0-75A43A9F204B@tik.ee.ethz.ch> <4B8FC6F3.9040004@net.in.tum.de>
To: Gerhard Muenz <muenz@net.in.tum.de>
X-Mailer: Apple Mail (2.1077)
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] WG last call on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 04 Mar 2010 15:43:53 -0000

Hi Gerhard,

On Mar 4, 2010, at 3:42 PM, Gerhard Muenz wrote:
> Brian Trammell wrote:
>> hi Gerhard, all,
>> Short, short answers inline below. A much longer message follows.
>> Cheers,
>> Brian
>> On Mar 3, 2010, at 10:36 AM, Gerhard Muenz wrote:
>>> Hi Lothar,
>>>=20
>>> Ok, we have an open issue regarding the Cache Layout, which is =
important.
>>>=20
>>> So, other IPFIX people are encouraged to also read this e-mail and =
comment :-)
>>>=20
>>> It's about section 4.3.1, which looks like this after some =
revisions:
>>>=20
>>>> 4.3.1.  CacheLayout Class
>>>>           +--------------+
>>>>           | CacheLayout  |
>>>>           +--------------+   1..* +--------------------------+
>>>>           |              |<>------| CacheField               |
>>>>           |              |        +--------------------------+
>>>>           |              |        | name                     |
>>>>           |              |        | ieId/ieName              |
>>>>           |              |        | ieLength {opt.}          |
>>>>           |              |        | ieEnterpriseNumber[0..1] |
>>>>           |              |        | isFlowKey[0..1]          |
>>>>           +--------------+        +--------------------------+
>>>>                      Figure 12: CacheLayout class
>>>>  A Cache generates and maintains Packet Reports or Flow Records
>>>>  containing information that has been extracted from the incoming
>>>>  stream of packets.  Using the CacheField class, the CacheLayout =
class
>>>>  specifies the superset of fields that are included in the Packet
>>>>  Reports or Flow Records (see Figure 12).
>>>>  If Packet Reports are generated (i.e., if Cache Mode is =
"immediate"),
>>>>  every field specified by the Cache Layout MUST be included in the
>>>>  resulting Packet Report unless the corresponding Information =
Element
>>>>  is not applicable or cannot be derived from the content or =
treatment
>>>>  of the incoming packet.
>>>>  For example, if a field is configured to contain the TCP source =
port
>>>>  (Information Element tcpSourcePort [RFC5102]), the field MUST be
>>>>  included in all Packet Reports which are related to TCP packets.
>>>>  Obviously, the field value cannot be determined for non-TCP =
packets.
>>>>  If Flow Records are generated (i.e., if Cache Mode is "timeout" or
>>>>  "permanent"), every Flow Key field specified by the Cache Layout =
MUST
>>>>  be included as Flow Key in the resulting Flow Record unless the
>>>>  corresponding Information Element is not applicable or cannot be
>>>>  derived from the content or treatment of the incoming packet.  Two
>>>>  packets MUST be accounted by different Flow Records if they differ =
in
>>>>  at least one Flow Key field.  Hence, if a Flow Key field can be
>>>>  determined for one packet but not for the other, the two packets =
are
>>>>  accounted in different Flow Records.
>>>>  Every non-key field specified by the Cache Layout MUST be included =
in
>>>>  the resulting Flow Record unless the corresponding Information
>>>>  Element is not applicable or cannot be derived for the given Flow.
>>>>  Two packets are accounted by the same Flow Record if they only =
differ
>>>>  in non-key fields but not in any of the Flow Key fields.  The same
>>>>  applies if one or more non-key fields can be determined for one
>>>>  packet but not for the other.
>>>>  For example, if a non-key field specifies an Information Element
>>>>  whose value is determined by the first packet observed within a =
Flow
>>>>  (which is the default rule according to [RFC5102]), this field =
MUST
>>>>  be included in the resulting Flow Record if it can be determined =
from
>>>>  the first packet of the Flow.
>>> First question:
>>>=20
>>> What happens if the Cache Layout contains an Information Element the =
Monitoring Device cannot handle?
>>>=20
>>> In my opinion, the only possible answer is that the Monitoring =
Device returns an error. In general, the Monitoring Device MUST notify =
the user about any part of the configuration which is not supported. It =
MUST NOT silently accept a configuration which cannot be completely =
implemented.
>>>=20
>>> I propose to add these two sentences to the introduction. This will =
implicitly answer the question.
>> Yes.
>=20
> This text will appear in Section 5 which is about "Adaptation to =
Device Capabilities" and therefore should be the right place in the =
document:
>=20
>   The deviation concept enables a device to announce deviations from
>   the standard model using the "deviation" statement.  For example, it
>   is possible to restrict the value range of a specific parameter or =
to
>   define that the configuration of a certain parameter is not =
supported
>   at all.  Hence, deviations are typically used to specify limitations
>   due to resource constraints or functional restrictions.  Deviations
>   concern existing parameters of the original configuration data model
>   and must not be confused with model extensions.  Model extension are
>   specified with the "augment" statement and allow adding new
>   parameters to the original configuration data model.
>=20
>   If certain device-specific constraints cannot be formally specified
>   with YANG, they MUST be expressed with human-readable text using the
>   "description" statement.  The provided information MUST enable the
>   user to define a configuration which is entirely supported by the
>   Monitoring Device.  On the other hand, if a Monitoring Device is
>   configured, it MUST notify the user about any part of the
>   configuration which is not supported.  The Monitoring Device MUST =
NOT
>   silently accept configuration data which cannot be completely
>   enforced.
>=20
>   Just like features, deviations and model extensions are announced in
>   NETCONF's <hello> message.  A usage example of deviations is given =
in
>   Section 7.5.

It is unclear to me what happens when there is a configuration that =
cannot be completely enforced.

The device MUST inform the user that the configuration is not fully =
supported. But what about the configuration: Does the monitor have to =
apply the configuration anyways or must it reject the configuration with =
an error? Or can the implementer decide on what to do?

It is not clear to me if "MUST NOT silently accept" corresponds to =
"reject" or "protest against the configuration but accept it anyways".=20=


>=20
>>> Second question:
>>>=20
>>> What happens to a field in the Cache Layout which cannot be =
determined from an incoming packet?
>>>=20
>>> Possible answers:
>>>=20
>>> 1) It MUST NOT be present in the resulting Packet Report or Flow =
Record.
>>>=20
>>> 2) It MAY still be present in the resulting Packet Report or Flow
>>>  Record, but it MUST be obvious from the field value that the field
>>>  was not be determined from the packet.
>>>=20
>>> 3) It MUST still be present in the resulting Packet Report or Flow
>>>  Record and contain a default value which is configured by the user.
>>>=20
>>> 4) We use a bit vector field which marks the fields with valid data.
>>>=20
>>>=20
>>> Currently, we do not specify this case, having in mind that 2) is =
allowed. Example: transport port 0 is invalid and may be used for ICMP.
>>>=20
>>> 1) is a clean solution, but it may results in 2^n different =
Templates per Cache, n being the number of Flow Key fields. Implementing =
this is more complicated.
>>>=20
>>> 3) does not work because the Collector cannot distinguish between =
default values and real values.
>>>=20
>>> 4) is disliked by many people, e.g.:
>>> http://www.ietf.org/mail-archive/web/ipfix/current/msg05275.html
>>>=20
>>>=20
>>> So, I see 1) and 2) as the only feasible solutions.
>>> Depending on the decision, I will add some text to the document.
>> (2).
>=20
> Fine.

I would also prefer Option 2 but I have no idea if this is possible for =
all fields.

Best regards,
  Lothar

--
Lothar Braun
Chair for Network Architectures and Services (I8)
Department of Informatics
Technische Universit=E4t M=FCnchen
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18010       Fax: +49 89 289-18033
E-mail: braun@net.in.tum.de=20







From akoba@nttv6.net  Fri Mar  5 01:06:51 2010
Return-Path: <akoba@nttv6.net>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 2618D28C1CC for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 01:06:51 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.611
X-Spam-Level: 
X-Spam-Status: No, score=0.611 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HOST_MISMATCH_COM=0.311, RDNS_DYNAMIC=0.1, SUBJ_RE_NUM=2.799]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Q8CaXf5F2Wov for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 01:06:50 -0800 (PST)
Received: from mail.nttv6.net (mail.nttv6.net [IPv6:2001:fa8::25]) by core3.amsl.com (Postfix) with ESMTP id 9AD8E3A8D52 for <ipfix@ietf.org>; Fri,  5 Mar 2010 01:06:49 -0800 (PST)
Received: from [127.0.0.1] (dhcp-3-152.nttv6.com [192.47.163.152]) by mail.nttv6.net (8.14.3/8.14.3) with ESMTP id o2596oxT037961; Fri, 5 Mar 2010 18:06:50 +0900 (JST) (envelope-from akoba@nttv6.net)
Date: Fri, 05 Mar 2010 18:01:59 +0900
From: Atsushi Kobayashi <akoba@nttv6.net>
To: Gerhard Muenz <muenz@net.in.tum.de>
In-Reply-To: <4B8D1926.8030905@net.in.tum.de>
References: <20100302203953.40CF.17391CF2@nttv6.net> <4B8D1926.8030905@net.in.tum.de>
Message-Id: <20100305174408.415E.17391CF2@nttv6.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
X-Mailer: Becky! ver. 2.50.05 [ja] (Unregistered)
X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.2.3 (mail.nttv6.net [192.16.178.5]); Fri, 05 Mar 2010 18:06:50 +0900 (JST)
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] WGLC comments on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 05 Mar 2010 09:06:51 -0000

Hi Gerhard,

On Tue, 02 Mar 2010 14:56:54 +0100
Gerhard Muenz <muenz@net.in.tum.de> wrote:
>>     sourceIpAddress:  This parameter specifies the source IP address(es)
>>        used by the Exporting Process.
>>        If the transport protocol is SCTP, this parameter MAY appear
>>        multiple times to specify the list of eligible local IP addresses
>>        of a multi-homed Exporting Process [RFC4960].  If this parameter
>>        is omitted and the transport protocol is SCTP, all locally
>>        assigned IP addresses are eligible local IP addresses.
>>        If the transport protocol is UDP or TCP, this parameter MAY appear
>>        only once. If this parameter is omitted and the transport protocol
>>        is UDP or TCP, the IP address assigned to the outgoing interface
>>        is used as source IP address.
>>
>>     ifIndex/ifName:  Either the index or the name of the interface used
>>        by the Exporting Process to export IPFIX messages to the given
>>        destination MAY be specified according to corresponding objects in
>>        the IF-MIB <xref target="RFC2863"/>.  If omitted, the Exporting
>>        Process selects the outgoing interface based on local routing
>>        decision and accepts return traffic, such as transport layer
>>        acknowledgments, on all available interfaces.

I understood. They are fine. Thank you.

Regards,
Atsushi

> This is something which confused my as well, but Paul put me on the 
> right track.
> 
> I try to explain:
> 
> If you do not bind the socket to a specific (link-layer) interface, the 
> OS will look into the routing table and choose the outgoing interface 
> with the lowest routing costs to the destination. The source IP address 
> does not matter at all in this decision. Even in the case of TCP, 
> asymmetric routing may lead to the effect that outgoing packets are sent 
> on one router interface and incoming packets are received on another 
> interface. This is a valid TCP connection.
> 
> As administrator, you want to enforce that IPFIX messages leave the 
> router on the management interface, independently of the routing table. 
> This is done with ifName/ifIndex.
> 
> Is it clear now?


--- 
Atsushi KOBAYASHI  <akoba@nttv6.net>
NTT Information Sharing Platform Lab.
tel:+81-(0)422-59-3978 fax:+81-(0)422-59-5637



From muenz@net.in.tum.de  Fri Mar  5 02:49:48 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id F27803A8522 for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 02:49:48 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id l88GtsfBuk9f for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 02:49:46 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 1ABEC3A8760 for <ipfix@ietf.org>; Fri,  5 Mar 2010 02:49:45 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 2DB9C200A791; Fri,  5 Mar 2010 11:49:46 +0100 (CET)
Message-ID: <4B90E1D5.90506@net.in.tum.de>
Date: Fri, 05 Mar 2010 11:49:57 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Brian Trammell <trammell@tik.ee.ethz.ch>
References: <3146E35B-9877-44AE-B884-17541E6ED5B8@tik.ee.ethz.ch> <4B8C0943.6070309@net.in.tum.de> <A89AA32C-086C-46B9-8B6F-66BBEA645602@tik.ee.ethz.ch>
In-Reply-To: <A89AA32C-086C-46B9-8B6F-66BBEA645602@tik.ee.ethz.ch>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms020702090903060702000507"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] WGLC comments on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 05 Mar 2010 10:49:49 -0000

This is a cryptographically signed message in MIME format.

--------------ms020702090903060702000507
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Brian,

Thanks a lot for your feedback.

Maybe you can have a quick look at my proposals, so that I can close=20
these issues (or at least most of them).


Brian Trammell wrote:
> Hi, Gerhard, all,
>=20
> On Mar 1, 2010, at 7:36 PM, Gerhard Muenz wrote:
>=20
>> Hi Brian,
>>=20
>> Thanks a lot for your review.
>>=20
>> Data models in network management will never satisfy all people.
>> So, please keep in mind that the goal of this document is to
>> standardize a configuration data model that can be easily
>> implemented by many vendors. There will always be additional
>> vendor-specific parameters, and there will always be
>> vendor-specific restrictions of the model. Thanks to YANG, there
>> are possibilities to make restrictions and extension.
>>=20
>> So, ask yourself if the parameters which are contained in the
>> configuration data model can be easily mapped to one of the YAF
>> command line options. Not the other way round.
>=20
> Yep, I understand the point of the data model here... and YAF's
> configuration by its nature will most probably remain command-line
> (indeed I was thinking of implementing this in a Ruby script that
> kicked off YAF); the references to YAF herein were merely an attempt
> to use one specific implementation (with which I am familiar) as a
> tool for evaluating the flexibility and general applicability of the
> configuration data model. "This is out of scope" is a perfectly
> reasonable response. bue I wanted to make sure in reviewing that the
> flexibility we can get out of yang is presented in the right way...
> Indeed, in the selector->cache->ep chain, you can probably just model
> intermediate processes (like YAF's partial reassembly) as being part
> of the Cache, and that's just fine.

In an ipfix-mediator-config draft, you can use YANG to extend the model=20
to allow  CP->IP->EP  and  OP->SP->Cache->IP->EP  setups.

Speaking in UML, we are not restricted to the existing classes, we=20
extend the model by adding new ones.

>> More comments inline.
>>=20
>> Gerhard
>>=20
>>=20
>> Brian Trammell wrote:
>>> Greetings, all, I have reviewed
>>> draft-ietf-ipfix-configuration-model-04 in response to the WGLC,
>>> and have the following comments, ranked in order of impact; most
>>> of these deal with the applicability of the configuration data
>>> model to implementations other than the "exemplar"=20
>>> implementation(s) the authors had in mind while while writing.  I
>>>  will say that overall, the draft does an more than reasonable
>>> job of being generally applicable. I apologize for starting a new
>>> thread, but I have not yet had time to properly review other
>>> comments, but plan on being in Anaheim prepared to discuss those
>>> too. I have also explicitly not reviewed the YANG definition; I
>>> leave that to people with actual YANG expertise. First and
>>> foremost is the issue of caches, as it is in cache implementation
>>> that IPFIX packet-to-flow MPs will differentiate the most from
>>> each other, and indeed most vendors should be expected to=20
>>> subclass this cache class for their own configuration parameters;
>>>  therefore getting this right is probably the most crucial.
>>> cacheMode immediate seems like a bit of a hack, as it invalidates
>>> basically every other cache field. However, I suppose I can see
>>> it as a way to keep the class design clean (otherwise, there
>>> would be nothing to connect OP to layout/EP in a PSAMP
>>> implementation...).
>> Right.
>>=20
>>> 1. I'm concerned about the definition of cacheMode timeout (4.3,
>>> p. 24), as it seems to make it impossible to configure a cache to
>>> allow cache exit on TCP state tracking. Perhaps this is merely an
>>> editorial oversight?
>> We can allow a timeout Cache to expire a flow after a certain flow
>> termination event. Is this what you propose?
>=20
> Basically, yes. Most flow generators will expire a flow after a TCP
> FIN or RST (or some number of these), so there is a difference
> between expiring flows only on timeout and expiring on timeout as
> well as "natural" end of flow (see the definition of flowEndReason,
> IE 136). So I would propose the following:
>=20
> cacheMode:  Configures the Cache Mode.  The following parameter=20
> values are specified by the configuration data model: *  immediate:
> Data Records expire after the first packet *  timeout: Data Records
> expire after active or inactive timeout |    *  natural: Data Records
> expire after active or inactive timeout, or |       on natural
> termination (e.g. TCP FIN, or TCP RST) as determined |       by the
> Metering Process. *  permanent: Data Records never expire, but are
> periodically exported with interval set by exportInterval In the case
> of "immediate", PSAMP Packet Reports are generated. Otherwise, IPFIX
> Flow Records are generated.

Ok. I change your text as follows (to avoid mentioning Metering Process=20
here):

   * natural: Data Records expire after active or inactive timeout, or on=

              natural termination (e.g. TCP FIN, or TCP RST) of the Flow

We must be aware of the fact that there are reasons to prematurely=20
expire flows (e.g., full cache). I assume that we do not have to mention =

these, do we?

> Alternately, if "timeout" meant "timeout + natural" in this
> terminology, we could change it as follows:
>=20
> cacheMode:  Configures the Cache Mode.  The following parameter=20
> values are specified by the configuration data model: *  immediate:
> Data Records expire after the first packet |    *  timeout: Data
> Records expire after active or inactive timeout, or |       on
> natural termination (e.g. TCP FIN, or TCP RST) as determined |
> by the Metering Process. *  permanent: Data Records never expire, but
> are periodically exported with interval set by exportInterval In the
> case of "immediate", PSAMP Packet Reports are generated. Otherwise,
> IPFIX Flow Records are generated.
>=20
> I'm inclined toward the first alternative, with 'natural' as default
> and 'timeout' provided for MPs that don't want to be bothered with
> TCP state.

Currently, there is no default value. The user must configure the=20
cacheMode parameter. I think, this is useful because it forces the user=20
to think about what he wants to meter.

>>> 2. Destination (4.4.1, pp .27-29) appears not to match the
>>> properties of the particular transports:
>>>=20
>>> a. can't specify a local IP address for TCP or UDP. This seems=20
>>> necessary on devices with multiple management interfaces, which I
>>>  don't think is all that weird a deployment plan.
>> This will be fixed as proposed in this mail unless anybody objects:
>>  http://www.ietf.org/mail-archive/web/ipfix/current/msg05316.html
>=20
> Perfect.
>=20
>>> b. can't specify 1..n remote IP addresses for SCTP. This breaks
>>> one of the advantages of SCTP.
>> I see send_connectx() in the SCTP API:=20
>> http://tools.ietf.org/html/draft-ietf-tsvwg-sctpsocket-21#section-8.9
>>=20
>>=20
>> However, what happens if you configure multiple remote IP addresses
>> and they do not belong to the same SCTP endpoint? What shall the
>> Exporting Process do in this situation?
>=20
> What sctp_connectx() does:
>=20
> "The way the SCTP stack uses the list of addresses to set up the
> association is implementation dependent.  This function only
> specifies that the stack will try to make use of all the addresses in
> the list when needed." The assumptions are 1. that the list of
> addresses all correspond to a logically unitary CP and 2. that the
> admin knows what he's doing.
>=20
>> I'm not sure if it makes sense to configure multiple remote
>> addresses for a single destination. It seems to be safer to
>> configure multiple export destinations and to use export member
>> type primary and secondary.
>>=20
>> What do you think?
>=20
> These are logically separate concepts, and should not be conflated by
> the data model. Multiple Destinations means multiple CPs: endpoint
> redundancy, and the assumption is that each Destination is its own CP
> with separate app-layer state. Multiple addresses means multiple
> endpoints on a single CP for session setup: _path_ redundancy, and
> the assumption is that every given address leads to the same CP, with
> the same app-layer state. Different tools for different situations.
> Of course you can configure to export to two addresses that aren't
> the same CP, just like you can configure an EP to export to a
> destination that _isn't_ a CP. In both cases this is a configuration
> error.

New multiplicity factor for destinationIPAddress:

    +--------------------------------------------------+
    | Destination                                      |
    +--------------------------------------------------+
    | name                                             |<>-----+
    | exportMemberType =3D parallel                      |       | 0..1
    | ipfixVersion =3D 10                                |       |
    | transportProtocol                                |   +------------+=

    | sourceIpAddress[0..*] {[0..1] for UDP and TCP}   |   | Transport- |=

    | destinationIpAddress[1..*] {[1] for UDP and TCP} |   | Layer-     |=

    | destinationPort =3D 4739|4740                      |   | Security  =
 |
    | ifName/ifIndex[0..1]                             |   +------------+=

    | sendBufferSize {opt.}                            |
    | rateLimit[0..1]                                  |
    | timedReliability =3D 0 {SCTP only}                 |
    | numberOfStreams {opt.} {SCTP only}               |
    | templateRefreshTimeout =3D 600 {UDP only}          |
    | optionsTemplateRefreshTimeout =3D 600 {UDP only}   |
    | templateRefreshPacket[0..1] {UDP only}           |
    | optionsTemplateRefreshPacket[0..1] {UDP only}    |
    +--------------------------------------------------+

                        Figure 14: Destination class

[...]

    destinationIpAddress:  Destination IP address to which IPFIX Messages=

       are sent (i.e., the IP address of the Collector).  This parameter
       MUST be configured by the user.
       If the transport protocol is SCTP, the parameter MAY appear
       multiple times to specify multiple destination IP addresses.  The
       user MUST ensure that all configured IP addresses belong to the
       same Collector.  The Exporting Process tries to establish an SCTP
       association to any of the configured destination IP addresses.


Is this ok?


>>> c. can't configure SCTP reliability per-template, as opposed to
>>> per destination. see 2.b.
>> As discussed long ago, Templates are not configured. The are
>> created by the Exporting Process depending on the Data Records to
>> be exported.
>=20
> I'm not convinced this choice is the right one (both the Cache and
> the EP are valid places to make choices about what, precisely, will
> be exported: see below the discussion on

Some people think that it is more natural to configure what you want to=20
measure, and not how the information is transported to the Collector.

>> Therefore, there is no good place for a per-Template reliability
>> parameter in the configuration data model. The only possibility
>> would be to configure the reliability in the Cache, and this would
>> be a very strange place to do it.
>>=20
>> I had discussed this issue with Benoit. The conclusion was that we
>> have a per-Destination parameter. If people want to configure more,
>> they can come up with an extension.
>>=20
>> If you have another convincing solution, please let me know.
>=20
> Not really. But now we as a WG are saying logically inconsistent
> things: I can't reconcile the "templates are applications and the
> reliability settings are per-application; here is a way to estimate
> loss across multiple templates to the same destination" view of sctp
> per stream export and the "destination-level reliability" view
> presented here. It is an inelegant contradiction.

Surely, the parameter must not inhibit the utilization of the=20
per-SCTP-stream extension.

Therefore, I changed the parameter description as follows:

    timedReliability:  Lifetime in milliseconds until an IPFIX Message
       containing Data Sets only is "abandoned" due to the timed
       reliability mechanism of PR-SCTP [RFC3758].  If this parameter is
       set to zero, reliable SCTP transport is used for all Data Records.=

       Regardless of the value of this parameter, the Exporting Process
       MAY use reliable SCTP transport for Data Sets associated with
       Options Templates.

Ok?

> My personal opinion would be to provide template-level configuration
> as an EP option in an extension, and to simply ignore this
> Destination option (much the same way as I work as hard as I can to
> ignore UDP ;) )
 >
>>> 3. Options (4.4.3, p. 30) configuration is rather odd and highly=20
>>> restrictive, even to the point of being unusable. The biggest
>>> problem is that RFC5473 common properties are something
>>> completely different from the options specified in 5101 and 5476.
>>> "Enable common properties", without some definition of which
>>> fields to use as common properties and under what circumstances,
>>> is not a useful directive.
>> The idea is that the Exporting Process defines common properties
>> depending on the Data Records to be exported. This can be done
>> dynamically (depending on the value distribution) or using
>> knowledge about the configuration (e.g., if a filter ensures that
>> the accounted packets are from/to a specific endpoint, this can be
>> encoded as common property).
>>=20
>> If you think that reducingRedundancy is not useful, I will remove
>> from the list.
>=20
> It's certainly useful to have a switch, but we must then be very
> explicit about the fact that when you turn this switch on, it's up to
> the EP to determine precisely what redundancy to reduce and how, that
> this is implementation dependent, and that implementations are
> encouraged to provide implementation-specific configuration
> extensions to cover this case.

    reducingRedundancy:  Enables the utilization of Options Templates to
       reduce redundancy in the exported Data Records according to
       [RFC5473].  The Exporting Process decides when to apply these
       Options Templates.

Ok?

>>> Secondarily, options specified in 5610 and 5655 are not covered;
>>> do we really want to amend this document every time we specify
>>> another WG-blessed options template?
>> Apart from reducingRedundancy, all mentioned Options type appear in
>> the protocol RFCs (RFC5101, RFC5476).
>=20
> 5655 options don't make much sense to configure, as runtime
> determination and implementation dependency probably dominate there.
> However, type export as in 5610 seems like it would be a useful
> configuration switch.

    extendedTypeInformation:  Export of extended type information for
       enterprise-specific Information Elements used in the exported
       Templates [RFC5610].

Ok?

> Although this discussion reminds me of another issue I missed,
> elaborated below.
>=20
>> Adding more parameter values for optionsType can be easily achieved
>> with YANG extensions because they are specified as identities (it
>> would note be possible with an enum type).
>=20
> Hm. Okay. Will we create then a "defined Options Template" registry
> somewhere?

We could think about this, yes.
At the moment, they are spread over many, many documents.

>>> I'd suggest using an OptionsTemplate subclass of Template,
>>> instead, and providing the definition of common (non-5473)
>>> options templates in an appendix to the configuration data model.
>>>=20
>> I do not understand your proposal. There is no Template class for
>> configuration.
>=20
> I missed this point somehow (perhaps in my assumption that Templates
> indeed should be configurable, but I did not show up to that
> discussion so I accept the decision of the rest of the WG)... but
> yes, indeed, so, withdrawn.
>=20
>>> 4. There does not seem to be any room for logical configurable=20
>>> processes between selection and cache, or cache and EP. The
>>> example that comes to mind is perhaps implementation-specific
>>> [I'm mentioning this because yaf does it]: fragmentation behavior
>>> (reassemble, ignore, drop). How could I allow then for the
>>> configuration of the fragment table (given that a selection
>>> process can only reference a cache)?
>> You can always extend the model. For example, if fragmentation
>> reassembly can be regarded as part of the Cache, you can easily
>> define your own new cacheType (again, identities and identityref in
>> YANG).
>>=20
>> We also have implemented a lot of cool additional functionalities
>> which we are going to configure by extending the model. It is not
>> the purpose of this document to cover all of these.
>=20
> Indeed. As long as you can collapse these into Cache (which as I said
> above it seems you can), that's more than sufficient.
>=20
>>> 5. CacheLayout (4.3.1, p. 25) does not appear to be able to
>>> support adaptive reduced-length export (exporting, say, counters
>>> with unsigned32 unless one of them actually requires unsigned64)
>>> [yes, yaf does this too]. Though this could probably be solved
>>> with a deviation, and an additional vendor-specific CacheField
>>> attribute...
>> Is reduced size encoding performed by the Cache or by the Exporting
>> Process?
>=20
> In YAF, there's no process separation, but from a functional block
> separation this is done in the EP.
>=20
>> If you implement a Cache in hardware, it is probably easier to
>> always use the maximum length and to reduce it later in the
>> Exporting Process. We assume that the Cache reserves a certain
>> field length for all Data Records in the Cache.  (Variable-length
>> fields are treated differently)
>=20
> Indeed, I would expect that only caches written in dynamically typed
> languages, where the language itself autopromotes storage on
> overflow, would do variable-sized caching of fixed length elements,
> if only because doing it in C is rather time expensive and difficult
> for relatively little benefit. And in this case the variable-sized
> caching is invisible to the programmer.
>=20
>> The ieLength specifies the field length in the Cache. The length in
>> the exported Data Record can be different. Note again: we are not
>> configuring a Template, here. The exported Data Record may even
>> contain a completely different set of fields, for example in the
>> case of reducing redundancy.
>=20
> Ah, okay, so it supports it, it just doesn't support configuring it
> (which can be an implementation dependent extension in either the
> cache, the EP, or both, which makes sense depending on how it's
> implemented). This is okay.
>=20
>> What about removing the following sentence to avoid any confusion:
>>=20
>> For Information Elements of integer and float type, the field
>> length MAY be set to a smaller value than the standard length of
>> the abstract data type if the rules of reduced size encoding are
>> fulfilled (see [RFC5101], Section 6.2).
>=20
> No reason to cut that, as it wasn't the point of confusion...

I want to make this clear. What about the following new paragraph?

    Note that the Data Records generated by the Cache MAY be modified by
    the Exporting Process before sending them to a Collector.  For
    example, the Exporting Process MAY shorten the length of a field
    according to the rules of reduced size encoding [RFC5101].  The
    Exporting Process MAY also export certain fields in a separate Data
    Record as described in [RFC5476].


>>> 6. CacheLayout (4.3.1, p. 25) could do with an explicit example
>>> to illustrate the "unless the corresponding IE is not applicable
>>> or cannot be derived" case: I presume this is intended for
>>> dual-stack IPv4/IPv6 measurement, RFC5103 biflows, and the like?
>>> It does appear than this clause is sufficient for such cases, but
>>> I had to read this a couple of times before I understood that's
>>> what was meant.
>> Right. I tried to improve this. Please look into my reply to
>> Lothar's review:=20
>> http://www.ietf.org/mail-archive/web/ipfix/current/msg05325.html
>=20
> Perfect.
>=20
>>> 7. Real File Readers and File Writers will probably need more
>>> than a single filename endpoint on the file system (however, most
>>> file readers and file writers in document-based workflows won't
>>> be configured using this information model...) How to handle this
>>> is probably implementation-specific; in any case, I don't have a
>>> good suggestion here, yet; perhaps we can discuss in Anaheim.
>> I assume that the URI filename parameter is useful for many
>> FileWriter/Reader implementations.  All additional parameters are
>> probably very specific to the implementation.
>=20
> Indeed. Okay, makes sense to leave it then, maybe with a statement
> that the handling of the URI is implementation dependent.

For simple FileWriter/Readers, a single file is enough, and the=20
parameter description is clear in this case.

More extended FileWriter/Readers need to extend the model anyway, which=20
means that they can also adapt the meaning of the file parameter.

So, rather no further statement which may cause confusion.

>>> 8. Is there any benefit to cascading caches as with selection=20
>>> processes, or rather, are the benefits of cascading caches
>>> considered in scope for this application? The simplest example
>>> would be simultaneous export of flow data as well as flow
>>> aggregate data (e.g. traffic counters per AS pair), perhaps as
>>> part of a cascading-aggregation mediator. I ask only because it
>>> seems a rather small and easily supportable change to greatly
>>> increase flexibility.
>> Uff. I already have problems to convince people that there may be
>> more than one parallel Caches in a Metering Process. Now, you even
>> propose multiple Caches in a series :)
>>=20
>> IMO, aggregation is done in an Intermediate Process, not in a
>> Metering Process and therefore not in the Cache of a Metering
>> Process.
>>=20
>> The configuration of Mediators and Intermediate Process is out of
>> scope of this document. Otherwise, we will never finish...
>=20
> Accepted. But now I have an idea of at least one subsection heading
> in the Mediator Configuration Data Model draft :)
>=20
>>> 9. Destination exportMemberType appears to reference a renamed
>>> MIB object as of MIB-10; this is called ipfixExportMemberType
>>> now. I would suggest copying the complete description from MIB,
>>> because the semantics are kind of confusing otherwise (I had to
>>> read MIB, which I probably should have done by now but oh well,
>>> in order to figure out what the difference between parallel and
>>> primary was). A tertiary problem: these exportMemberTypes seem a
>>> bit restrictive (e.g. what if I want to do parallel export and
>>> also have secondaries?), but I suppose that's a MIB comment.
>> Andrew suggested to improve this as well.  See:=20
>> http://www.ietf.org/mail-archive/web/ipfix/current/msg05323.html=20
>> I'm open to concrete suggestions.
>=20
> Update the text in the configuration model to match that in the MIB,
> copy the definitions in full from the MIB so you don't have to go
> read the MIB to figure out what these mean, and call it finished. I
> expect most people will either find the MIB approach sufficient, or
> simply leave these parallel and implement their own redundancy.

Ok, I will follow your proposal unless I find a better way to configure=20
this stuff.

>>> 10. draft-ietf-ipfix-file is now RFC5655 (the editor will catch=20
>>> this); since this is now an RFC, references to things "proposed"
>>> in this draft should probably be changed to things "specified" in
>>> RFC 5655.
>> Done.
>=20
> And, as promised, a new 11. Cache should have a way to configure
> biflow assembly and export. The simplest one is simply to state that
> the presence of reverse Information Elements in a CacheLayout implies
> that the Cache will assemble biflows suitable for export as described
> in RFC 5103.

Please check the following:


    ieId, ieName, ieEnterpriseNumber:  These parameters specify a field
       by the combination of the Information Element identifier or name,
       and the Information Element enterprise number.  Either ieId or
       ieName MUST be specified.  ieEnterpriseNumber MUST be used for
       enterprise-specific Information Elements.  If ieEnterpriseNumber
       is omitted or zero, this is Information Element is not enterprise-=

       specific but registered at IANA.
       If the enterprise number is set to 29305, this field contains a
       Reverse Information Element.  In this case, the Cache MUST
       generate Data Records in accordance to [RFC5103].

[...]

    isFlowKey:  If present, this field is a Flow Key. If the field
       contains a Reverse Information Element, it MUST NOT be configured
       as Flow Key.


> This opens a slightly larger question, though: is this a
> configuration data model for 5101 and 5102, for all the RFCs
> published by the WG to date (i.e., adding 5103, 5473, 5610, and maybe
> 5655), for all the work completed by the WG (adding per-stream), or
> for all the work accepted aside from mediation (adding flow
> selection, and possibly anonymisation in a non-mediator context)? I
> think the last is almost certainly out of scope (otherwise we're
> never done). Thoughts?

For many of the RFC you mention above do not specify any configuration=20
requirements. Often, there exist zero to one implementation. So, it is=20
difficult to say whether the parameters required by these=20
implementations can be assumed to be "common" configuration parameters.

So, I assume, we need more implementations until we can define a=20
complete configuration data model for them.

Regards,
Gerhard


>=20
> Cheers,
>=20
> Brian

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms020702090903060702000507
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature

MIAGCSqGSIb3DQEHAqCAMIACAQExCzAJBgUrDgMCGgUAMIAGCSqGSIb3DQEHAQAAoIIO5jCC
BEYwggOvoAMCAQICEGb9R+PCGeToms2Z3fU6yyQwDQYJKoZIhvcNAQEFBQAwXzELMAkGA1UE
BhMCVVMxFzAVBgNVBAoTDlZlcmlTaWduLCBJbmMuMTcwNQYDVQQLEy5DbGFzcyAxIFB1Ymxp
YyBQcmltYXJ5IENlcnRpZmljYXRpb24gQXV0aG9yaXR5MB4XDTA1MTAyODAwMDAwMFoXDTE1
MTAyNzIzNTk1OVowgd0xCzAJBgNVBAYTAlVTMRcwFQYDVQQKEw5WZXJpU2lnbiwgSW5jLjEf
MB0GA1UECxMWVmVyaVNpZ24gVHJ1c3QgTmV0d29yazE7MDkGA1UECxMyVGVybXMgb2YgdXNl
IGF0IGh0dHBzOi8vd3d3LnZlcmlzaWduLmNvbS9ycGEgKGMpMDUxHjAcBgNVBAsTFVBlcnNv
bmEgTm90IFZhbGlkYXRlZDE3MDUGA1UEAxMuVmVyaVNpZ24gQ2xhc3MgMSBJbmRpdmlkdWFs
IFN1YnNjcmliZXIgQ0EgLSBHMjCCASIwDQYJKoZIhvcNAQEBBQADggEPADCCAQoCggEBAMnf
rOfq+PgDFMQAktXBfjbCPO98chXLwKuMPRyVzm8eECw/AO2XJua2x+atQx0/pIdHR0w+VPhs
+Mf8sZ69MHC8l7EDBeqV8a1AxUR6SwWi8mD81zplYu//EHuiVrvFTnAt1qIfPO2wQuhejVch
rKaZ2RHp0hoHwHRHQgv8xTTq/ea6JNEdCBU3otdzzwFBL2OyOj++pRpu9MlKWz2VphW7NQIZ
+dTvvI8OcXZZu0u2Ptb8Whb01g6J8kn+bAztFenZiHWcec5gJ925rXXOL3OVekA6hXVJsLjf
aLyrzROChRFQo+A8C67AClPN1zBvhTJGG+RJEMJs4q8fef/btLUCAwEAAaOB/zCB/DASBgNV
HRMBAf8ECDAGAQH/AgEAMEQGA1UdIAQ9MDswOQYLYIZIAYb4RQEHFwEwKjAoBggrBgEFBQcC
ARYcaHR0cHM6Ly93d3cudmVyaXNpZ24uY29tL3JwYTALBgNVHQ8EBAMCAQYwEQYJYIZIAYb4
QgEBBAQDAgEGMC4GA1UdEQQnMCWkIzAhMR8wHQYDVQQDExZQcml2YXRlTGFiZWwzLTIwNDgt
MTU1MB0GA1UdDgQWBBQRfV4ZfTwE32ps1qKKGj8x2DuUUjAxBgNVHR8EKjAoMCagJKAihiBo
dHRwOi8vY3JsLnZlcmlzaWduLmNvbS9wY2ExLmNybDANBgkqhkiG9w0BAQUFAAOBgQA8o9oC
YzrEk6qrctPcrVA4HgyeFkqIt+7r2f8PjZWg1rv6aguuYYTYaEeJ70+ssh9JQZtJM3aTi55u
uUMcYL3C3Ioth8FFwBFyBBprJCpsb+f8BxMp0Hc6I+f1wYVoGb/GAVQgGa41gsxiPGEJxvTV
67APpp8zhZrTcY5Qj5ndYjCCBUowggQyoAMCAQICEF0kYdlVNx5E1qcmxun5IrswDQYJKoZI
hvcNAQEFBQAwgd0xCzAJBgNVBAYTAlVTMRcwFQYDVQQKEw5WZXJpU2lnbiwgSW5jLjEfMB0G
A1UECxMWVmVyaVNpZ24gVHJ1c3QgTmV0d29yazE7MDkGA1UECxMyVGVybXMgb2YgdXNlIGF0
IGh0dHBzOi8vd3d3LnZlcmlzaWduLmNvbS9ycGEgKGMpMDUxHjAcBgNVBAsTFVBlcnNvbmEg
Tm90IFZhbGlkYXRlZDE3MDUGA1UEAxMuVmVyaVNpZ24gQ2xhc3MgMSBJbmRpdmlkdWFsIFN1
YnNjcmliZXIgQ0EgLSBHMjAeFw0xMDAxMDIwMDAwMDBaFw0xMTAxMDIyMzU5NTlaMIIBEzEX
MBUGA1UEChMOVmVyaVNpZ24sIEluYy4xHzAdBgNVBAsTFlZlcmlTaWduIFRydXN0IE5ldHdv
cmsxRjBEBgNVBAsTPXd3dy52ZXJpc2lnbi5jb20vcmVwb3NpdG9yeS9SUEEgSW5jb3JwLiBi
eSBSZWYuLExJQUIuTFREKGMpOTgxHjAcBgNVBAsTFVBlcnNvbmEgTm90IFZhbGlkYXRlZDEz
MDEGA1UECxMqRGlnaXRhbCBJRCBDbGFzcyAxIC0gTmV0c2NhcGUgRnVsbCBTZXJ2aWNlMRYw
FAYDVQQDFA1HZXJoYXJkIE11ZW56MSIwIAYJKoZIhvcNAQkBFhNtdWVuekBuZXQuaW4udHVt
LmRlMIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAl/YDW9cnAXJjesGhwtGwj9JN
IHWXQ7QA18s+wGZNsQY0k4eKBeLTk/RAiC1hP+uQAXLoGQHQF/ljjQCQeRmnsN6w04iAzKVo
Ru+NP7Y6ipn8uztx+z0Y3w43CO+LWUudp3PIF3+Fm2zwsAFXj/IYy/8Kq6m9+nQgYO/D4HoN
DDR3//GepXghK0x93gR0GCkm8pgKINHNBD47/IZbDe/cIpAC9HaoxCeLiB1QaCDIYPIrJ0Vx
kk0crAA0tPaMDjdSQTmNRAn2P+7Wg7K+DLvauBHfxLs8PrsolI+XdwiBu6henK9nPxvtoHmm
mfxcdslKLbS9Xz2rHxi6PVVgv1ubWwIDAQABo4HMMIHJMAkGA1UdEwQCMAAwRAYDVR0gBD0w
OzA5BgtghkgBhvhFAQcXATAqMCgGCCsGAQUFBwIBFhxodHRwczovL3d3dy52ZXJpc2lnbi5j
b20vcnBhMAsGA1UdDwQEAwIFoDAdBgNVHSUEFjAUBggrBgEFBQcDBAYIKwYBBQUHAwIwSgYD
VR0fBEMwQTA/oD2gO4Y5aHR0cDovL0luZEMxRGlnaXRhbElELWNybC52ZXJpc2lnbi5jb20v
SW5kQzFEaWdpdGFsSUQuY3JsMA0GCSqGSIb3DQEBBQUAA4IBAQBPU7OeWNo2LRzsdu/RuuAk
L5cj/RzeZ0xv9q4K/eosy9dRu45XACpPvb0fjc3u/dsY4m3jDqzUsB02IrrnF9jpEo10BJPw
b/RZ7NaU7jC/gEknL3bSXiD2PVwL4ZI5ChH5+TJUboWhVh+FpOwOfP/yKk7wQM/46iXXzRIz
enk0rcJp71k4aQgQTML3/8MDeD6TUipuQwQMY9n6T6DsNd/6JpGa+x9ZX4HIEG6+EWwQ85G5
vGfDBaq3dFrksHE75u7b7pf75wm9HIFU2SDHP2tymTHJt/XM9sqP1URIC57s+uZV0/d5KRc1
uPcJ0FAnAiC9o7ybPIh27shiq0FvOEwfMIIFSjCCBDKgAwIBAgIQXSRh2VU3HkTWpybG6fki
uzANBgkqhkiG9w0BAQUFADCB3TELMAkGA1UEBhMCVVMxFzAVBgNVBAoTDlZlcmlTaWduLCBJ
bmMuMR8wHQYDVQQLExZWZXJpU2lnbiBUcnVzdCBOZXR3b3JrMTswOQYDVQQLEzJUZXJtcyBv
ZiB1c2UgYXQgaHR0cHM6Ly93d3cudmVyaXNpZ24uY29tL3JwYSAoYykwNTEeMBwGA1UECxMV
UGVyc29uYSBOb3QgVmFsaWRhdGVkMTcwNQYDVQQDEy5WZXJpU2lnbiBDbGFzcyAxIEluZGl2
aWR1YWwgU3Vic2NyaWJlciBDQSAtIEcyMB4XDTEwMDEwMjAwMDAwMFoXDTExMDEwMjIzNTk1
OVowggETMRcwFQYDVQQKEw5WZXJpU2lnbiwgSW5jLjEfMB0GA1UECxMWVmVyaVNpZ24gVHJ1
c3QgTmV0d29yazFGMEQGA1UECxM9d3d3LnZlcmlzaWduLmNvbS9yZXBvc2l0b3J5L1JQQSBJ
bmNvcnAuIGJ5IFJlZi4sTElBQi5MVEQoYyk5ODEeMBwGA1UECxMVUGVyc29uYSBOb3QgVmFs
aWRhdGVkMTMwMQYDVQQLEypEaWdpdGFsIElEIENsYXNzIDEgLSBOZXRzY2FwZSBGdWxsIFNl
cnZpY2UxFjAUBgNVBAMUDUdlcmhhcmQgTXVlbnoxIjAgBgkqhkiG9w0BCQEWE211ZW56QG5l
dC5pbi50dW0uZGUwggEiMA0GCSqGSIb3DQEBAQUAA4IBDwAwggEKAoIBAQCX9gNb1ycBcmN6
waHC0bCP0k0gdZdDtADXyz7AZk2xBjSTh4oF4tOT9ECILWE/65ABcugZAdAX+WONAJB5Gaew
3rDTiIDMpWhG740/tjqKmfy7O3H7PRjfDjcI74tZS52nc8gXf4WbbPCwAVeP8hjL/wqrqb36
dCBg78Pgeg0MNHf/8Z6leCErTH3eBHQYKSbymAog0c0EPjv8hlsN79wikAL0dqjEJ4uIHVBo
IMhg8isnRXGSTRysADS09owON1JBOY1ECfY/7taDsr4Mu9q4Ed/Euzw+uyiUj5d3CIG7qF6c
r2c/G+2geaaZ/Fx2yUottL1fPasfGLo9VWC/W5tbAgMBAAGjgcwwgckwCQYDVR0TBAIwADBE
BgNVHSAEPTA7MDkGC2CGSAGG+EUBBxcBMCowKAYIKwYBBQUHAgEWHGh0dHBzOi8vd3d3LnZl
cmlzaWduLmNvbS9ycGEwCwYDVR0PBAQDAgWgMB0GA1UdJQQWMBQGCCsGAQUFBwMEBggrBgEF
BQcDAjBKBgNVHR8EQzBBMD+gPaA7hjlodHRwOi8vSW5kQzFEaWdpdGFsSUQtY3JsLnZlcmlz
aWduLmNvbS9JbmRDMURpZ2l0YWxJRC5jcmwwDQYJKoZIhvcNAQEFBQADggEBAE9Ts55Y2jYt
HOx279G64CQvlyP9HN5nTG/2rgr96izL11G7jlcAKk+9vR+Nze792xjibeMOrNSwHTYiuucX
2OkSjXQEk/Bv9Fns1pTuML+ASScvdtJeIPY9XAvhkjkKEfn5MlRuhaFWH4Wk7A58//IqTvBA
z/jqJdfNEjN6eTStwmnvWThpCBBMwvf/wwN4PpNSKm5DBAxj2fpPoOw13/omkZr7H1lfgcgQ
br4RbBDzkbm8Z8MFqrd0WuSwcTvm7tvul/vnCb0cgVTZIMc/a3KZMcm39cz2yo/VREgLnuz6
5lXT93kpFzW49wnQUCcCIL2jvJs8iHbuyGKrQW84TB8xggTsMIIE6AIBATCB8jCB3TELMAkG
A1UEBhMCVVMxFzAVBgNVBAoTDlZlcmlTaWduLCBJbmMuMR8wHQYDVQQLExZWZXJpU2lnbiBU
cnVzdCBOZXR3b3JrMTswOQYDVQQLEzJUZXJtcyBvZiB1c2UgYXQgaHR0cHM6Ly93d3cudmVy
aXNpZ24uY29tL3JwYSAoYykwNTEeMBwGA1UECxMVUGVyc29uYSBOb3QgVmFsaWRhdGVkMTcw
NQYDVQQDEy5WZXJpU2lnbiBDbGFzcyAxIEluZGl2aWR1YWwgU3Vic2NyaWJlciBDQSAtIEcy
AhBdJGHZVTceRNanJsbp+SK7MAkGBSsOAwIaBQCgggLOMBgGCSqGSIb3DQEJAzELBgkqhkiG
9w0BBwEwHAYJKoZIhvcNAQkFMQ8XDTEwMDMwNTEwNDk1OFowIwYJKoZIhvcNAQkEMRYEFEUA
rlm9n2BkmTjqGpVOSmsng0m9MF8GCSqGSIb3DQEJDzFSMFAwCwYJYIZIAWUDBAECMAoGCCqG
SIb3DQMHMA4GCCqGSIb3DQMCAgIAgDANBggqhkiG9w0DAgIBQDAHBgUrDgMCBzANBggqhkiG
9w0DAgIBKDCCAQMGCSsGAQQBgjcQBDGB9TCB8jCB3TELMAkGA1UEBhMCVVMxFzAVBgNVBAoT
DlZlcmlTaWduLCBJbmMuMR8wHQYDVQQLExZWZXJpU2lnbiBUcnVzdCBOZXR3b3JrMTswOQYD
VQQLEzJUZXJtcyBvZiB1c2UgYXQgaHR0cHM6Ly93d3cudmVyaXNpZ24uY29tL3JwYSAoYykw
NTEeMBwGA1UECxMVUGVyc29uYSBOb3QgVmFsaWRhdGVkMTcwNQYDVQQDEy5WZXJpU2lnbiBD
bGFzcyAxIEluZGl2aWR1YWwgU3Vic2NyaWJlciBDQSAtIEcyAhBdJGHZVTceRNanJsbp+SK7
MIIBBQYLKoZIhvcNAQkQAgsxgfWggfIwgd0xCzAJBgNVBAYTAlVTMRcwFQYDVQQKEw5WZXJp
U2lnbiwgSW5jLjEfMB0GA1UECxMWVmVyaVNpZ24gVHJ1c3QgTmV0d29yazE7MDkGA1UECxMy
VGVybXMgb2YgdXNlIGF0IGh0dHBzOi8vd3d3LnZlcmlzaWduLmNvbS9ycGEgKGMpMDUxHjAc
BgNVBAsTFVBlcnNvbmEgTm90IFZhbGlkYXRlZDE3MDUGA1UEAxMuVmVyaVNpZ24gQ2xhc3Mg
MSBJbmRpdmlkdWFsIFN1YnNjcmliZXIgQ0EgLSBHMgIQXSRh2VU3HkTWpybG6fkiuzANBgkq
hkiG9w0BAQEFAASCAQB1lKaIomdoZe4qXwxoZfpCtq72I5DtWOI6PTI6155ZHM144PYT780F
nXfc8lYBgf4V/wKGNuf7u6mkaTsi7PXpBsGaykhn/u63tu5YuFfAgLLJUqtRYqboF+6V8Gfc
fuQqcgIwhW5ym0PL7R/hAY11R2q9sSJUYJ/M9tqotwONA8D0CB1wHeJq5aedi6wOz4S2+Q90
1XYn0PmuaupVmSz9KqPDf69cS8h6Sk66j0PewpV1hFFWH1Qjf06/kkh8lA+LHBODYmUIEVD5
NyNe//wctQ2E2+X20Vd3JH/MVARu0VB10jH2MISwLwpfitjCInLewRYbJvbmGBviY73Mf0ut
AAAAAAAA
--------------ms020702090903060702000507--

From trammell@tik.ee.ethz.ch  Fri Mar  5 03:55:34 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 1E0E83A8972 for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 03:55:34 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.599
X-Spam-Level: 
X-Spam-Status: No, score=-6.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id kDSQrap5KvGQ for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 03:55:31 -0800 (PST)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id EC9C33A89AA for <ipfix@ietf.org>; Fri,  5 Mar 2010 03:55:30 -0800 (PST)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id 11719D93D5; Fri,  5 Mar 2010 12:55:31 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id CPXPjItj2z8G; Fri,  5 Mar 2010 12:55:30 +0100 (MET)
Received: from pb-10072.ethz.ch (pb-10072.ethz.ch [82.130.103.195]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id 4074DD9333; Fri,  5 Mar 2010 12:55:30 +0100 (MET)
Mime-Version: 1.0 (Apple Message framework v1077)
Content-Type: text/plain; charset=us-ascii
From: Brian Trammell <trammell@tik.ee.ethz.ch>
In-Reply-To: <4B90E1D5.90506@net.in.tum.de>
Date: Fri, 5 Mar 2010 12:55:29 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <C3BB9847-7EF2-410E-AF8C-CDFB69661CE5@tik.ee.ethz.ch>
References: <3146E35B-9877-44AE-B884-17541E6ED5B8@tik.ee.ethz.ch> <4B8C0943.6070309@net.in.tum.de> <A89AA32C-086C-46B9-8B6F-66BBEA645602@tik.ee.ethz.ch> <4B90E1D5.90506@net.in.tum.de>
To: Gerhard Muenz <muenz@net.in.tum.de>
X-Mailer: Apple Mail (2.1077)
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] WGLC comments on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 05 Mar 2010 11:55:34 -0000

Hi, Gerhard,

In short, thanks: your proposed changes address all my concerns. As far =
as I'm concerned, this one's ready to go.

Details inline.

Cheers,

Brian

On Mar 5, 2010, at 11:49 AM, Gerhard Muenz wrote:

>=20
> Hi Brian,
>=20
> Thanks a lot for your feedback.
>=20
> Maybe you can have a quick look at my proposals, so that I can close =
these issues (or at least most of them).
>=20
>=20
> Brian Trammell wrote:
>> Hi, Gerhard, all,
>> On Mar 1, 2010, at 7:36 PM, Gerhard Muenz wrote:
>>> Hi Brian,
>>> Thanks a lot for your review.
>>> Data models in network management will never satisfy all people.
>>> So, please keep in mind that the goal of this document is to
>>> standardize a configuration data model that can be easily
>>> implemented by many vendors. There will always be additional
>>> vendor-specific parameters, and there will always be
>>> vendor-specific restrictions of the model. Thanks to YANG, there
>>> are possibilities to make restrictions and extension.
>>> So, ask yourself if the parameters which are contained in the
>>> configuration data model can be easily mapped to one of the YAF
>>> command line options. Not the other way round.
>> Yep, I understand the point of the data model here... and YAF's
>> configuration by its nature will most probably remain command-line
>> (indeed I was thinking of implementing this in a Ruby script that
>> kicked off YAF); the references to YAF herein were merely an attempt
>> to use one specific implementation (with which I am familiar) as a
>> tool for evaluating the flexibility and general applicability of the
>> configuration data model. "This is out of scope" is a perfectly
>> reasonable response. bue I wanted to make sure in reviewing that the
>> flexibility we can get out of yang is presented in the right way...
>> Indeed, in the selector->cache->ep chain, you can probably just model
>> intermediate processes (like YAF's partial reassembly) as being part
>> of the Cache, and that's just fine.
>=20
> In an ipfix-mediator-config draft, you can use YANG to extend the =
model to allow  CP->IP->EP  and  OP->SP->Cache->IP->EP  setups.
>=20
> Speaking in UML, we are not restricted to the existing classes, we =
extend the model by adding new ones.

Great. This is okay then.

>>> More comments inline.
>>> Gerhard
>>> Brian Trammell wrote:
>>>> Greetings, all, I have reviewed
>>>> draft-ietf-ipfix-configuration-model-04 in response to the WGLC,
>>>> and have the following comments, ranked in order of impact; most
>>>> of these deal with the applicability of the configuration data
>>>> model to implementations other than the "exemplar" =
implementation(s) the authors had in mind while while writing.  I
>>>> will say that overall, the draft does an more than reasonable
>>>> job of being generally applicable. I apologize for starting a new
>>>> thread, but I have not yet had time to properly review other
>>>> comments, but plan on being in Anaheim prepared to discuss those
>>>> too. I have also explicitly not reviewed the YANG definition; I
>>>> leave that to people with actual YANG expertise. First and
>>>> foremost is the issue of caches, as it is in cache implementation
>>>> that IPFIX packet-to-flow MPs will differentiate the most from
>>>> each other, and indeed most vendors should be expected to subclass =
this cache class for their own configuration parameters;
>>>> therefore getting this right is probably the most crucial.
>>>> cacheMode immediate seems like a bit of a hack, as it invalidates
>>>> basically every other cache field. However, I suppose I can see
>>>> it as a way to keep the class design clean (otherwise, there
>>>> would be nothing to connect OP to layout/EP in a PSAMP
>>>> implementation...).
>>> Right.
>>>> 1. I'm concerned about the definition of cacheMode timeout (4.3,
>>>> p. 24), as it seems to make it impossible to configure a cache to
>>>> allow cache exit on TCP state tracking. Perhaps this is merely an
>>>> editorial oversight?
>>> We can allow a timeout Cache to expire a flow after a certain flow
>>> termination event. Is this what you propose?
>> Basically, yes. Most flow generators will expire a flow after a TCP
>> FIN or RST (or some number of these), so there is a difference
>> between expiring flows only on timeout and expiring on timeout as
>> well as "natural" end of flow (see the definition of flowEndReason,
>> IE 136). So I would propose the following:
>> cacheMode:  Configures the Cache Mode.  The following parameter =
values are specified by the configuration data model: *  immediate:
>> Data Records expire after the first packet *  timeout: Data Records
>> expire after active or inactive timeout |    *  natural: Data Records
>> expire after active or inactive timeout, or |       on natural
>> termination (e.g. TCP FIN, or TCP RST) as determined |       by the
>> Metering Process. *  permanent: Data Records never expire, but are
>> periodically exported with interval set by exportInterval In the case
>> of "immediate", PSAMP Packet Reports are generated. Otherwise, IPFIX
>> Flow Records are generated.
>=20
> Ok. I change your text as follows (to avoid mentioning Metering =
Process here):
>=20
>  * natural: Data Records expire after active or inactive timeout, or =
on
>             natural termination (e.g. TCP FIN, or TCP RST) of the Flow
>=20
> We must be aware of the fact that there are reasons to prematurely =
expire flows (e.g., full cache). I assume that we do not have to mention =
these, do we?

You can't very well tell a Cache to go out and buy more storage for =
itself, now can you? So, no. :)

>> Alternately, if "timeout" meant "timeout + natural" in this
>> terminology, we could change it as follows:
>> cacheMode:  Configures the Cache Mode.  The following parameter =
values are specified by the configuration data model: *  immediate:
>> Data Records expire after the first packet |    *  timeout: Data
>> Records expire after active or inactive timeout, or |       on
>> natural termination (e.g. TCP FIN, or TCP RST) as determined |
>> by the Metering Process. *  permanent: Data Records never expire, but
>> are periodically exported with interval set by exportInterval In the
>> case of "immediate", PSAMP Packet Reports are generated. Otherwise,
>> IPFIX Flow Records are generated.
>> I'm inclined toward the first alternative, with 'natural' as default
>> and 'timeout' provided for MPs that don't want to be bothered with
>> TCP state.
>=20
> Currently, there is no default value. The user must configure the =
cacheMode parameter. I think, this is useful because it forces the user =
to think about what he wants to meter.

Yep, makes sense.

>>>> 2. Destination (4.4.1, pp .27-29) appears not to match the
>>>> properties of the particular transports:
>>>> a. can't specify a local IP address for TCP or UDP. This seems =
necessary on devices with multiple management interfaces, which I
>>>> don't think is all that weird a deployment plan.
>>> This will be fixed as proposed in this mail unless anybody objects:
>>> http://www.ietf.org/mail-archive/web/ipfix/current/msg05316.html
>> Perfect.
>>>> b. can't specify 1..n remote IP addresses for SCTP. This breaks
>>>> one of the advantages of SCTP.
>>> I see send_connectx() in the SCTP API: =
http://tools.ietf.org/html/draft-ietf-tsvwg-sctpsocket-21#section-8.9
>>> However, what happens if you configure multiple remote IP addresses
>>> and they do not belong to the same SCTP endpoint? What shall the
>>> Exporting Process do in this situation?
>> What sctp_connectx() does:
>> "The way the SCTP stack uses the list of addresses to set up the
>> association is implementation dependent.  This function only
>> specifies that the stack will try to make use of all the addresses in
>> the list when needed." The assumptions are 1. that the list of
>> addresses all correspond to a logically unitary CP and 2. that the
>> admin knows what he's doing.
>>> I'm not sure if it makes sense to configure multiple remote
>>> addresses for a single destination. It seems to be safer to
>>> configure multiple export destinations and to use export member
>>> type primary and secondary.
>>> What do you think?
>> These are logically separate concepts, and should not be conflated by
>> the data model. Multiple Destinations means multiple CPs: endpoint
>> redundancy, and the assumption is that each Destination is its own CP
>> with separate app-layer state. Multiple addresses means multiple
>> endpoints on a single CP for session setup: _path_ redundancy, and
>> the assumption is that every given address leads to the same CP, with
>> the same app-layer state. Different tools for different situations.
>> Of course you can configure to export to two addresses that aren't
>> the same CP, just like you can configure an EP to export to a
>> destination that _isn't_ a CP. In both cases this is a configuration
>> error.
>=20
> New multiplicity factor for destinationIPAddress:
>=20
>   +--------------------------------------------------+
>   | Destination                                      |
>   +--------------------------------------------------+
>   | name                                             |<>-----+
>   | exportMemberType =3D parallel                      |       | 0..1
>   | ipfixVersion =3D 10                                |       |
>   | transportProtocol                                |   =
+------------+
>   | sourceIpAddress[0..*] {[0..1] for UDP and TCP}   |   | Transport- =
|
>   | destinationIpAddress[1..*] {[1] for UDP and TCP} |   | Layer-     =
|
>   | destinationPort =3D 4739|4740                      |   | Security  =
 |
>   | ifName/ifIndex[0..1]                             |   =
+------------+
>   | sendBufferSize {opt.}                            |
>   | rateLimit[0..1]                                  |
>   | timedReliability =3D 0 {SCTP only}                 |
>   | numberOfStreams {opt.} {SCTP only}               |
>   | templateRefreshTimeout =3D 600 {UDP only}          |
>   | optionsTemplateRefreshTimeout =3D 600 {UDP only}   |
>   | templateRefreshPacket[0..1] {UDP only}           |
>   | optionsTemplateRefreshPacket[0..1] {UDP only}    |
>   +--------------------------------------------------+
>=20
>                       Figure 14: Destination class
>=20
> [...]
>=20
>   destinationIpAddress:  Destination IP address to which IPFIX =
Messages
>      are sent (i.e., the IP address of the Collector).  This parameter
>      MUST be configured by the user.
>      If the transport protocol is SCTP, the parameter MAY appear
>      multiple times to specify multiple destination IP addresses.  The
>      user MUST ensure that all configured IP addresses belong to the
>      same Collector.  The Exporting Process tries to establish an SCTP
>      association to any of the configured destination IP addresses.
>=20
>=20
> Is this ok?

Perfect.

>=20
>>>> c. can't configure SCTP reliability per-template, as opposed to
>>>> per destination. see 2.b.
>>> As discussed long ago, Templates are not configured. The are
>>> created by the Exporting Process depending on the Data Records to
>>> be exported.
>> I'm not convinced this choice is the right one (both the Cache and
>> the EP are valid places to make choices about what, precisely, will
>> be exported: see below the discussion on
>=20
> Some people think that it is more natural to configure what you want =
to measure, and not how the information is transported to the Collector.

MPs that need to do post-Cache modification can easily extend the model, =
so no problem.

>>> Therefore, there is no good place for a per-Template reliability
>>> parameter in the configuration data model. The only possibility
>>> would be to configure the reliability in the Cache, and this would
>>> be a very strange place to do it.
>>> I had discussed this issue with Benoit. The conclusion was that we
>>> have a per-Destination parameter. If people want to configure more,
>>> they can come up with an extension.
>>> If you have another convincing solution, please let me know.
>> Not really. But now we as a WG are saying logically inconsistent
>> things: I can't reconcile the "templates are applications and the
>> reliability settings are per-application; here is a way to estimate
>> loss across multiple templates to the same destination" view of sctp
>> per stream export and the "destination-level reliability" view
>> presented here. It is an inelegant contradiction.
>=20
> Surely, the parameter must not inhibit the utilization of the =
per-SCTP-stream extension.
>=20
> Therefore, I changed the parameter description as follows:
>=20
>   timedReliability:  Lifetime in milliseconds until an IPFIX Message
>      containing Data Sets only is "abandoned" due to the timed
>      reliability mechanism of PR-SCTP [RFC3758].  If this parameter is
>      set to zero, reliable SCTP transport is used for all Data =
Records.
>      Regardless of the value of this parameter, the Exporting Process
>      MAY use reliable SCTP transport for Data Sets associated with
>      Options Templates.
>=20
> Ok?

Ping Benoit on this one, but it looks good to me.

>> My personal opinion would be to provide template-level configuration
>> as an EP option in an extension, and to simply ignore this
>> Destination option (much the same way as I work as hard as I can to
>> ignore UDP ;) )
> >
>>>> 3. Options (4.4.3, p. 30) configuration is rather odd and highly =
restrictive, even to the point of being unusable. The biggest
>>>> problem is that RFC5473 common properties are something
>>>> completely different from the options specified in 5101 and 5476.
>>>> "Enable common properties", without some definition of which
>>>> fields to use as common properties and under what circumstances,
>>>> is not a useful directive.
>>> The idea is that the Exporting Process defines common properties
>>> depending on the Data Records to be exported. This can be done
>>> dynamically (depending on the value distribution) or using
>>> knowledge about the configuration (e.g., if a filter ensures that
>>> the accounted packets are from/to a specific endpoint, this can be
>>> encoded as common property).
>>> If you think that reducingRedundancy is not useful, I will remove
>>> from the list.
>> It's certainly useful to have a switch, but we must then be very
>> explicit about the fact that when you turn this switch on, it's up to
>> the EP to determine precisely what redundancy to reduce and how, that
>> this is implementation dependent, and that implementations are
>> encouraged to provide implementation-specific configuration
>> extensions to cover this case.
>=20
>   reducingRedundancy:  Enables the utilization of Options Templates to
>      reduce redundancy in the exported Data Records according to
>      [RFC5473].  The Exporting Process decides when to apply these
>      Options Templates.
>=20
> Ok?

Yep.

>>>> Secondarily, options specified in 5610 and 5655 are not covered;
>>>> do we really want to amend this document every time we specify
>>>> another WG-blessed options template?
>>> Apart from reducingRedundancy, all mentioned Options type appear in
>>> the protocol RFCs (RFC5101, RFC5476).
>> 5655 options don't make much sense to configure, as runtime
>> determination and implementation dependency probably dominate there.
>> However, type export as in 5610 seems like it would be a useful
>> configuration switch.
>=20
>   extendedTypeInformation:  Export of extended type information for
>      enterprise-specific Information Elements used in the exported
>      Templates [RFC5610].
>=20
> Ok?

Also yep.

>> Although this discussion reminds me of another issue I missed,
>> elaborated below.
>>> Adding more parameter values for optionsType can be easily achieved
>>> with YANG extensions because they are specified as identities (it
>>> would note be possible with an enum type).
>> Hm. Okay. Will we create then a "defined Options Template" registry
>> somewhere?
>=20
> We could think about this, yes.
> At the moment, they are spread over many, many documents.

This is probably a good point for a future "what you need to know to =
nicely extend the protocol" effort...

>>>> I'd suggest using an OptionsTemplate subclass of Template,
>>>> instead, and providing the definition of common (non-5473)
>>>> options templates in an appendix to the configuration data model.
>>> I do not understand your proposal. There is no Template class for
>>> configuration.
>> I missed this point somehow (perhaps in my assumption that Templates
>> indeed should be configurable, but I did not show up to that
>> discussion so I accept the decision of the rest of the WG)... but
>> yes, indeed, so, withdrawn.
>>>> 4. There does not seem to be any room for logical configurable =
processes between selection and cache, or cache and EP. The
>>>> example that comes to mind is perhaps implementation-specific
>>>> [I'm mentioning this because yaf does it]: fragmentation behavior
>>>> (reassemble, ignore, drop). How could I allow then for the
>>>> configuration of the fragment table (given that a selection
>>>> process can only reference a cache)?
>>> You can always extend the model. For example, if fragmentation
>>> reassembly can be regarded as part of the Cache, you can easily
>>> define your own new cacheType (again, identities and identityref in
>>> YANG).
>>> We also have implemented a lot of cool additional functionalities
>>> which we are going to configure by extending the model. It is not
>>> the purpose of this document to cover all of these.
>> Indeed. As long as you can collapse these into Cache (which as I said
>> above it seems you can), that's more than sufficient.
>>>> 5. CacheLayout (4.3.1, p. 25) does not appear to be able to
>>>> support adaptive reduced-length export (exporting, say, counters
>>>> with unsigned32 unless one of them actually requires unsigned64)
>>>> [yes, yaf does this too]. Though this could probably be solved
>>>> with a deviation, and an additional vendor-specific CacheField
>>>> attribute...
>>> Is reduced size encoding performed by the Cache or by the Exporting
>>> Process?
>> In YAF, there's no process separation, but from a functional block
>> separation this is done in the EP.
>>> If you implement a Cache in hardware, it is probably easier to
>>> always use the maximum length and to reduce it later in the
>>> Exporting Process. We assume that the Cache reserves a certain
>>> field length for all Data Records in the Cache.  (Variable-length
>>> fields are treated differently)
>> Indeed, I would expect that only caches written in dynamically typed
>> languages, where the language itself autopromotes storage on
>> overflow, would do variable-sized caching of fixed length elements,
>> if only because doing it in C is rather time expensive and difficult
>> for relatively little benefit. And in this case the variable-sized
>> caching is invisible to the programmer.
>>> The ieLength specifies the field length in the Cache. The length in
>>> the exported Data Record can be different. Note again: we are not
>>> configuring a Template, here. The exported Data Record may even
>>> contain a completely different set of fields, for example in the
>>> case of reducing redundancy.
>> Ah, okay, so it supports it, it just doesn't support configuring it
>> (which can be an implementation dependent extension in either the
>> cache, the EP, or both, which makes sense depending on how it's
>> implemented). This is okay.
>>> What about removing the following sentence to avoid any confusion:
>>> For Information Elements of integer and float type, the field
>>> length MAY be set to a smaller value than the standard length of
>>> the abstract data type if the rules of reduced size encoding are
>>> fulfilled (see [RFC5101], Section 6.2).
>> No reason to cut that, as it wasn't the point of confusion...
>=20
> I want to make this clear. What about the following new paragraph?
>=20
>   Note that the Data Records generated by the Cache MAY be modified by
>   the Exporting Process before sending them to a Collector.  For
>   example, the Exporting Process MAY shorten the length of a field
>   according to the rules of reduced size encoding [RFC5101].  The
>   Exporting Process MAY also export certain fields in a separate Data
>   Record as described in [RFC5476].

Okay, yes; that's better than what's there now.

>>>> 6. CacheLayout (4.3.1, p. 25) could do with an explicit example
>>>> to illustrate the "unless the corresponding IE is not applicable
>>>> or cannot be derived" case: I presume this is intended for
>>>> dual-stack IPv4/IPv6 measurement, RFC5103 biflows, and the like?
>>>> It does appear than this clause is sufficient for such cases, but
>>>> I had to read this a couple of times before I understood that's
>>>> what was meant.
>>> Right. I tried to improve this. Please look into my reply to
>>> Lothar's review: =
http://www.ietf.org/mail-archive/web/ipfix/current/msg05325.html
>> Perfect.
>>>> 7. Real File Readers and File Writers will probably need more
>>>> than a single filename endpoint on the file system (however, most
>>>> file readers and file writers in document-based workflows won't
>>>> be configured using this information model...) How to handle this
>>>> is probably implementation-specific; in any case, I don't have a
>>>> good suggestion here, yet; perhaps we can discuss in Anaheim.
>>> I assume that the URI filename parameter is useful for many
>>> FileWriter/Reader implementations.  All additional parameters are
>>> probably very specific to the implementation.
>> Indeed. Okay, makes sense to leave it then, maybe with a statement
>> that the handling of the URI is implementation dependent.
>=20
> For simple FileWriter/Readers, a single file is enough, and the =
parameter description is clear in this case.
>=20
> More extended FileWriter/Readers need to extend the model anyway, =
which means that they can also adapt the meaning of the file parameter.
>=20
> So, rather no further statement which may cause confusion.

Good. Makes sense.

>>>> 8. Is there any benefit to cascading caches as with selection =
processes, or rather, are the benefits of cascading caches
>>>> considered in scope for this application? The simplest example
>>>> would be simultaneous export of flow data as well as flow
>>>> aggregate data (e.g. traffic counters per AS pair), perhaps as
>>>> part of a cascading-aggregation mediator. I ask only because it
>>>> seems a rather small and easily supportable change to greatly
>>>> increase flexibility.
>>> Uff. I already have problems to convince people that there may be
>>> more than one parallel Caches in a Metering Process. Now, you even
>>> propose multiple Caches in a series :)
>>> IMO, aggregation is done in an Intermediate Process, not in a
>>> Metering Process and therefore not in the Cache of a Metering
>>> Process.
>>> The configuration of Mediators and Intermediate Process is out of
>>> scope of this document. Otherwise, we will never finish...
>> Accepted. But now I have an idea of at least one subsection heading
>> in the Mediator Configuration Data Model draft :)
>>>> 9. Destination exportMemberType appears to reference a renamed
>>>> MIB object as of MIB-10; this is called ipfixExportMemberType
>>>> now. I would suggest copying the complete description from MIB,
>>>> because the semantics are kind of confusing otherwise (I had to
>>>> read MIB, which I probably should have done by now but oh well,
>>>> in order to figure out what the difference between parallel and
>>>> primary was). A tertiary problem: these exportMemberTypes seem a
>>>> bit restrictive (e.g. what if I want to do parallel export and
>>>> also have secondaries?), but I suppose that's a MIB comment.
>>> Andrew suggested to improve this as well.  See: =
http://www.ietf.org/mail-archive/web/ipfix/current/msg05323.html I'm =
open to concrete suggestions.
>> Update the text in the configuration model to match that in the MIB,
>> copy the definitions in full from the MIB so you don't have to go
>> read the MIB to figure out what these mean, and call it finished. I
>> expect most people will either find the MIB approach sufficient, or
>> simply leave these parallel and implement their own redundancy.
>=20
> Ok, I will follow your proposal unless I find a better way to =
configure this stuff.
>=20
>>>> 10. draft-ietf-ipfix-file is now RFC5655 (the editor will catch =
this); since this is now an RFC, references to things "proposed"
>>>> in this draft should probably be changed to things "specified" in
>>>> RFC 5655.
>>> Done.
>> And, as promised, a new 11. Cache should have a way to configure
>> biflow assembly and export. The simplest one is simply to state that
>> the presence of reverse Information Elements in a CacheLayout implies
>> that the Cache will assemble biflows suitable for export as described
>> in RFC 5103.
>=20
> Please check the following:
>=20
>=20
>   ieId, ieName, ieEnterpriseNumber:  These parameters specify a field
>      by the combination of the Information Element identifier or name,
>      and the Information Element enterprise number.  Either ieId or
>      ieName MUST be specified.  ieEnterpriseNumber MUST be used for
>      enterprise-specific Information Elements.  If ieEnterpriseNumber
>      is omitted or zero, this is Information Element is not =
enterprise-
>      specific but registered at IANA.
>      If the enterprise number is set to 29305, this field contains a
>      Reverse Information Element.  In this case, the Cache MUST
>      generate Data Records in accordance to [RFC5103].
>=20
> [...]
>=20
>   isFlowKey:  If present, this field is a Flow Key. If the field
>      contains a Reverse Information Element, it MUST NOT be configured
>      as Flow Key.

Perfect.=20

>> This opens a slightly larger question, though: is this a
>> configuration data model for 5101 and 5102, for all the RFCs
>> published by the WG to date (i.e., adding 5103, 5473, 5610, and maybe
>> 5655), for all the work completed by the WG (adding per-stream), or
>> for all the work accepted aside from mediation (adding flow
>> selection, and possibly anonymisation in a non-mediator context)? I
>> think the last is almost certainly out of scope (otherwise we're
>> never done). Thoughts?
>=20
> For many of the RFC you mention above do not specify any configuration =
requirements. Often, there exist zero to one implementation. So, it is =
difficult to say whether the parameters required by these =
implementations can be assumed to be "common" configuration parameters.
>=20
> So, I assume, we need more implementations until we can define a =
complete configuration data model for them.

Point. Indeed, as the draft stands with your proposed edits, it covers =
the protocol as of all applicable RFCs that have options that can be =
configured: 5101, 5102, 5103, 5473, and 5610, so my question is answered =
de facto, I suppose.



From muenz@net.in.tum.de  Fri Mar  5 04:33:08 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 715073A816C for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 04:33:08 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id jG-4bN4FGN4v for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 04:33:06 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 92FC33A6816 for <ipfix@ietf.org>; Fri,  5 Mar 2010 04:33:06 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 8AD4D200A791; Fri,  5 Mar 2010 13:33:07 +0100 (CET)
Message-ID: <4B90FA0F.3030801@net.in.tum.de>
Date: Fri, 05 Mar 2010 13:33:19 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Lothar Braun <braun@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu>	<4AD793E9-3463-43F1-8840-13BB6FF0B1E6@net.in.tum.de> <33C32EC7-A6D6-4238-B376-3F91FA3B74D4@net.in.tum.de> <4B8BF243.7090303@net.in.tum.de> <D71955E5-D24F-4859-B4F0-1C080B8CF192@net.in.tum.de> <4B8E2DAD.2010308@net.in.tum.de> <A7039173-D573-42F5-90C0-75A43A9F204B@tik.ee.ethz.ch> <4B8FC6F3.9040004@net.in.tum.de> <1CEC4FEC-8C83-4A43-B7BA-F3C7F50A9CBC@net.in.tum.de>
In-Reply-To: <1CEC4FEC-8C83-4A43-B7BA-F3C7F50A9CBC@net.in.tum.de>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms030009070402050703030909"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] WG last call on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 05 Mar 2010 12:33:08 -0000

This is a cryptographically signed message in MIME format.

--------------ms030009070402050703030909
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Lothar, Brian, all,

Please read my comments inline.

Lothar Braun wrote:
> Hi Gerhard,
>=20
> On Mar 4, 2010, at 3:42 PM, Gerhard Muenz wrote:
>> Brian Trammell wrote:
>>> hi Gerhard, all,
>>> Short, short answers inline below. A much longer message follows.
>>> Cheers,
>>> Brian
>>> On Mar 3, 2010, at 10:36 AM, Gerhard Muenz wrote:
>>>> Hi Lothar,
>>>>
>>>> Ok, we have an open issue regarding the Cache Layout, which is impor=
tant.
>>>>
>>>> So, other IPFIX people are encouraged to also read this e-mail and c=
omment :-)
>>>>
>>>> It's about section 4.3.1, which looks like this after some revisions=
:
>>>>
>>>>> 4.3.1.  CacheLayout Class
>>>>>           +--------------+
>>>>>           | CacheLayout  |
>>>>>           +--------------+   1..* +--------------------------+
>>>>>           |              |<>------| CacheField               |
>>>>>           |              |        +--------------------------+
>>>>>           |              |        | name                     |
>>>>>           |              |        | ieId/ieName              |
>>>>>           |              |        | ieLength {opt.}          |
>>>>>           |              |        | ieEnterpriseNumber[0..1] |
>>>>>           |              |        | isFlowKey[0..1]          |
>>>>>           +--------------+        +--------------------------+
>>>>>                      Figure 12: CacheLayout class
>>>>>  A Cache generates and maintains Packet Reports or Flow Records
>>>>>  containing information that has been extracted from the incoming
>>>>>  stream of packets.  Using the CacheField class, the CacheLayout cl=
ass
>>>>>  specifies the superset of fields that are included in the Packet
>>>>>  Reports or Flow Records (see Figure 12).
>>>>>  If Packet Reports are generated (i.e., if Cache Mode is "immediate=
"),
>>>>>  every field specified by the Cache Layout MUST be included in the
>>>>>  resulting Packet Report unless the corresponding Information Eleme=
nt
>>>>>  is not applicable or cannot be derived from the content or treatme=
nt
>>>>>  of the incoming packet.
>>>>>  For example, if a field is configured to contain the TCP source po=
rt
>>>>>  (Information Element tcpSourcePort [RFC5102]), the field MUST be
>>>>>  included in all Packet Reports which are related to TCP packets.
>>>>>  Obviously, the field value cannot be determined for non-TCP packet=
s.
>>>>>  If Flow Records are generated (i.e., if Cache Mode is "timeout" or=

>>>>>  "permanent"), every Flow Key field specified by the Cache Layout M=
UST
>>>>>  be included as Flow Key in the resulting Flow Record unless the
>>>>>  corresponding Information Element is not applicable or cannot be
>>>>>  derived from the content or treatment of the incoming packet.  Two=

>>>>>  packets MUST be accounted by different Flow Records if they differ=
 in
>>>>>  at least one Flow Key field.  Hence, if a Flow Key field can be
>>>>>  determined for one packet but not for the other, the two packets a=
re
>>>>>  accounted in different Flow Records.
>>>>>  Every non-key field specified by the Cache Layout MUST be included=
 in
>>>>>  the resulting Flow Record unless the corresponding Information
>>>>>  Element is not applicable or cannot be derived for the given Flow.=

>>>>>  Two packets are accounted by the same Flow Record if they only dif=
fer
>>>>>  in non-key fields but not in any of the Flow Key fields.  The same=

>>>>>  applies if one or more non-key fields can be determined for one
>>>>>  packet but not for the other.
>>>>>  For example, if a non-key field specifies an Information Element
>>>>>  whose value is determined by the first packet observed within a Fl=
ow
>>>>>  (which is the default rule according to [RFC5102]), this field MUS=
T
>>>>>  be included in the resulting Flow Record if it can be determined f=
rom
>>>>>  the first packet of the Flow.
>>>> First question:
>>>>
>>>> What happens if the Cache Layout contains an Information Element the=
 Monitoring Device cannot handle?
>>>>
>>>> In my opinion, the only possible answer is that the Monitoring Devic=
e returns an error. In general, the Monitoring Device MUST notify the use=
r about any part of the configuration which is not supported. It MUST NOT=
 silently accept a configuration which cannot be completely implemented.
>>>>
>>>> I propose to add these two sentences to the introduction. This will =
implicitly answer the question.
>>> Yes.
>> This text will appear in Section 5 which is about "Adaptation to Devic=
e Capabilities" and therefore should be the right place in the document:
>>
>>   The deviation concept enables a device to announce deviations from
>>   the standard model using the "deviation" statement.  For example, it=

>>   is possible to restrict the value range of a specific parameter or t=
o
>>   define that the configuration of a certain parameter is not supporte=
d
>>   at all.  Hence, deviations are typically used to specify limitations=

>>   due to resource constraints or functional restrictions.  Deviations
>>   concern existing parameters of the original configuration data model=

>>   and must not be confused with model extensions.  Model extension are=

>>   specified with the "augment" statement and allow adding new
>>   parameters to the original configuration data model.
>>
>>   If certain device-specific constraints cannot be formally specified
>>   with YANG, they MUST be expressed with human-readable text using the=

>>   "description" statement.  The provided information MUST enable the
>>   user to define a configuration which is entirely supported by the
>>   Monitoring Device.  On the other hand, if a Monitoring Device is
>>   configured, it MUST notify the user about any part of the
>>   configuration which is not supported.  The Monitoring Device MUST NO=
T
>>   silently accept configuration data which cannot be completely
>>   enforced.
>>
>>   Just like features, deviations and model extensions are announced in=

>>   NETCONF's <hello> message.  A usage example of deviations is given i=
n
>>   Section 7.5.
>=20
> It is unclear to me what happens when there is a configuration that can=
not be completely enforced.
>=20
> The device MUST inform the user that the configuration is not fully sup=
ported. But what about the configuration: Does the monitor have to apply =
the configuration anyways or must it reject the configuration with an err=
or? Or can the implementer decide on what to do?
>=20
> It is not clear to me if "MUST NOT silently accept" corresponds to "rej=
ect" or "protest against the configuration but accept it anyways".=20

In general, I do not want to specify any constraints regarding the=20
handling of configuration errors.

Anyway, if NETCONF is used, the error handling is specified by the error =

option [RFC4741]:

       error-option:

          The error-option element has one of the following values:

          stop-on-error: Abort the edit-config operation on first error.
             This is the default error-option.

          continue-on-error: Continue to process configuration data on
             error; error is recorded, and negative response is generated=

             if any errors occur.

          rollback-on-error: If an error condition occurs such that an
             error severity <rpc-error> element is generated, the server
             will stop processing the edit-config operation and restore
             the specified configuration to its complete state at the
             start of this edit-config operation.  This option requires
             the server to support the :rollback-on-error capability
             described in Section 8.5.

This option is sent with the configuration data (e.g., in an edit-config =

operation) and instructs the NETCONF agent what to do in case of a=20
configuration error.

So, I added one more sentence to the paragraph to refer to NETCONF error =

handling:

    If certain device-specific constraints cannot be formally specified
    with YANG, they MUST be expressed with human-readable text using the
    "description" statement.  The provided information MUST enable the
    user to define a configuration which is entirely supported by the
    Monitoring Device.  On the other hand, if a Monitoring Device is
    configured, it MUST notify the user about any part of the
    configuration which is not supported.  The Monitoring Device MUST NOT=

    silently accept configuration data which cannot be completely
    enforced.  If the NETCONF protocol is used to send configuration data=

    to the Monitoring Device, the error handling is specified in the
    NETCONF protocol specification [RFC4741].


Is this sufficient in your opinion?

>>>> Second question:
>>>>
>>>> What happens to a field in the Cache Layout which cannot be determin=
ed from an incoming packet?
>>>>
>>>> Possible answers:
>>>>
>>>> 1) It MUST NOT be present in the resulting Packet Report or Flow Rec=
ord.
>>>>
>>>> 2) It MAY still be present in the resulting Packet Report or Flow
>>>>  Record, but it MUST be obvious from the field value that the field
>>>>  was not be determined from the packet.
>>>>
>>>> 3) It MUST still be present in the resulting Packet Report or Flow
>>>>  Record and contain a default value which is configured by the user.=

>>>>
>>>> 4) We use a bit vector field which marks the fields with valid data.=

>>>>
>>>>
>>>> Currently, we do not specify this case, having in mind that 2) is al=
lowed. Example: transport port 0 is invalid and may be used for ICMP.
>>>>
>>>> 1) is a clean solution, but it may results in 2^n different Template=
s per Cache, n being the number of Flow Key fields. Implementing this is =
more complicated.
>>>>
>>>> 3) does not work because the Collector cannot distinguish between de=
fault values and real values.
>>>>
>>>> 4) is disliked by many people, e.g.:
>>>> http://www.ietf.org/mail-archive/web/ipfix/current/msg05275.html
>>>>
>>>>
>>>> So, I see 1) and 2) as the only feasible solutions.
>>>> Depending on the decision, I will add some text to the document.
>>> (2).
>> Fine.
>=20
> I would also prefer Option 2 but I have no idea if this is possible for=
 all fields.

No, it is not possible for all Information Elements because there is not =

always an invalid value.

For example, icmpTypeIPv4 =3D 0 is valid (Echo reply).

However, we might conclude that the icmpTypeIPv4 field is invalid if we=20
have transportProtocol !=3D 1 in the same record.

The description then would be more complicated:

2) It MAY still be present in the resulting Packet Report or Flow
    Record, but it MUST be obvious from the field value itself or from
    the values of other fields in same Data Record that the field
    was not be determined from the packet.

Opinions?

Gerhard

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms030009070402050703030909
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms030009070402050703030909--

From braun@net.in.tum.de  Fri Mar  5 04:45:05 2010
Return-Path: <braun@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 96AE23A89DC for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 04:45:05 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id uDCb8VVxDXOc for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 04:45:04 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id D44E73A8F62 for <ipfix@ietf.org>; Fri,  5 Mar 2010 04:45:03 -0800 (PST)
Received: from honshu.net.in.tum.de (honshu.net.in.tum.de [131.159.20.100]) by mail.net.in.tum.de (Postfix) with ESMTPSA id A7E50200AD9A; Fri,  5 Mar 2010 13:45:04 +0100 (CET)
Mime-Version: 1.0 (Apple Message framework v1077)
Content-Type: text/plain; charset=iso-8859-1
From: Lothar Braun <braun@net.in.tum.de>
In-Reply-To: <4B90FA0F.3030801@net.in.tum.de>
Date: Fri, 5 Mar 2010 13:45:04 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <87511595-B597-4E9A-8C34-8C58A1F93FDE@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu>	<4AD793E9-3463-43F1-8840-13BB6FF0B1E6@net.in.tum.de> <33C32EC7-A6D6-4238-B376-3F91FA3B74D4@net.in.tum.de> <4B8BF243.7090303@net.in.tum.de> <D71955E5-D24F-4859-B4F0-1C080B8CF192@net.in.tum.de> <4B8E2DAD.2010308@net.in.tum.de> <A7039173-D573-42F5-90C0-75A43A9F204B@tik.ee.ethz.ch> <4B8FC6F3.9040004@net.in.tum.de> <1CEC4FEC-8C83-4A43-B7BA-F3C7F50A9CBC@net.in.tum.de> <4B90FA0F.3030801@net.in.tum.de>
To: Gerhard Muenz <muenz@net.in.tum.de>
X-Mailer: Apple Mail (2.1077)
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] WG last call on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 05 Mar 2010 12:45:05 -0000

Hi,

On Mar 5, 2010, at 1:33 PM, Gerhard Muenz wrote:

>=20
> Hi Lothar, Brian, all,
>=20
> Please read my comments inline.
>=20
> Lothar Braun wrote:
>> Hi Gerhard,
>> On Mar 4, 2010, at 3:42 PM, Gerhard Muenz wrote:
>>> Brian Trammell wrote:
>>>> hi Gerhard, all,
>>>> Short, short answers inline below. A much longer message follows.
>>>> Cheers,
>>>> Brian
>>>> On Mar 3, 2010, at 10:36 AM, Gerhard Muenz wrote:
>>>>> Hi Lothar,
>>>>>=20
>>>>> Ok, we have an open issue regarding the Cache Layout, which is =
important.
>>>>>=20
>>>>> So, other IPFIX people are encouraged to also read this e-mail and =
comment :-)
>>>>>=20
>>>>> It's about section 4.3.1, which looks like this after some =
revisions:
>>>>>=20
>>>>>> 4.3.1.  CacheLayout Class
>>>>>>          +--------------+
>>>>>>          | CacheLayout  |
>>>>>>          +--------------+   1..* +--------------------------+
>>>>>>          |              |<>------| CacheField               |
>>>>>>          |              |        +--------------------------+
>>>>>>          |              |        | name                     |
>>>>>>          |              |        | ieId/ieName              |
>>>>>>          |              |        | ieLength {opt.}          |
>>>>>>          |              |        | ieEnterpriseNumber[0..1] |
>>>>>>          |              |        | isFlowKey[0..1]          |
>>>>>>          +--------------+        +--------------------------+
>>>>>>                     Figure 12: CacheLayout class
>>>>>> A Cache generates and maintains Packet Reports or Flow Records
>>>>>> containing information that has been extracted from the incoming
>>>>>> stream of packets.  Using the CacheField class, the CacheLayout =
class
>>>>>> specifies the superset of fields that are included in the Packet
>>>>>> Reports or Flow Records (see Figure 12).
>>>>>> If Packet Reports are generated (i.e., if Cache Mode is =
"immediate"),
>>>>>> every field specified by the Cache Layout MUST be included in the
>>>>>> resulting Packet Report unless the corresponding Information =
Element
>>>>>> is not applicable or cannot be derived from the content or =
treatment
>>>>>> of the incoming packet.
>>>>>> For example, if a field is configured to contain the TCP source =
port
>>>>>> (Information Element tcpSourcePort [RFC5102]), the field MUST be
>>>>>> included in all Packet Reports which are related to TCP packets.
>>>>>> Obviously, the field value cannot be determined for non-TCP =
packets.
>>>>>> If Flow Records are generated (i.e., if Cache Mode is "timeout" =
or
>>>>>> "permanent"), every Flow Key field specified by the Cache Layout =
MUST
>>>>>> be included as Flow Key in the resulting Flow Record unless the
>>>>>> corresponding Information Element is not applicable or cannot be
>>>>>> derived from the content or treatment of the incoming packet.  =
Two
>>>>>> packets MUST be accounted by different Flow Records if they =
differ in
>>>>>> at least one Flow Key field.  Hence, if a Flow Key field can be
>>>>>> determined for one packet but not for the other, the two packets =
are
>>>>>> accounted in different Flow Records.
>>>>>> Every non-key field specified by the Cache Layout MUST be =
included in
>>>>>> the resulting Flow Record unless the corresponding Information
>>>>>> Element is not applicable or cannot be derived for the given =
Flow.
>>>>>> Two packets are accounted by the same Flow Record if they only =
differ
>>>>>> in non-key fields but not in any of the Flow Key fields.  The =
same
>>>>>> applies if one or more non-key fields can be determined for one
>>>>>> packet but not for the other.
>>>>>> For example, if a non-key field specifies an Information Element
>>>>>> whose value is determined by the first packet observed within a =
Flow
>>>>>> (which is the default rule according to [RFC5102]), this field =
MUST
>>>>>> be included in the resulting Flow Record if it can be determined =
from
>>>>>> the first packet of the Flow.
>>>>> First question:
>>>>>=20
>>>>> What happens if the Cache Layout contains an Information Element =
the Monitoring Device cannot handle?
>>>>>=20
>>>>> In my opinion, the only possible answer is that the Monitoring =
Device returns an error. In general, the Monitoring Device MUST notify =
the user about any part of the configuration which is not supported. It =
MUST NOT silently accept a configuration which cannot be completely =
implemented.
>>>>>=20
>>>>> I propose to add these two sentences to the introduction. This =
will implicitly answer the question.
>>>> Yes.
>>> This text will appear in Section 5 which is about "Adaptation to =
Device Capabilities" and therefore should be the right place in the =
document:
>>>=20
>>>  The deviation concept enables a device to announce deviations from
>>>  the standard model using the "deviation" statement.  For example, =
it
>>>  is possible to restrict the value range of a specific parameter or =
to
>>>  define that the configuration of a certain parameter is not =
supported
>>>  at all.  Hence, deviations are typically used to specify =
limitations
>>>  due to resource constraints or functional restrictions.  Deviations
>>>  concern existing parameters of the original configuration data =
model
>>>  and must not be confused with model extensions.  Model extension =
are
>>>  specified with the "augment" statement and allow adding new
>>>  parameters to the original configuration data model.
>>>=20
>>>  If certain device-specific constraints cannot be formally specified
>>>  with YANG, they MUST be expressed with human-readable text using =
the
>>>  "description" statement.  The provided information MUST enable the
>>>  user to define a configuration which is entirely supported by the
>>>  Monitoring Device.  On the other hand, if a Monitoring Device is
>>>  configured, it MUST notify the user about any part of the
>>>  configuration which is not supported.  The Monitoring Device MUST =
NOT
>>>  silently accept configuration data which cannot be completely
>>>  enforced.
>>>=20
>>>  Just like features, deviations and model extensions are announced =
in
>>>  NETCONF's <hello> message.  A usage example of deviations is given =
in
>>>  Section 7.5.
>> It is unclear to me what happens when there is a configuration that =
cannot be completely enforced.
>> The device MUST inform the user that the configuration is not fully =
supported. But what about the configuration: Does the monitor have to =
apply the configuration anyways or must it reject the configuration with =
an error? Or can the implementer decide on what to do?
>> It is not clear to me if "MUST NOT silently accept" corresponds to =
"reject" or "protest against the configuration but accept it anyways".=20=

>=20
> In general, I do not want to specify any constraints regarding the =
handling of configuration errors.
>=20
> Anyway, if NETCONF is used, the error handling is specified by the =
error option [RFC4741]:
>=20
>      error-option:
>=20
>         The error-option element has one of the following values:
>=20
>         stop-on-error: Abort the edit-config operation on first error.
>            This is the default error-option.
>=20
>         continue-on-error: Continue to process configuration data on
>            error; error is recorded, and negative response is =
generated
>            if any errors occur.
>=20
>         rollback-on-error: If an error condition occurs such that an
>            error severity <rpc-error> element is generated, the server
>            will stop processing the edit-config operation and restore
>            the specified configuration to its complete state at the
>            start of this edit-config operation.  This option requires
>            the server to support the :rollback-on-error capability
>            described in Section 8.5.
>=20
> This option is sent with the configuration data (e.g., in an =
edit-config operation) and instructs the NETCONF agent what to do in =
case of a configuration error.
>=20
> So, I added one more sentence to the paragraph to refer to NETCONF =
error handling:
>=20
>   If certain device-specific constraints cannot be formally specified
>   with YANG, they MUST be expressed with human-readable text using the
>   "description" statement.  The provided information MUST enable the
>   user to define a configuration which is entirely supported by the
>   Monitoring Device.  On the other hand, if a Monitoring Device is
>   configured, it MUST notify the user about any part of the
>   configuration which is not supported.  The Monitoring Device MUST =
NOT
>   silently accept configuration data which cannot be completely
>   enforced.  If the NETCONF protocol is used to send configuration =
data
>   to the Monitoring Device, the error handling is specified in the
>   NETCONF protocol specification [RFC4741].
>=20
>=20
> Is this sufficient in your opinion?

Perfect.

>>>>> Second question:
>>>>>=20
>>>>> What happens to a field in the Cache Layout which cannot be =
determined from an incoming packet?
>>>>>=20
>>>>> Possible answers:
>>>>>=20
>>>>> 1) It MUST NOT be present in the resulting Packet Report or Flow =
Record.
>>>>>=20
>>>>> 2) It MAY still be present in the resulting Packet Report or Flow
>>>>> Record, but it MUST be obvious from the field value that the field
>>>>> was not be determined from the packet.
>>>>>=20
>>>>> 3) It MUST still be present in the resulting Packet Report or Flow
>>>>> Record and contain a default value which is configured by the =
user.
>>>>>=20
>>>>> 4) We use a bit vector field which marks the fields with valid =
data.
>>>>>=20
>>>>>=20
>>>>> Currently, we do not specify this case, having in mind that 2) is =
allowed. Example: transport port 0 is invalid and may be used for ICMP.
>>>>>=20
>>>>> 1) is a clean solution, but it may results in 2^n different =
Templates per Cache, n being the number of Flow Key fields. Implementing =
this is more complicated.
>>>>>=20
>>>>> 3) does not work because the Collector cannot distinguish between =
default values and real values.
>>>>>=20
>>>>> 4) is disliked by many people, e.g.:
>>>>> http://www.ietf.org/mail-archive/web/ipfix/current/msg05275.html
>>>>>=20
>>>>>=20
>>>>> So, I see 1) and 2) as the only feasible solutions.
>>>>> Depending on the decision, I will add some text to the document.
>>>> (2).
>>> Fine.
>> I would also prefer Option 2 but I have no idea if this is possible =
for all fields.
>=20
> No, it is not possible for all Information Elements because there is =
not always an invalid value.
>=20
> For example, icmpTypeIPv4 =3D 0 is valid (Echo reply).
>=20
> However, we might conclude that the icmpTypeIPv4 field is invalid if =
we have transportProtocol !=3D 1 in the same record.
>=20
> The description then would be more complicated:
>=20
> 2) It MAY still be present in the resulting Packet Report or Flow
>   Record, but it MUST be obvious from the field value itself or from
>   the values of other fields in same Data Record that the field
>   was not be determined from the packet.
>=20
> Opinions?

I think this text is good. Can you add examples for this two conditions?=20=


Like "value 0 as field value, if the TCP source port is extracted from =
an ICMP Message" for a scenario where the field value is invalid itself =
and "icmpTypeIpv4 =3D 0 and transportProtocol =3D TCP" for the other =
case.

Best regards,
  Lothar

--
Lothar Braun
Chair for Network Architectures and Services (I8)
Department of Informatics
Technische Universit=E4t M=FCnchen
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18010       Fax: +49 89 289-18033
E-mail: braun@net.in.tum.de=20







From muenz@net.in.tum.de  Fri Mar  5 05:32:06 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 194653A8B5C for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 05:32:06 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[AWL=0.000,  BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id cjNAPlpruLKT for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 05:32:04 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 281783A8B55 for <ipfix@ietf.org>; Fri,  5 Mar 2010 05:32:03 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id A1B20200AD9A; Fri,  5 Mar 2010 14:32:04 +0100 (CET)
Message-ID: <4B9107E1.1050308@net.in.tum.de>
Date: Fri, 05 Mar 2010 14:32:17 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: "ipfix@ietf.org" <ipfix@ietf.org>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms070903000101000002050405"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: Daniel Mentz <mentz@in.tum.de>
Subject: [IPFIX] IPFIX-CONFIG: maxPacketSize parameter for UDP export
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 05 Mar 2010 13:32:06 -0000

This is a cryptographically signed message in MIME format.

--------------ms070903000101000002050405
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Dear all,

I've missed a configuration issue mentioned in RFC5101:

10.3.3. MTU

    The maximum size of exported messages MUST be configured such that
    the total packet size does not exceed the path MTU.  If the path MTU
    is unknown, a maximum packet size of 512 octets SHOULD be used.


Although the text mentions the configuration of the IPFIX Message size,=20
I think that it makes more sense to configure the maximum packet size=20
(including IP and UDP header).

So, I propose to add a maxPacketSize parameter to the Destination class:


    +--------------------------------------------------+
    | Destination                                      |
    +--------------------------------------------------+
    | name                                             |<>-----+
    | exportMemberType =3D parallel                      |       | 0..1
    | ipfixVersion =3D 10                                |       |
    | transportProtocol                                |   +------------+=

    | sourceIpAddress[0..*] {[0..1] for UDP and TCP}   |   | Transport- |=

    | destinationIpAddress[1..*] {[1] for UDP and TCP} |   | Layer-     |=

    | destinationPort =3D 4739|4740                      |   | Security  =
 |
    | ifName/ifIndex[0..1]                             |   +------------+=

    | sendBufferSize {opt.}                            |
    | rateLimit[0..1]                                  |
    | timedReliability =3D 0 {SCTP only}                 |
    | numberOfStreams {opt.} {SCTP only}               |
=3D> | maxPacketSize[0..1] {UDP only}                   |
    | templateRefreshTimeout =3D 600 {UDP only}          |
    | optionsTemplateRefreshTimeout =3D 600 {UDP only}   |
    | templateRefreshPacket[0..1] {UDP only}           |
    | optionsTemplateRefreshPacket[0..1] {UDP only}    |
    +--------------------------------------------------+


    maxPacketSize:  This parameter allow the user to specify the maximum
       size of packets sent to the Collector.  If configured, the user
       MUST ensure that the parameter value does not exceed the path MTU.=

       If not configured, the Exporting Process MAY derive the maximum
       packet size from path MTU discovery mechanisms.  Otherwise, the
       packet size MUST not exceed 512 octets as specified in [RFC5101].


Please let me know if you like the parameter description.

I have noticed that path MTU discovery for UDP is not yet mentioned in=20
any IPFIX documents. Not sure, if it is ok to mention it here.

Gerhard


--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms070903000101000002050405
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms070903000101000002050405--

From muenz@net.in.tum.de  Fri Mar  5 06:52:32 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 0304528C102 for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 06:52:32 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 3bQ9oH4BaWfa for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 06:52:30 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 3E6923A8FB7 for <ipfix@ietf.org>; Fri,  5 Mar 2010 06:52:30 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id EC88C200A791; Fri,  5 Mar 2010 15:52:30 +0100 (CET)
Message-ID: <4B911ABA.1070002@net.in.tum.de>
Date: Fri, 05 Mar 2010 15:52:42 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Lothar Braun <braun@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu>	<4AD793E9-3463-43F1-8840-13BB6FF0B1E6@net.in.tum.de> <33C32EC7-A6D6-4238-B376-3F91FA3B74D4@net.in.tum.de> <4B8BF243.7090303@net.in.tum.de> <D71955E5-D24F-4859-B4F0-1C080B8CF192@net.in.tum.de> <4B8E2DAD.2010308@net.in.tum.de> <A7039173-D573-42F5-90C0-75A43A9F204B@tik.ee.ethz.ch> <4B8FC6F3.9040004@net.in.tum.de> <1CEC4FEC-8C83-4A43-B7BA-F3C7F50A9CBC@net.in.tum.de> <4B90FA0F.3030801@net.in.tum.de> <87511595-B597-4E9A-8C34-8C58A1F93FDE@net.in.tum.de>
In-Reply-To: <87511595-B597-4E9A-8C34-8C58A1F93FDE@net.in.tum.de>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms030504090504050201040702"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] WG last call on draft-ietf-ipfix-configuration-model-04
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 05 Mar 2010 14:52:32 -0000

This is a cryptographically signed message in MIME format.

--------------ms030504090504050201040702
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Lothar, Brian, all,

Still about the Cache Layout...

>>>>>> What happens to a field in the Cache Layout which cannot be determ=
ined from an incoming packet?
>>>>>>
>>>>>> Possible answers:
>>>>>>
>>>>>> 1) It MUST NOT be present in the resulting Packet Report or Flow R=
ecord.
>>>>>>
>>>>>> 2) It MAY still be present in the resulting Packet Report or Flow
>>>>>> Record, but it MUST be obvious from the field value that the field=

>>>>>> was not be determined from the packet.
>>>>>>
>>>>>> 3) It MUST still be present in the resulting Packet Report or Flow=

>>>>>> Record and contain a default value which is configured by the user=
=2E
>>>>>>
>>>>>> 4) We use a bit vector field which marks the fields with valid dat=
a.
>>>>>>
>>>>>>
>>>>>> Currently, we do not specify this case, having in mind that 2) is =
allowed. Example: transport port 0 is invalid and may be used for ICMP.
>>>>>>
>>>>>> 1) is a clean solution, but it may results in 2^n different Templa=
tes per Cache, n being the number of Flow Key fields. Implementing this i=
s more complicated.
>>>>>>
>>>>>> 3) does not work because the Collector cannot distinguish between =
default values and real values.
>>>>>>
>>>>>> 4) is disliked by many people, e.g.:
>>>>>> http://www.ietf.org/mail-archive/web/ipfix/current/msg05275.html
>>>>>>
>>>>>>
>>>>>> So, I see 1) and 2) as the only feasible solutions.
>>>>>> Depending on the decision, I will add some text to the document.
>>>>> (2).
>>>> Fine.
>>> I would also prefer Option 2 but I have no idea if this is possible f=
or all fields.
>> No, it is not possible for all Information Elements because there is n=
ot always an invalid value.
>>
>> For example, icmpTypeIPv4 =3D 0 is valid (Echo reply).
>>
>> However, we might conclude that the icmpTypeIPv4 field is invalid if w=
e have transportProtocol !=3D 1 in the same record.
>>
>> The description then would be more complicated:
>>
>> 2) It MAY still be present in the resulting Packet Report or Flow
>>   Record, but it MUST be obvious from the field value itself or from
>>   the values of other fields in same Data Record that the field
>>   was not be determined from the packet.
>>
>> Opinions?
>=20
> I think this text is good. Can you add examples for this two conditions=
?=20
>=20
> Like "value 0 as field value, if the TCP source port is extracted from =
an ICMP Message" for a scenario where the field value is invalid itself a=
nd "icmpTypeIpv4 =3D 0 and transportProtocol =3D TCP" for the other case.=


I've realized that packets with zero ports exist in reality:
http://isc.sans.org/port.html?port=3D0
So, this is a bad example for an invalid field value.
And I do not have a better example for invalid field values.

This is how the text looks like now (new parts underlined):

    If Packet Reports are generated (i.e., if Cache Mode is "immediate"),=

    every field specified by the Cache Layout MUST be included in the
    resulting Packet Report unless the corresponding Information Element
    is not applicable or cannot be derived from the content or treatment
    of the incoming packet.  _Any other field specified by the Cache
    Layout MAY only be included in the Packet Report if it is obvious
    from the field value itself or from the values of other fields in
    same Packet Report that the field value was not determined from the
    packet._

    For example, if a field is configured to contain the TCP source port
    (Information Element tcpSourcePort [RFC5102]), the field MUST be
    included in all Packet Reports which are related to TCP packets.
    _Although the field value cannot be determined for non-TCP packets,
    the field MAY be included in the Packet Reports if another field
    contains the transport protocol identifier (Information Element
    protocolIdentifier [RFC5102])._

    If Flow Records are generated (i.e., if Cache Mode is "timeout",
    "natural", or "permanent"), every Flow Key field specified by the
    Cache Layout MUST be included as Flow Key in the resulting Flow
    Record unless the corresponding Information Element is not applicable=

    or cannot be derived from the content or treatment of the incoming
    packet.  _Any other Flow Key field specified by the Cache Layout MAY
    only be included in the Flow Record if it is obvious from the field
    value itself or from the values of other Flow Key fields in same Flow=

    Record that the field value was not be determined from the packet._
    Two packets MUST be accounted by different Flow Records if they
    differ in at least one Flow Key field.  Hence, if a Flow Key field
    can be determined for one packet but not for the other, the two
    packets are accounted in different Flow Records.

    Every non-key field specified by the Cache Layout MUST be included in=

    the resulting Flow Record unless the corresponding Information
    Element is not applicable or cannot be derived for the given Flow.
    _Any other non-key field specified by the Cache Layout MAY only be
    included in the Flow Record if it is obvious from the field value
    itself or from the values of other fields in same Flow Record that
    the field value was not be determined from the packet._  Two packets
    are accounted by the same Flow Record if they only differ in non-key
    fields but not in any of the Flow Key fields.  The same applies if
    one or more non-key fields can be determined for one packet but not
    for the other.

    For example, if a non-key field specifies an Information Element
    whose value is determined by the first packet observed within a Flow
    (which is the default rule according to [RFC5102]), this field MUST
    be included in the resulting Flow Record if it can be determined from=

    the first packet of the Flow.

Opinions?

Regards,
Gerhard


--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms030504090504050201040702
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms030504090504050201040702--

From andrjohn@cisco.com  Fri Mar  5 14:15:38 2010
Return-Path: <andrjohn@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id B919E28C3BA for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 14:15:38 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -10.599
X-Spam-Level: 
X-Spam-Status: No, score=-10.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id bMa3iC1+1Abs for <ipfix@core3.amsl.com>; Fri,  5 Mar 2010 14:15:37 -0800 (PST)
Received: from ams-iport-1.cisco.com (ams-iport-1.cisco.com [144.254.224.140]) by core3.amsl.com (Postfix) with ESMTP id D33FD28C3A4 for <ipfix@ietf.org>; Fri,  5 Mar 2010 14:15:27 -0800 (PST)
Authentication-Results: ams-iport-1.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: As0AADMRkUuQ/uCWe2dsb2JhbACbSRUBAQsLJAYcnheYZIR3BA
X-IronPort-AV: E=Sophos;i="4.49,589,1262563200"; d="scan'208";a="57765738"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 05 Mar 2010 22:15:25 +0000
Received: from cisco.com (mrwint.cisco.com [64.103.71.48]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o25MFPcH013657; Fri, 5 Mar 2010 22:15:25 GMT
Received: from [10.55.163.37] (ams-andrjohn-8714.cisco.com [10.55.163.37]) by cisco.com (8.11.7p3+Sun/8.8.8) with ESMTP id o25MFOX00504; Fri, 5 Mar 2010 22:15:24 GMT
Message-ID: <4B918279.4040609@cisco.com>
Date: Fri, 05 Mar 2010 22:15:21 +0000
From: Andrew Johnson <andrjohn@cisco.com>
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de>
In-Reply-To: <4B8BD178.8030802@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 05 Mar 2010 22:15:38 -0000

Hi Gerhard

In this first part of my reply I'd like to address a specific point.

Gerhard Muenz wrote:
[SNIP]
>> I can see why it would be convenient to configure multiple caches to
>> be populated from a single Selection Process, and it could even be
>> implemented that way (to some extent), but that is not the PSAMP model.
>> The configuration model doesn't have to exactly match this model, but
>> you should ensure that the differences are made clear.
> 
> The PSAMP model has no notion of Caches.
> 
> Could you please show me the paragraphs in the PSAMP documents which 
> would indirectly prevent the existence of more than one Cache per 
> Metering Process?

I didn't say that you were PSAMP restricted you to one cache per 
Metering Process.  My point was about Selection Processes and my concern 
is over the statistics that PSAMP requires exported for each Selection 
Process.

For example, imagine a simple selection process configuration that 
selects 1 in 10 packets.  That configuration could first be used with 
Cache A:

     Selection Process A1:
       Observed packet stream count: 100
       Selected packet stream count:  10

     Cache A:
       Total packets in: 10


Now that same configuration can be used to populate another Cache later, 
but the statistics need kept separate for each Metering Process:

     Selection Process A1:
       Observed packet stream count: 130
       Selected packet stream count:  13

     Cache A:
       Total packets in: 10


     Selection Process B1:
       Observed packet stream count:  30
       Selected packet stream count:   3

     Cache B:
       Total packets in:  3


PSAMP requires that these stats are kept and exported, which means that 
the stats for A1 and B1 need to be exported under different Selection 
Process IDs.  I guess the real restriction is that a new cache cannot be 
populated from an existing Selection Process.  While a template 
configuration can be used it needs to spawn a separate instances so that 
the stats can kept separate.


Cheers

Andrew

From muenz@net.in.tum.de  Sat Mar  6 01:50:41 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 79E483A8F31 for <ipfix@core3.amsl.com>; Sat,  6 Mar 2010 01:50:41 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id NiIWg4jst3uF for <ipfix@core3.amsl.com>; Sat,  6 Mar 2010 01:50:40 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 237EA3A8F2E for <ipfix@ietf.org>; Sat,  6 Mar 2010 01:50:39 -0800 (PST)
Received: from [192.168.128.99] (ppp-88-217-27-28.dynamic.mnet-online.de [88.217.27.28]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 9C9B321CBD46; Sat,  6 Mar 2010 10:50:36 +0100 (CET)
Message-ID: <4B9225A5.5030709@net.in.tum.de>
Date: Sat, 06 Mar 2010 10:51:33 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Andrew Johnson <andrjohn@cisco.com>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com>
In-Reply-To: <4B918279.4040609@cisco.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 8bit
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sat, 06 Mar 2010 09:50:41 -0000

Hi Andrew,

Thanks for you reply.

Andrew Johnson wrote:
> Hi Gerhard
> 
> In this first part of my reply I'd like to address a specific point.
> 
> Gerhard Muenz wrote:
> [SNIP]
>>> I can see why it would be convenient to configure multiple caches to
>>> be populated from a single Selection Process, and it could even be
>>> implemented that way (to some extent), but that is not the PSAMP model.
>>> The configuration model doesn't have to exactly match this model, but
>>> you should ensure that the differences are made clear.
>>
>> The PSAMP model has no notion of Caches.
>>
>> Could you please show me the paragraphs in the PSAMP documents which
>> would indirectly prevent the existence of more than one Cache per
>> Metering Process?
> 
> I didn't say that you were PSAMP restricted you to one cache per
> Metering Process.  My point was about Selection Processes and my concern
> is over the statistics that PSAMP requires exported for each Selection
> Process.
>
> For example, imagine a simple selection process configuration that
> selects 1 in 10 packets.  That configuration could first be used with
> Cache A:
> 
>     Selection Process A1:
>       Observed packet stream count: 100
>       Selected packet stream count:  10
> 
>     Cache A:
>       Total packets in: 10
> 
> 
> Now that same configuration can be used to populate another Cache later,
> but the statistics need kept separate for each Metering Process:
> 
>     Selection Process A1:
>       Observed packet stream count: 130
>       Selected packet stream count:  13
> 
>     Cache A:
>       Total packets in: 10
> 
> 
>     Selection Process B1:
>       Observed packet stream count:  30
>       Selected packet stream count:   3
> 
>     Cache B:
>       Total packets in:  3
>
> PSAMP requires that these stats are kept and exported, which means that
> the stats for A1 and B1 need to be exported under different Selection
> Process IDs.  I guess the real restriction is that a new cache cannot be
> populated from an existing Selection Process.  While a template
> configuration can be used it needs to spawn a separate instances so that
> the stats can kept separate.

This is a good example. I see it differently, though.

In order to forward the output of the Selection Process to a second
Cache, you need to change the configuration of the Selection Process.
See how it looks in XML:

On Cache:

SP A1 -> Cache A

  <selectionProcess>
    <name>A1</name>
    <selector>
      ...
    </selector>
    <cache>A</cache>
  </selectionProcess>

After adding the second Cache:

       +-> Cache A
SP A1 -+
       +-> Cache B

  <selectionProcess>
    <name>A1</name>
    <selector>
      ...
    </selector>
    <cache>A</cache>
    <cache>B</cache>
  </selectionProcess>

So, switching from the first configuration to the second one involves a
reconfiguration of the Selection Process A1.

Regardless of how it is realized, the counters need to be reset because
the configuration of the SP has changed.

Before the configuration of the SP is changed, it makes sense to export
all old counters. Otherwise, they will be lost.

Probably, the cacheDataRecords counter of the Cache should also be reset
if the Cache is reconfigured.

Should I add a sentence which says that packetsObserved, packetsDropped,
cacheDataRecords need to be reset upon reconfiguration?

BTW: There are several ways to implement a switch from one configuration
to a new one. If your processes do not support reconfiguration at
runtime, you can replace them by newly instantiated processes. In this
case, the counters will be reset, anyway.

Cheers,
Gerhard

-- 
Dipl.-Ing. Gerhard Münz
Chair for Network Architectures and Services (I8)
Technische Universität München - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei München, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz

From andrjohn@cisco.com  Sat Mar  6 05:25:46 2010
Return-Path: <andrjohn@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 40A1E3A87A5 for <ipfix@core3.amsl.com>; Sat,  6 Mar 2010 05:25:46 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.599
X-Spam-Level: 
X-Spam-Status: No, score=-6.599 tagged_above=-999 required=5 tests=[AWL=-4.000, BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id GAoRLjhjlKqs for <ipfix@core3.amsl.com>; Sat,  6 Mar 2010 05:25:43 -0800 (PST)
Received: from ams-iport-2.cisco.com (ams-iport-2.cisco.com [144.254.224.141]) by core3.amsl.com (Postfix) with ESMTP id 9E0043A8307 for <ipfix@ietf.org>; Sat,  6 Mar 2010 05:25:42 -0800 (PST)
Authentication-Results: ams-iport-2.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AsEAAD/mkUuQ/uCWe2dsb2JhbACbSxUBAQsLJAYcnRaYHYR4BA
X-IronPort-AV: E=Sophos;i="4.49,594,1262563200";  d="scan'208";a="4051857"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-2.cisco.com with ESMTP; 06 Mar 2010 12:52:51 +0000
Received: from cisco.com (mrwint.cisco.com [64.103.71.48]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o26DPi4q020679; Sat, 6 Mar 2010 13:25:44 GMT
Received: from [10.55.163.37] (ams-andrjohn-8714.cisco.com [10.55.163.37]) by cisco.com (8.11.7p3+Sun/8.8.8) with ESMTP id o26DPgX04131; Sat, 6 Mar 2010 13:25:42 GMT
Message-ID: <4B9257CE.6040409@cisco.com>
Date: Sat, 06 Mar 2010 13:25:34 +0000
From: Andrew Johnson <andrjohn@cisco.com>
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de>
In-Reply-To: <4B8BD178.8030802@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sat, 06 Mar 2010 13:25:46 -0000

Hi Gerhard


Gerhard Muenz wrote:
> 
> Hi Andrew,
> 
> Thanks a lot for your thorough review.
> 
> Please find answers and comments inline.
> 
> Regards,
> Gerhard
> 
> 
> Andrew Johnson wrote:
>> One general comment:
>>
>>   Why do only some of the UML fields have settings given>
>>   e.g. "direction = both"
> 
> The meaning of "= X" is explained in Section 3.1:
> 
>    A default value is indicated after the attribute behind the equality
>    symbol ("=").  It means that this default value MUST be used by the
>    Monitoring Device if the parameter is not configured by the user.
> 
> However, nothing is written about what happens if "= X" is not present.
> 
> So, I have rewritten the first paragraphs of the section to clarify the 
> meaning of default values, multiplicity indicators, and attribute 
> properties, as well as their mapping to YANG
> 
> Please let me know if the text is ok and answers your question.
> 
> NEW:
> 
> 3.1.  UML Representation
> 
>    We use UML class diagrams [UML] to explain the structure of the
>    configuration data model.  The attributes of the classes are the
>    configuration or state parameters of the Monitoring Device.
> 
> 
>     +------------------------------------------------+
>     | Destination                                    |
>     +------------------------------------------------+
>     | name                                           |<>-----+
>     | exportMemberType = parallel                    |       | 0..1
>     | ipfixVersion = 10                              |       |
>     | transportProtocol                              |   +------------+
>     | sourceIpAddress[0..*] {[0..1] for UDP and TCP} |   | Transport- |
>     | destinationIpAddress                           |   | Layer-     |
>     | destinationPort = 4739|4740                    |   | Security   |
>     | ifName/ifIndex[0..1]                           |   +------------+
>     | sendBufferSize {opt.}                          |
>     | rateLimit[0..1]                                |
>     | timedReliability = 0 {SCTP only}               |
>     | numberOfStreams {opt.} {SCTP only}             |
>     | templateRefreshTimeout = 600 {UDP only}        |
>     | optionsTemplateRefreshTimeout = 600 {UDP only} |
>     | templateRefreshPacket[0..1] {UDP only}         |
>     | optionsTemplateRefreshPacket[0..1] {UDP only}  |
>     +------------------------------------------------+
> 
>                  Figure 2: UML example: Destination class
> 
>    As an example, Figure 2 shows the UML diagram of the Destination
>    class, which is specified in more detail in Section 4.4.1.  The upper
>    box contains the name of the class.  The lower box lists the
>    attributes of the class.  Each attribute corresponds to a parameter
>    of the configuration data model.
> 
>    Behind an attribute's name, there may appear a multiplicity indicator
>    in brackets (i.e., between "[" and "]").  An attribute with a
>    multiplicity indicator of "[0..1]" represents an OPTIONAL
>    configuration parameter which is only included in the configuration
>    data if the user configures it.  Typically, the absence of an
>    OPTIONAL parameter has a specific meaning.  For example, not
>    configuring rateLimit in the Destination class means that no rate
>    limiting will be applied to the exported data.  In YANG, an OPTIONAL
>    parameter is specified as a "leaf" without "mandatory true"
>    substatement.  The "description" substatement specifies the behavior
>    for the case that the parameter is not configured.
> 
>    A multiplicity indicator of "[0..*]" means that this parameter MAY be
>    configured multiple times with different values.  In the example,
>    multiple source IP addresses may be configured for a multi-homed
>    Exporting Process if SCTP is transport protocol.  In YANG, an
>    attribute with multiplicity indicator "[0..*]" corresponds to a
>    "leaf-list".
> 
>    Attributes without multiplicity indicator may be endued with a
>    default value which is indicated behind the equality symbol ("=").
>    If a default value exists, the parameter does not have to be
>    explicitly configured by the user.  If the parameter is not
>    configured by the user, the Monitoring Device MUST use the given
>    default value for the given parameter.  In the example, IPFIX version
>    10 must be used unless a different value is configured for
>    ipfixVersion.  In YANG, an attribute with default value corresponds
>    to a "leaf" with "default" substatement.
> 
>    In the example, there exist two default values for the destination
>    port, namely the registered ports for IPFIX with and without
>    transport layer security (i.e., DTLS or TLS), which are 4740 and
>    4739, respectively.  In the UML diagram, the two default values are
>    separated by a vertical bar ("|").  In YANG, such alternative default
>    values cannot be specified formally.  Instead, they are defined in
>    the "description" substatement of the "leaf".
> 
>    Further attribute properties are denoted in braces (i.e., between "{"
>    and "}").  An attribute with property "{opt.}", such as
>    sendBufferSize in the Destination class, represents a parameter that
>    MAY be configured by the user.  If not configured by the user, the
>    Monitoring Device MUST assign an appropriate value for this
>    parameter.  As a result, the parameter will always exist in the
>    configuration data, yet it is not mandatory for the user to configure
>    it.  In YANG, this behavior cannot be specified formally.  Therefore,
>    it is specified in the "description" substatement of the "leaf".
> 
>    The availability of a parameter may depend on another parameter
>    value.  In the UML diagram, such restrictions are indicated as
>    attribute properties.  In the example, the parameter timedReliability
>    can only be configured if the transport protocol is SCTP ("{SCTP
>    only}").  In YANG, such restrictions are formalized in the "when"
>    substatement of the "leaf".
> 
>    If the multiplicity of a parameter changes depending on another
>    parameter value, this is also denoted as an attribute property.  In
>    the example, the parameter sourceIpAddress may be configured only
>    once if the transport protocol is UDP or TCP ("{[0..1] for UDP and
>    TCP}").  In YANG, this is formalized in the "must" substatement of
>    the "leaf".
> 
>    Another attribute property not shown in the example is "{readOnly}"
>    specifying state parameters which cannot be configured.  In YANG,
>    this corresponds to the "config false" substatement.
> 
>    Attributes without multiplicity indicator, without default value, and
>    without "{readOnly}" property are mandatory configuration parameters.
>    These parameters MUST be configured by the user unless an attribute
>    property determines that the parameter is not available.  In YANG, a
>    mandatory parameter corresponds to a "leaf" with "mandatory true"
>    substatement.  In the example, the user MUST configure the parameters
>    name, transportProtocol, and destinationIpAddress.  The remaining
>    parameters have either a default value, the "{opt.}" property, or a
>    multiplicity indicator starting with zero.

I think this looks good.


>> ========================================================================
>> Section 3:
>> ========================================================================
>> If a Selection Process is part of a Metering Process and a single
>> Selection Process can feed into more than one Cache, then it is
>> implied that a single Metering Process can have more than once cache.
>
> Note: The text has slightly changed after Atsushi's review. Now, it 
> explicitly says that a Metering Process may contain multiple Selection 
> Processes and multiple Caches.

If a Metering Process can have multiple caches and an IPFIX device can 
have multiple Metering Processes, how do we associate the caches with 
Metering Processes?  Can this configuration system only be used to 
configure a single Metering Process within a device?


> Furthermore, it says that only packets observed at a single Observation 
> Point may enter a Selection Process. IMO, this results from the PSAMP 
> definitions of Observed Packet Stream and Selection Process. Please tell 
> me if this conclusion is wrong.

A Selection process should only have a single feed, but can have 
multiple observers.  This results in the number of possible chains 
(Selection Sequences) being equal to the number of Selectors that 
directly feed a cache.


> This is how the text reads now:
> 
> 
>                +-----------------------------------+
>                | Metering Process                  |
>                | +-----------+ Selected            |
>       Observed | | Selection | Packet    +-------+ |  Stream of
>       Packet  -->| Process   |---------->| Cache |--> Flow Records or
>       Stream   | +-----------+ Stream    +-------+ |  Packet Reports
>                +-----------------------------------+
> 
>      Figure 1: Selection Process and Cache forming a Metering Process
> 
>    The configuration data model adopts the separation of Selection
>    Processes and Caches in order to support the flexible configuration
>    and combination of these functional blocks.  As defined in [RFC5476],
>    the Selection Process takes the Observed Packet Stream as its input
>    and selects a subset of that stream as its output (Selected Packet
>    Stream).  The Observed Packet Stream at the input of a Selection
>    Process is composed of packets from a single Observation Point.  The
>    action of the Selection Process on a single packet of its input is
>    defined by a Primitive Selector or a Composite Selector.  The Cache
>    generates Flow Records or Packet Reports from the Selected Packet
>    Stream, depending on the configured Cache Mode.
> 
>    Beyond the basic Metering Process shown in Figure 1, the
>    configuration data model enables the specification of more complex
>    configurations.  A Metering Process may be composed of multiple
>    Selection Processes and multiple Caches.  A single Cache may process
>    the output of multiple Selection Processes.  The Selected Packet
>    Stream may be copied to enter several Caches, for example one Cache
>    that generates Flow Records and another Cache that generates Packet
>    Reports.  Selection Processes can also be cascaded, such that the
>    Selected Packet Stream at the output of one Selection Process is
>    passed to another Selection Process.  Cascading Selection Processes
>    can be useful in specific contexts such as the example in
>    Section 7.3.
> 
>> I can see why it would be convenient to configure multiple caches to
>> be populated from a single Selection Process, and it could even be
>> implemented that way (to some extent), but that is not the PSAMP model.
>> The configuration model doesn't have to exactly match this model, but
>> you should ensure that the differences are made clear.
> 
> The PSAMP model has no notion of Caches.
> 
> Could you please show me the paragraphs in the PSAMP documents which 
> would indirectly prevent the existence of more than one Cache per 
> Metering Process?

See other mail.


>> ========================================================================
>> From section 3.1:
>> ========================================================================
>> Why have so many SCTP or UDP only options?  Can't you use UML
>> inheritance to do something like:
>>
>>     +------------------------------------------------+
>>     | Destination                                    |
>>     +------------------------------------------------+
>>     | name                                           |
>>     | exportMemberType = parallel                    |
>>     | ipfixVersion = 10                              |
>>     | transportProtocol                              |
>>     | destinationIpAddress                           |
>>     | destinationPort = 4739|4740                    |
>>     | sendBufferSize {opt.}                          |
>>     | rateLimit[0..1]                                |
>>     +------------------------------------------------+
>>           ^
>>           | 1
>>      +----+-------------------+
>>      |                        |
>>      |    +------------------------------------------------+
>>      |    | SCTP Destination                               |
>>      |    +------------------------------------------------+
>>      |    | localIpAddress[0..*] {SCTP only}               
>> |<>-----+                |    | timedReliability = 0 {SCTP 
>> only}               |       | 0..1         |    | numberOfStreams 
>> {opt.} {SCTP only}             |       |                |    | 
>> sourceIpAddress[0..1] {UDP only}               |   +------------+
>>      |    | templateRefreshTimeout = 600 {UDP only}        |   | 
>> Transport- |
>>      |    | optionsTemplateRefreshTimeout = 600 {UDP only} |   | 
>> Layer-     |
>>      |    | templateRefreshPacket[0..1] {UDP only}         |   | 
>> Security   |
>>      |    | optionsTemplateRefreshPacket[0..1] {UDP only}  |   
>> +------------+
>>      |    +------------------------------------------------+
>>      |
>>      +---------------------+
>>                            |
>>     +------------------------------------------------+
>>     | UDP Destination                                |
>>     +------------------------------------------------+
>>     | sourceIpAddress[0..1] {UDP only}               
>> |<>-----+               | templateRefreshTimeout = 600 {UDP 
>> only}        |       | 0..1        | optionsTemplateRefreshTimeout = 
>> 600 {UDP only} |       |               | templateRefreshPacket[0..1] 
>> {UDP only}         |   +------------+
>>     | optionsTemplateRefreshPacket[0..1] {UDP only}  |   | Transport- |
>>     +------------------------------------------------+   | Layer-     |
>>                                    | Security   |
>>                                                          +------------+
> 
> This looks nice. Nevertheless, there are some drawbacks:
> 
> Since YANG does not support inheritance, it cannot be mapped to YANG 
> directly.

I'm not that familiar with YANG, perhaps it deals with the situation 
more gracefully than the UML diagram shows.  Can you point the 
Destination to one of a choice of protocol specific objects?


> Furthermore, I would have to explain even more UML symbols and stuff in 
> the document.
> 
> Note that the Destination class might be changed as proposed in
> http://www.ietf.org/mail-archive/web/ipfix/current/msg05316.html
> 
>> Are there no options that are used with TCP?
> 
> You mean parameters which are available for TCP only?
> I do not know any. If you have some, please let me know.
> 
>> Are the security parameters the same across DTLS and TLS?
> 
> There was some discussion on the mailing list whether we should add any 0
> (D)TLS parameters to the IPFIX configuration data model.
> 
> The conclusion was that only parameters regarding the mutual 
> authentication of Exporters and Collectors should be added. Further 
> parameters are too specific.
> 
> Regarding authentication, the parameters are the same for DTLS and TLS.

OK.

>> ========================================================================
>> From section 4.1:
>> ========================================================================
>> Doesn't the direction belong in the Observation Point class, not the
>> Interface or LC class.
> 
> If the current position of the direction parameter bothers you, we can 
> move it.
> 
> However, my reasoning is the following:
> - If it appears in the ObservationPoint class, it looks like the 
> Observation Point is doing some filtering.
> - If it appears in the Interface/Linecard class, it looks like the 
> filtering is already done at the interface or linecard.
> 
> I thought that the interface/linecard is doing the filtering.

I don't see it as filtering, just part of the definition of the 
Observation Point.  The Interface is most likely able to send and 
receive traffic, but the observation point could be on only the send or 
receive channel.  i.e. It's not a input only interface, but the 
Observation Point could only see input packets.

Currently you have a 1:1 mapping of OP to interface.  If the direction 
is moved to the Observation Point then you could have two observation 
points on the same interface, one on the input side and one on the output.


> There is another small advantage: The current position of the parameter 
> makes it easier to add vendor or implementation-specific packet sources 
> for which the direction parameter has no meaning, for example:
> 
> +------------------+
> | PcapFile         |
> +------------------+
> | filename         |
> +------------------+
> 
>> ========================================================================
>> From Section 4.2.1
>> ========================================================================
>> Why packetsDropped and not packetsSelected, when PSAMP talks about
>> packetsSelected in the stats it needs to export?
> 
> 
> All state parameters are adopted from the IPFIX-MIB. There, we have:
> 
> +--ipfixSelectionProcessStatsTable(4)
>    |
>    +--ipfixSelectionProcessStatsEntry(1) 
> [ipfixMeteringProcessCacheId,ipfixSelectionProcessIndex,ipfixSelectionProcessSelectorIndex] 
> 
>       |
>       +-- r-n Counter64 ipfixSelectionProcessStatsPacketsObserved(1)
>       +-- r-n Counter64 ipfixSelectionProcessStatsPacketsDropped(2)
>       +-- r-n TimeStamp ipfixSelectionProcessStatsDiscontinuityTime(3)
> 
> 
> Probably, ipfixSelectionProcessStatsPacketsSelected would make more 
> sense in the MIB.
> 
> I assume that the MIB standardization cannot be stopped any more, right?
> 
> So, what do you suggest?

I don't know.  Why follow the MIB exactly?  Any idea about the history 
of this choice or dropped rather than selected?


>> ========================================================================
>> From Section 4.3.1
>> ========================================================================
>> Isn't is possible to generate a cache without any keys?
>> This would generate a cache with a single flow?  Do we allow this?
> 
> A the moment, there is no restriction which says that a flow key field 
> must exist.
> 
> Do you want to add such a restriction?

A cache without keys will result in a single flow, which seems a bit 
daft.  I guess there is no pressing reason to add this restriction though.


>> Would it be useful to have two lists, one for keys and one for
>> non-keys?
> 
> There is an argument for having a single list:
> 
> A Cache implementation may generate records using the specified field 
> order. So, the fields occur in the same order as specified in the 
> configuration.

If we're worried about implementation specifics can't they extend the 
YANG model?


> With separate lists for flow key fields and non-key fields, we cannot 
> arbitrarily mix these two types of fields.
> 
> Is there any advantage of having separate lists?

I was just thinking about how you'd specify that at least one key was 
required.


>> For a permanent cache, update counters should be used and for a
>> timeout cache the delta counters should be used.  Where is this
>> sort of thing enforced?
> 
> This is not enforced. I would not strictly enforce it because strange 
> configurations might make sense in specific use cases.

OK.

> However, we could add a note that totalCounters are typically used for 
> "permanent" and deltaCounters for "timeout".
> 
> Could you provide a sentence or two?

Perhaps a more general note that not all fields will be appropriate for 
all configurations.  Is the Exporting Device free to change the fields 
if it needs total counters for a permanent cache?


>> ========================================================================
>> From Section 4.4.1
>> ========================================================================
>>    exportMemberType:  Configures the export member type that corresponds
>>       to the ipfixTransportSessionGroupMemberType object in
>>       [I-D.ietf-ipfix-mib].  The following parameter values are
>>       specified by the configuration data model:
>>       *  primary: primary target of the Exporting Process
>>       *  secondary: secondary target of the Exporting Process used when
>>          the primary target is not reachable
>>       *  parallel: parallel exporting to all destinations and files of
>>          the Exporting Process
>>       *  loadBalancing: load-balancing between different destinations
>>          and files of the Exporting Process
>>       "parallel" is the default value if this parameter is not
>>       configured.  If one destination or file is configured as "primary"
>>
>> ------------------------------------------------------------------------
>> Comments
>> ------------------------------------------------------------------------
>> It's not clear to me that this parameter belongs to the Destination
>> class.  The number of restrictions on when you can use the various
>> types suggests that we could simplify things.
>>
>> A Cache can pass flow records to multiple Exporting Processes, so we
>> don't really need a parallel option.  This would mean that each
>> Exporting Process is responsible for sending one complete set of
>> records.  i.e. one record in, one record out.  It could do that, either
>> using a redundancy model or with a load-balancing one.
>>
>> In other words, the exportType would be an Exporting Process attribute
>> which would indicate whether the list of Destinations should be load-
>> balanced across or treated as Primary, Seondary, Tertiary, etc.
> 
> At the moment, the parameter is inline with the IPFIX-MIB:
> 
> ipfixExportMemberType OBJECT-TYPE
>     SYNTAX      INTEGER {
>                     unknown(0),
>                     primary(1),
>                     secondary(2),
>                     parallel(3),
>                     loadBalancing(4)
>                 }
>     MAX-ACCESS  read-only
>     STATUS      current
>     DESCRIPTION
>         "The type of a member Transport Session in a Transport
>         Session group (identified by the value of ipfixExportIndex,
>         ipfixObservationDomainId and ipfixMeteringProcessCacheId).
>         The following values are valid:
> 
>         unknown(0)
>             This value MUST be used if the status of the group
>             membership cannot be detected by the equipment. This
>             value should be avoided as far as possible.
> 
>         primary(1)
>             This value is used for a group member that is used as
>             the primary target of an Exporter. Other group members
>             (with the same ipfixExportIndex and
>             ipfixMeteringProcessCacheId) MUST NOT have the value
>             primary(1) but MUST have the value secondary(2).
>             This value MUST also be specified if the Exporter does
>             not support Transport Session grouping. In this case
>             the group contains only one Transport Session.
> 
>         secondary(2)
>             This value is used for a group member that is used as a
>             secondary target of an Exporter. The Exporter will use
>             one of the targets specified as secondary(2) within the
>             same Transport Session group when the primary target is
>             not reachable.
> 
>         parallel(3)
>             This value is used for a group member that is used for
>             duplicate exporting i.e., all group members identified
>             by the ipfixExportIndex are exporting the same Records
>             in parallel. This implies that all group members MUST
>             have the same membertype parallel(3).
> 
>         loadBalancing(4)
>             This value is used for a group member that is used
>             as one target for load-balancing. This means that a
>             Record is sent to one of the group members in this
>             group identified by ipfixExportIndex.
>             This implies that all group members MUST have the same
>             membertype loadBalancing(4)."
>     ::= { ipfixExportEntry 2 }
> 
> 
> I'm open to better proposals.
> 
> "parallel" and "loadBalancing" can be configured once with a single 
> ExportingProcess class parameter.
> 
> How would you configure "primary" and "secondary with a single Exporting 
> Process parameter?
> 
> 
>> ========================================================================
>> From Section 4.7: TransportSession class
>> ========================================================================
>> I think you want to add a timestamp for when the session came up
>> and when it went down (unless it's still active).
> 
> Currently, there is no such parameter in ipfixTransportSessionTable.
> 
> Hm, it seems that we should have asked you to review the IPFIX-MIB :)
> 
> Shall we go beyond IPFIX-MIB and add such state parameters?

I don't see why not.  Is it possible to add to MIB after the initial 
definition?  Not necessarily now, but perhaps after it's been in use a 
while and certain additions are in demand.


>> There's no association with the relevant destination.  It seems to me
>> that your really want this as part of each Destination or FileReader.
>> Specifically, I think you want to separate the Transport Session from
>> the File Reader stats.  Potentially there could be a one to many
>> relationship of Destination and Transport Sessions, each with their own
>> timestamp for coming up and going down.
> 
> Currently, TransportSession is a mere copy of ipfixTransportSessionTable 
> which exists for both, Exporters and Collectors.
> 
> In the case of the _Exporting Process_ , we could move the state 
> parameters to the _Destination_ and _FileWriter_ class.
> 
> I think that there can be at most one Transport Session per Destination 
> and per FileWriter object. At least at the same point in time.
> However, it may happen that no Transport Session exists, for example in 
> the case of a secondary export destination.

Well let's imagine and Exporting process with a Primary Destination and 
a Secondary Destination.  The Primary Destination starts a Transport 
Session, which works for a while and then is lost.  The Secondary 
Destination then opens a Transport Session, which works fine.  The 
Exporting Process then manages to reconnect with Primary Destination and 
opens a new Transport Session with it and closes the Secondary 
Destination's Transport Session.

So, we have seen three Transport Sessions and if the Exporting Process 
kept a full history of the statistics it could show:

   Primary Destination (10.2.100.2 port 4739)
     - Transport Session (Open)
         Start Time: 10:01.19
         End Time: --:--.--
         Records sent: 5345
         ...
    -  Transport Session (Closed)
         Start Time: 08:00.00
         End Time: 09:45.14
         Records sent: 123210

   Secondary Destination (10.3.0.15 port 4739)
    - Transport Session (Closed)
        Start Time: 09:45.16
        End Time: 10:01.20
        Records sent: 14022

I guess it would be up to the implementation to control how much history 
was kept.


> So, we could move a single set of Transport Session state parameters to 
> the Destination and FileWriter classes. The parameters would appear or 
> not, depending on whether the Transport Session exists.

I think this makes sense, otherwise you have to match the Destination 
configuration to the Transport Session statistics by matching the 
various address and port field.


> In the case of the _Collecting Process_ , we can also move the Transport 
> Session state parameters to the _Receiver_ and _FileReader_ classes. As 
> you say, there can be multiple Transport Sessions per Receiver.

Well it seems that every field in the Destination Class is present in 
the Transport Session Class.  In fact, the Transport Session seems to be 
a copy of the Destination Class plus some statistics.  This is not true 
of the Receiver class.

Perhaps the Destination Class could be replaced by a Transport Session 
class that has some readonly stats and the Receiver Class can then 
contain multiple Transport Sessions.

In this case and Exporting Process wouldn't have a history of closed 
Transport Sessions, but would have information about the most recent one 
per Destination.  The above example would become:

   Primary Transport Session
     Destination IP address: 10.2.100.2
     Destination port:       4739
     Status:                 Open
     Start Time:             10:01.19
     End Time:               --:--.--
     Records sent:           5345
     ...

   Secondary Transport Session
     Destination IP address: 10.3.0.15
     Destination port:       4739
     Status:                 Closed
     Start Time:             09:45.16
     End Time:               10:01.20
     Records sent:           14022


> Probably, it makes sense to have a container 
> <transportSessionStatistics> which groups the state parameters of a 
> Transport Session. This container may appear multiple times in the case 
> of the Receiver class, and only once in all other classes.
> 
> Would you agree with these changes?

I think we need a little more discussion :)

Cheers

Andrew

From andrjohn@cisco.com  Sat Mar  6 12:26:18 2010
Return-Path: <andrjohn@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id CC58328C119 for <ipfix@core3.amsl.com>; Sat,  6 Mar 2010 12:26:18 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -8.599
X-Spam-Level: 
X-Spam-Status: No, score=-8.599 tagged_above=-999 required=5 tests=[AWL=2.000,  BAYES_00=-2.599, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id qrmtIc+pX5IR for <ipfix@core3.amsl.com>; Sat,  6 Mar 2010 12:26:17 -0800 (PST)
Received: from ams-iport-1.cisco.com (ams-iport-1.cisco.com [144.254.224.140]) by core3.amsl.com (Postfix) with ESMTP id 85A0A28C0F0 for <ipfix@ietf.org>; Sat,  6 Mar 2010 12:26:16 -0800 (PST)
Authentication-Results: ams-iport-1.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AvgAACZJkkuQ/uCWe2dsb2JhbACbTBUBAQsLJAYcnheYF4R4BA
X-IronPort-AV: E=Sophos;i="4.49,595,1262563200"; d="scan'208";a="57779093"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 06 Mar 2010 20:26:18 +0000
Received: from cisco.com (mrwint.cisco.com [64.103.71.48]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o26KQIKq021845; Sat, 6 Mar 2010 20:26:18 GMT
Received: from [10.55.163.37] (ams-andrjohn-8714.cisco.com [10.55.163.37]) by cisco.com (8.11.7p3+Sun/8.8.8) with ESMTP id o26KQHX16332; Sat, 6 Mar 2010 20:26:17 GMT
Message-ID: <4B92BA5F.1060601@cisco.com>
Date: Sat, 06 Mar 2010 20:26:07 +0000
From: Andrew Johnson <andrjohn@cisco.com>
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com> <4B9225A5.5030709@net.in.tum.de>
In-Reply-To: <4B9225A5.5030709@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sat, 06 Mar 2010 20:26:19 -0000

Hi Gerhard

Gerhard Muenz wrote:
> Hi Andrew,
> 
> Thanks for you reply.
> 
> Andrew Johnson wrote:
>> Hi Gerhard
>>
>> In this first part of my reply I'd like to address a specific point.
>>
>> Gerhard Muenz wrote:
>> [SNIP]
>>>> I can see why it would be convenient to configure multiple caches to
>>>> be populated from a single Selection Process, and it could even be
>>>> implemented that way (to some extent), but that is not the PSAMP model.
>>>> The configuration model doesn't have to exactly match this model, but
>>>> you should ensure that the differences are made clear.
>>> The PSAMP model has no notion of Caches.
>>>
>>> Could you please show me the paragraphs in the PSAMP documents which
>>> would indirectly prevent the existence of more than one Cache per
>>> Metering Process?
>> I didn't say that you were PSAMP restricted you to one cache per
>> Metering Process.  My point was about Selection Processes and my concern
>> is over the statistics that PSAMP requires exported for each Selection
>> Process.
>>
>> For example, imagine a simple selection process configuration that
>> selects 1 in 10 packets.  That configuration could first be used with
>> Cache A:
>>
>>     Selection Process A1:
>>       Observed packet stream count: 100
>>       Selected packet stream count:  10
>>
>>     Cache A:
>>       Total packets in: 10
>>
>>
>> Now that same configuration can be used to populate another Cache later,
>> but the statistics need kept separate for each Metering Process:
>>
>>     Selection Process A1:
>>       Observed packet stream count: 130
>>       Selected packet stream count:  13
>>
>>     Cache A:
>>       Total packets in: 10
>>
>>
>>     Selection Process B1:
>>       Observed packet stream count:  30
>>       Selected packet stream count:   3
>>
>>     Cache B:
>>       Total packets in:  3
>>
>> PSAMP requires that these stats are kept and exported, which means that
>> the stats for A1 and B1 need to be exported under different Selection
>> Process IDs.  I guess the real restriction is that a new cache cannot be
>> populated from an existing Selection Process.  While a template
>> configuration can be used it needs to spawn a separate instances so that
>> the stats can kept separate.
> 
> This is a good example. I see it differently, though.
> 
> In order to forward the output of the Selection Process to a second
> Cache, you need to change the configuration of the Selection Process.
> See how it looks in XML:
> 
> On Cache:
> 
> SP A1 -> Cache A
> 
>   <selectionProcess>
>     <name>A1</name>
>     <selector>
>       ...
>     </selector>
>     <cache>A</cache>
>   </selectionProcess>
> 
> After adding the second Cache:
> 
>        +-> Cache A
> SP A1 -+
>        +-> Cache B
> 
>   <selectionProcess>
>     <name>A1</name>
>     <selector>
>       ...
>     </selector>
>     <cache>A</cache>
>     <cache>B</cache>
>   </selectionProcess>
> 
> So, switching from the first configuration to the second one involves a
> reconfiguration of the Selection Process A1.
> 
> Regardless of how it is realized, the counters need to be reset because
> the configuration of the SP has changed.

Well, it might not be clear that the stats need cleared because of the 
nature of the change doesn't effect the selection behaviour of the 
Selection Process.

What about this case:

     SP A1 -> SP B1 -> Cache A

   becomes

                     +-> Cache A
     SP A1 -> SP B1 -+
                     +-> Cache B

In this case SP A1 didn't have its configuration changed at all, yet it 
must have its stats reset for Cache B.  In fact, if you only have one 
set of stats for each SP, then adding a new Cache means resetting the 
stats of the entire tree.  i.e. if the above was part of a tree like 
this, then all every SP in the tree would need its stats reset when 
Cache B was added:

                           +-> Cache A
                 +-> SP Y -+
                 |         +-> Cache C
    OP1 -> SP Z -+
                 |                   +-> Cache A
                 +-> SP A1 -> SP B1 -+
                                     +-> Cache B

Since Cache B needs fresh stats, B1 needs reset, therefore A1 needs 
reset, therefore Z needs reset, which then means all its children need 
reset.


> Before the configuration of the SP is changed, it makes sense to export
> all old counters. Otherwise, they will be lost.
> 
> Probably, the cacheDataRecords counter of the Cache should also be reset
> if the Cache is reconfigured.

Is the cache being reconfigured or just populated from a different source?

You seem to be making assumptions here about how I need to manage my 
stats.  I can conceive of a PSAMP compliant implementation that can be 
configured using your YANG model, that doesn't need to reset any stats 
when a new cache is added to a Selection Sequence.

It would be acceptable for each selection process in the chain to keep 
separate stats for each cache that is feeds.  In fact the configuration 
would act as a template and a new instance of each Selector would be 
created for each use.


> Should I add a sentence which says that packetsObserved, packetsDropped,
> cacheDataRecords need to be reset upon reconfiguration?

That would make sense.  However, my concern was over the stats that are 
exported by PSAMP, which is a different story.


> BTW: There are several ways to implement a switch from one configuration
> to a new one. If your processes do not support reconfiguration at
> runtime, you can replace them by newly instantiated processes. In this
> case, the counters will be reset, anyway.
> 
> Cheers,
> Gerhard
> 

From dromasca@avaya.com  Sun Mar  7 01:39:06 2010
Return-Path: <dromasca@avaya.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 2128E3A90A4 for <ipfix@core3.amsl.com>; Sun,  7 Mar 2010 01:39:06 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.537
X-Spam-Level: 
X-Spam-Status: No, score=-2.537 tagged_above=-999 required=5 tests=[AWL=0.062,  BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id EJ+8pPeToVYf for <ipfix@core3.amsl.com>; Sun,  7 Mar 2010 01:39:04 -0800 (PST)
Received: from p-us1-iereast-outbound-tmp.us1.avaya.com (p-us1-iereast-outbound-tmp.us1.avaya.com [135.11.29.16]) by core3.amsl.com (Postfix) with ESMTP id 94E5C3A909D for <ipfix@ietf.org>; Sun,  7 Mar 2010 01:39:04 -0800 (PST)
X-IronPort-AV: E=Sophos;i="4.49,597,1262581200";  d="scan'208";a="6508749"
Received: from unknown (HELO co300216-co-erhwest.avaya.com) ([198.152.7.5]) by p-us1-iereast-outbound-tmp.us1.avaya.com with ESMTP; 07 Mar 2010 04:39:07 -0500
X-IronPort-AV: E=Sophos;i="4.49,597,1262581200"; d="scan'208";a="452968523"
Received: from unknown (HELO 307622ANEX5.global.avaya.com) ([135.64.140.15]) by co300216-co-erhwest-out.avaya.com with ESMTP; 07 Mar 2010 04:39:06 -0500
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Date: Sun, 7 Mar 2010 10:38:20 +0100
Message-ID: <EDC652A26FB23C4EB6384A4584434A0401FC3DE6@307622ANEX5.global.avaya.com>
In-Reply-To: <4B9107E1.1050308@net.in.tum.de>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [IPFIX] IPFIX-CONFIG: maxPacketSize parameter for UDP export
Thread-Index: Acq8aEP5dFNfX3m0Rw2Se9He8fPu/gBcXS1Q
References: <4B9107E1.1050308@net.in.tum.de>
From: "Romascanu, Dan (Dan)" <dromasca@avaya.com>
To: "Gerhard Muenz" <muenz@net.in.tum.de>, <ipfix@ietf.org>
Cc: Daniel Mentz <mentz@in.tum.de>
Subject: Re: [IPFIX] IPFIX-CONFIG: maxPacketSize parameter for UDP export
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sun, 07 Mar 2010 09:39:06 -0000

RFC 5101 states the requirement as a SHOULD. Why do you propose to turn =
it into a MUST?=20

Thanks and Regards,

Dan
=20

> -----Original Message-----
> From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org]=20
> On Behalf Of Gerhard Muenz
> Sent: Friday, March 05, 2010 3:32 PM
> To: ipfix@ietf.org
> Cc: Daniel Mentz
> Subject: [IPFIX] IPFIX-CONFIG: maxPacketSize parameter for UDP export
>=20
>=20
> Dear all,
>=20
> I've missed a configuration issue mentioned in RFC5101:
>=20
> 10.3.3. MTU
>=20
>     The maximum size of exported messages MUST be configured such that
>     the total packet size does not exceed the path MTU.  If=20
> the path MTU
>     is unknown, a maximum packet size of 512 octets SHOULD be used.
>=20
>=20
> Although the text mentions the configuration of the IPFIX=20
> Message size, I think that it makes more sense to configure=20
> the maximum packet size (including IP and UDP header).
>=20
> So, I propose to add a maxPacketSize parameter to the=20
> Destination class:
>=20
>=20
>     +--------------------------------------------------+
>     | Destination                                      |
>     +--------------------------------------------------+
>     | name                                             |<>-----+
>     | exportMemberType =3D parallel                      |       | =
0..1
>     | ipfixVersion =3D 10                                |       |
>     | transportProtocol                                |  =20
> +------------+
>     | sourceIpAddress[0..*] {[0..1] for UDP and TCP}   |   |=20
> Transport- |
>     | destinationIpAddress[1..*] {[1] for UDP and TCP} |   |=20
> Layer-     |
>     | destinationPort =3D 4739|4740                      |   |=20
> Security   |
>     | ifName/ifIndex[0..1]                             |  =20
> +------------+
>     | sendBufferSize {opt.}                            |
>     | rateLimit[0..1]                                  |
>     | timedReliability =3D 0 {SCTP only}                 |
>     | numberOfStreams {opt.} {SCTP only}               |
> =3D> | maxPacketSize[0..1] {UDP only}                   |
>     | templateRefreshTimeout =3D 600 {UDP only}          |
>     | optionsTemplateRefreshTimeout =3D 600 {UDP only}   |
>     | templateRefreshPacket[0..1] {UDP only}           |
>     | optionsTemplateRefreshPacket[0..1] {UDP only}    |
>     +--------------------------------------------------+
>=20
>=20
>     maxPacketSize:  This parameter allow the user to specify=20
> the maximum
>        size of packets sent to the Collector.  If configured, the user
>        MUST ensure that the parameter value does not exceed=20
> the path MTU.
>        If not configured, the Exporting Process MAY derive the maximum
>        packet size from path MTU discovery mechanisms.  Otherwise, the
>        packet size MUST not exceed 512 octets as specified in=20
> [RFC5101].
>=20
>=20
> Please let me know if you like the parameter description.
>=20
> I have noticed that path MTU discovery for UDP is not yet=20
> mentioned in any IPFIX documents. Not sure, if it is ok to=20
> mention it here.
>=20
> Gerhard
>=20
>=20
> --
> Dipl.-Ing. Gerhard M=FCnz
> Chair for Network Architectures and Services (I8) Technische=20
> Universit=E4t M=FCnchen - Department of Informatics Boltzmannstr.=20
> 3, 85748 Garching bei M=FCnchen, Germany
> Phone:  +49 89 289-18008       Fax: +49 89 289-18033
> E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz
>=20
>=20
>=20

From root@core3.amsl.com  Sun Mar  7 05:45:02 2010
Return-Path: <root@core3.amsl.com>
X-Original-To: ipfix@ietf.org
Delivered-To: ipfix@core3.amsl.com
Received: by core3.amsl.com (Postfix, from userid 0) id 560FE3A8A05; Sun,  7 Mar 2010 05:45:02 -0800 (PST)
From: Internet-Drafts@ietf.org
To: i-d-announce@ietf.org
Content-Type: Multipart/Mixed; Boundary="NextPart"
Mime-Version: 1.0
Message-Id: <20100307134502.560FE3A8A05@core3.amsl.com>
Date: Sun,  7 Mar 2010 05:45:02 -0800 (PST)
Cc: ipfix@ietf.org
Subject: [IPFIX] I-D Action:draft-ietf-ipfix-flow-selection-tech-01.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sun, 07 Mar 2010 13:45:02 -0000

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts directories.
This draft is a work item of the IP Flow Information Export Working Group of the IETF.


	Title           : Flow Selection Techniques
	Author(s)       : L. Peluso, et al.
	Filename        : draft-ietf-ipfix-flow-selection-tech-01.txt
	Pages           : 24
	Date            : 2010-03-07

Flow selection is the process of selecting a subset of flows from all
flows observed at an observation point.  The objective of flow
selection is to reduce the effor for post-processing flow data and
for transfering flow records.  The flow selection process can be
enabled at different stages of the measurement process.  It can be
applied directly after classification or at recording/exporting time
by limiting the number of flows to be stored and/or exported to the
collecting process.  This document describes motivations for flow
selection and presents flow selection techniques.  It furthermore
provides an information model for configuring flow selection
techniques and discusses what information about a flow selection
process is beneficial to be exported by adopting a suitable
information model.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-ipfix-flow-selection-tech-01.txt

Internet-Drafts are also available by anonymous FTP at:
ftp://ftp.ietf.org/internet-drafts/

Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

--NextPart
Content-Type: Message/External-body;
	name="draft-ietf-ipfix-flow-selection-tech-01.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID: <2010-03-07053923.I-D@ietf.org>


--NextPart--

From n.brownlee@auckland.ac.nz  Sun Mar  7 18:13:05 2010
Return-Path: <n.brownlee@auckland.ac.nz>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 9A5D03A67E1 for <ipfix@core3.amsl.com>; Sun,  7 Mar 2010 18:13:05 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.299
X-Spam-Level: 
X-Spam-Status: No, score=-6.299 tagged_above=-999 required=5 tests=[AWL=-0.300, BAYES_00=-2.599, J_CHICKENPOX_13=0.6, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id DRMJHUQdnJNH for <ipfix@core3.amsl.com>; Sun,  7 Mar 2010 18:13:04 -0800 (PST)
Received: from mailhost.auckland.ac.nz (larry.its.auckland.ac.nz [130.216.12.34]) by core3.amsl.com (Postfix) with ESMTP id 100D43A6803 for <ipfix@ietf.org>; Sun,  7 Mar 2010 18:13:01 -0800 (PST)
Received: from localhost (localhost.localdomain [127.0.0.1]) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 2B4F6415C41; Mon,  8 Mar 2010 15:12:53 +1300 (NZDT)
X-Virus-Scanned: by amavisd-new at mailhost.auckland.ac.nz
Received: from mailhost.auckland.ac.nz ([127.0.0.1]) by localhost (larry.its.auckland.ac.nz [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id dxKGhYYzIpVS; Mon,  8 Mar 2010 15:12:53 +1300 (NZDT)
Received: from nevil-laptop.sfac.auckland.ac.nz (nevil-laptop.sfac.auckland.ac.nz [130.216.38.130]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 2035A415C52; Mon,  8 Mar 2010 15:12:49 +1300 (NZDT)
Message-ID: <4B945D21.3040804@auckland.ac.nz>
Date: Mon, 08 Mar 2010 15:12:49 +1300
From: Nevil Brownlee <n.brownlee@auckland.ac.nz>
User-Agent: Thunderbird 2.0.0.12 (Macintosh/20080213)
MIME-Version: 1.0
To: IPFIX Working Group <ipfix@ietf.org>
References: <AE36820147909644AD2A7CA014B1FB520A042922@xmb-sjc-222.amer.cisco.com> <20100218135814.C821.1AB7FA03@nttv6.net>
In-Reply-To: <20100218135814.C821.1AB7FA03@nttv6.net>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Subject: [IPFIX] Comments on draft-kashima-ipfix-data-link-layer-monitoring-01.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 08 Mar 2010 02:13:05 -0000

Hi all:

This draft introuces three new IPFIX Information Elements (IEs)
that would enable IPFIX to be used for monitoring (parts of) in
data-link-layer parts of packet headers.  It concentrates on Ethernet,
especially as used in providing layer-2 VPN services (IEEE 802.1ad
and 802.1ah).

Now that I've read it and thought about it, I have three comments:

1. The IPFIX/PSAMP architecture standardises the collection of
    *IP-related* data, this proposal would extend it to cover data from
    other parts of a packet's headers.  That would be a significant
    extension.  Is there support for such a generalisation?

2. It's presented as a simple, minimal extension to IPFIX/PSAMP,
    simply adding three new IEs PSAMP IEs, i.e with numbers above 338.
    Does it provide enough detail?  Should there be more than just three?
    For example, how about having separate IEs for S-VID, C-VID, B-VID
    and I=SID?  What other IEs are needed for PPP and WiMAX?

3. The description of proposed IE dataLinkFrameType seems rather
    minimal - eventually we'll want to add more data link types.  For
    example, the ifType-MIB (www.iana.org/assignments/ianaiftype-mib)
    has a much longer list.  Maybe we have to choose here between a long
    list of types, each with its own set of fields, or a short list of
    very general fields?  A short, simple list seems very appealing!

Cheers, Nevil

-- 
---------------------------------------------------------------------
  Nevil Brownlee                    Computer Science Department | ITS
  Phone: +64 9 373 7599 x88941             The University of Auckland
  FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand

From braun@net.in.tum.de  Sun Mar  7 23:42:03 2010
Return-Path: <braun@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 7D0833A6937 for <ipfix@core3.amsl.com>; Sun,  7 Mar 2010 23:42:03 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id zqCAOvz66TK7 for <ipfix@core3.amsl.com>; Sun,  7 Mar 2010 23:42:02 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 85ED83A686E for <ipfix@ietf.org>; Sun,  7 Mar 2010 23:42:02 -0800 (PST)
Received: from honshu.net.in.tum.de (honshu.net.in.tum.de [131.159.20.100]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 0459E200A793 for <ipfix@ietf.org>; Mon,  8 Mar 2010 08:42:05 +0100 (CET)
Content-Type: text/plain; charset=iso-8859-1
Mime-Version: 1.0 (Apple Message framework v1077)
From: Lothar Braun <braun@net.in.tum.de>
In-Reply-To: <930BF37B-7E33-440F-88CC-4771705A5C26@net.in.tum.de>
Date: Mon, 8 Mar 2010 08:42:04 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <E79680ED-37CD-401B-BB2D-26BEF3F2852A@net.in.tum.de>
References: <20100301094119.0F8423A8B2F@core3.amsl.com> <930BF37B-7E33-440F-88CC-4771705A5C26@net.in.tum.de>
To: IETF IPFIX Working Group <ipfix@ietf.org>
X-Mailer: Apple Mail (2.1077)
X-Virus-Scanned: ClamAV using ClamSMTP
Subject: Re: [IPFIX] Fwd: New Version Notification for draft-braun-core-compressed-ipfix-00
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 08 Mar 2010 07:42:03 -0000

Hi all,=20

we submitted an update on the draft "Compressed IPFIX for smart meters =
in constrained networks":

http://tools.ietf.org/html/draft-braun-core-compressed-ipfix-01

Best regards,
  Lothar

On Mar 3, 2010, at 10:33 AM, Lothar Braun wrote:

> Hi all,
>=20
> as you might have seen on the agenda, there is a draft Compressed =
IPFIX for smart meters in constrained networks:
>=20
> http://tools.ietf.org/id/draft-braun-core-compressed-ipfix-00.txt
>=20
> This draft replaces draft-schmitt-6lowapp-ipfix-ws and defines a =
modified version of IPFIX that can be used to export sensor data from =
constrained smart meters in 6lowpan networks. We are working on a =
01-version for the meeting (there should only be rather small fixes) and =
we would like to hear your comments on the document.
>=20
> Best regards,
> Lothar
>=20
> Begin forwarded message:
>=20
>> From: IETF I-D Submission Tool <idsubmission@ietf.org>
>> Date: March 1, 2010 10:41:19 AM GMT+01:00
>> To: braun@net.in.tum.de
>> Cc: schmitt@net.in.tum.de,bclaise@cisco.com,carle@net.in.tum.de
>> Subject: New Version Notification for =
draft-braun-core-compressed-ipfix-00=20
>>=20
>>=20
>> A new version of I-D, draft-braun-core-compressed-ipfix-00.txt has =
been successfuly submitted by Lothar Braun and posted to the IETF =
repository.
>>=20
>> Filename:	 draft-braun-core-compressed-ipfix
>> Revision:	 00
>> Title:		 Compressed IPFIX for smart meters in =
constrained networks
>> Creation_date:	 2010-03-01
>> WG ID:		 Independent Submission
>> Number_of_pages: 25
>>=20
>> Abstract:
>> This document specifies the Compressed IPFIX protocol that serves for
>> transmitting measurement data in 6LoWPAN networks [RFC4944].
>> Compressed IPFIX is derived from IPFIX [RFC5101] and adopted to the
>> needs of constrained networks.  This documents defines how the
>> Compressed IPFIX Data and Template Records are transmitted in 6LoWPAN
>> networks and how Compressed IPFIX data can be converted into
>> uncompressed IPFIX data.
>>=20
>>=20
>>=20
>> The IETF Secretariat.
>>=20
>>=20
>=20
> --
> Lothar Braun
> Chair for Network Architectures and Services (I8)
> Department of Informatics
> Technische Universit=E4t M=FCnchen
> Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
> Phone:  +49 89 289-18010       Fax: +49 89 289-18033
> E-mail: braun@net.in.tum.de=20
>=20
>=20
>=20
>=20
>=20
>=20
> _______________________________________________
> IPFIX mailing list
> IPFIX@ietf.org
> https://www.ietf.org/mailman/listinfo/ipfix

--
Lothar Braun
Chair for Network Architectures and Services (I8)
Department of Informatics
Technische Universit=E4t M=FCnchen
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18010       Fax: +49 89 289-18033
E-mail: braun@net.in.tum.de=20







From muenz@net.in.tum.de  Mon Mar  8 01:07:42 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id A79453A67CF for <ipfix@core3.amsl.com>; Mon,  8 Mar 2010 01:07:42 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id THbCDi-ema0a for <ipfix@core3.amsl.com>; Mon,  8 Mar 2010 01:07:41 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 53D883A67A1 for <ipfix@ietf.org>; Mon,  8 Mar 2010 01:07:40 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id D8945200A793; Mon,  8 Mar 2010 10:07:42 +0100 (CET)
Message-ID: <4B94BE6E.9010307@net.in.tum.de>
Date: Mon, 08 Mar 2010 10:07:58 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: "Romascanu, Dan (Dan)" <dromasca@avaya.com>
References: <4B9107E1.1050308@net.in.tum.de> <EDC652A26FB23C4EB6384A4584434A0401FC3DE6@307622ANEX5.global.avaya.com>
In-Reply-To: <EDC652A26FB23C4EB6384A4584434A0401FC3DE6@307622ANEX5.global.avaya.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms050801050205090101040000"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: Daniel Mentz <mentz@in.tum.de>, ipfix@ietf.org
Subject: Re: [IPFIX] IPFIX-CONFIG: maxPacketSize parameter for UDP export
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 08 Mar 2010 09:07:42 -0000

This is a cryptographically signed message in MIME format.

--------------ms050801050205090101040000
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit


Dear Dan, all,

Thanks for starting the discussion about this parameter.

Romascanu, Dan (Dan) wrote:
> RFC 5101 states the requirement as a SHOULD. Why do you propose to turn it into a MUST? 

The SHOULD in RFC 5101 and the MUST in IPFIX-CONFIG refer to different 
things.

I see
   "If the path MTU is unknown, a maximum packet size of 512 octets
    SHOULD be used" [RFC5101]
as a recommendation to the user:
If the user does not know the PMTU, he should configure
maxPacketSize = 512. He may configure maxPacketSize = 1400, though.

I want to specify how the Monitoring Device behaves if the maxPacketSize 
parameter is not configured by the user.

There are several possibilities how to do that. Here are three of them:

1) We specify a default value in IPFIX-CONFIG.
2) The devices configures the parameter value (like other parameters
    marked "{opt.} in UML).
3) We specify that the absence of maxPacketSize instructs the device
    to use PMTU discovery or 512 bytes.

Among the three options, only 3) allows using PMTU discovery.

1) is not good because it disallows PMTU discovery. And the default 
value would have to be 512.

2) means that the device configures maxPacketSize right after the user 
configures the Exporting Process without setting maxPacketSize. 
Thereafter, the device is expected not to change the parameter. Thus, 2) 
is incompatible with PMTU discovery as well because PMTU discovery may 
lead to changing PMTU estimates.

I would appreciate to hear more opinions. Maybe somebody has a better 
solution.

Apart from the configuration parameter, it could be useful to have a 
state parameter which allows the user to query the PMTU estimate for a 
UDP Transport Session if PMTU discovery is deployed.

Regards,
Gerhard


> 
> Thanks and Regards,
> 
> Dan
>  
> 
>> -----Original Message-----
>> From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] 
>> On Behalf Of Gerhard Muenz
>> Sent: Friday, March 05, 2010 3:32 PM
>> To: ipfix@ietf.org
>> Cc: Daniel Mentz
>> Subject: [IPFIX] IPFIX-CONFIG: maxPacketSize parameter for UDP export
>>
>>
>> Dear all,
>>
>> I've missed a configuration issue mentioned in RFC5101:
>>
>> 10.3.3. MTU
>>
>>     The maximum size of exported messages MUST be configured such that
>>     the total packet size does not exceed the path MTU.  If 
>> the path MTU
>>     is unknown, a maximum packet size of 512 octets SHOULD be used.
>>
>>
>> Although the text mentions the configuration of the IPFIX 
>> Message size, I think that it makes more sense to configure 
>> the maximum packet size (including IP and UDP header).
>>
>> So, I propose to add a maxPacketSize parameter to the 
>> Destination class:
>>
>>
>>     +--------------------------------------------------+
>>     | Destination                                      |
>>     +--------------------------------------------------+
>>     | name                                             |<>-----+
>>     | exportMemberType = parallel                      |       | 0..1
>>     | ipfixVersion = 10                                |       |
>>     | transportProtocol                                |   
>> +------------+
>>     | sourceIpAddress[0..*] {[0..1] for UDP and TCP}   |   | 
>> Transport- |
>>     | destinationIpAddress[1..*] {[1] for UDP and TCP} |   | 
>> Layer-     |
>>     | destinationPort = 4739|4740                      |   | 
>> Security   |
>>     | ifName/ifIndex[0..1]                             |   
>> +------------+
>>     | sendBufferSize {opt.}                            |
>>     | rateLimit[0..1]                                  |
>>     | timedReliability = 0 {SCTP only}                 |
>>     | numberOfStreams {opt.} {SCTP only}               |
>> => | maxPacketSize[0..1] {UDP only}                   |
>>     | templateRefreshTimeout = 600 {UDP only}          |
>>     | optionsTemplateRefreshTimeout = 600 {UDP only}   |
>>     | templateRefreshPacket[0..1] {UDP only}           |
>>     | optionsTemplateRefreshPacket[0..1] {UDP only}    |
>>     +--------------------------------------------------+
>>
>>
>>     maxPacketSize:  This parameter allow the user to specify 
>> the maximum
>>        size of packets sent to the Collector.  If configured, the user
>>        MUST ensure that the parameter value does not exceed 
>> the path MTU.
>>        If not configured, the Exporting Process MAY derive the maximum
>>        packet size from path MTU discovery mechanisms.  Otherwise, the
>>        packet size MUST not exceed 512 octets as specified in 
>> [RFC5101].
>>
>>
>> Please let me know if you like the parameter description.
>>
>> I have noticed that path MTU discovery for UDP is not yet 
>> mentioned in any IPFIX documents. Not sure, if it is ok to 
>> mention it here.
>>
>> Gerhard
>>


--------------ms050801050205090101040000
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms050801050205090101040000--

From trammell@tik.ee.ethz.ch  Mon Mar  8 01:23:10 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id A3D613A6811 for <ipfix@core3.amsl.com>; Mon,  8 Mar 2010 01:23:10 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.599
X-Spam-Level: 
X-Spam-Status: No, score=-6.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id BfW8swI+va04 for <ipfix@core3.amsl.com>; Mon,  8 Mar 2010 01:23:05 -0800 (PST)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id E90043A680D for <ipfix@ietf.org>; Mon,  8 Mar 2010 01:23:04 -0800 (PST)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id 1BF65D9399; Mon,  8 Mar 2010 10:23:07 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id wJCdkL+S6g1J; Mon,  8 Mar 2010 10:23:06 +0100 (MET)
Received: from pb-10072.ethz.ch (pb-10072.ethz.ch [82.130.103.195]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id CCCB9D9360; Mon,  8 Mar 2010 10:23:06 +0100 (MET)
Mime-Version: 1.0 (Apple Message framework v1077)
Content-Type: text/plain; charset=us-ascii
From: Brian Trammell <trammell@tik.ee.ethz.ch>
In-Reply-To: <4B94BE6E.9010307@net.in.tum.de>
Date: Mon, 8 Mar 2010 10:23:06 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <E318021F-9CC3-4173-96A2-1D0AD2DB47C1@tik.ee.ethz.ch>
References: <4B9107E1.1050308@net.in.tum.de> <EDC652A26FB23C4EB6384A4584434A0401FC3DE6@307622ANEX5.global.avaya.com> <4B94BE6E.9010307@net.in.tum.de>
To: Gerhard Muenz <muenz@net.in.tum.de>
X-Mailer: Apple Mail (2.1077)
Cc: Daniel Mentz <mentz@in.tum.de>, ipfix@ietf.org
Subject: Re: [IPFIX] IPFIX-CONFIG: maxPacketSize parameter for UDP export
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 08 Mar 2010 09:23:10 -0000

Hi, Gerhard, all,

I would propose 3a. The absence of maxPacketSize instructs the device to =
use PMTU discovery if supported, or the device default if not. The =
device default SHOULD be 512 bytes as per [RFC5101].

It has all the advantages of your approach, plus it doesn't supercede a =
SHOULD with a MUST.

Thoughts?

Cheers,

Brian

On Mar 8, 2010, at 10:07 AM, Gerhard Muenz wrote:

>=20
> Dear Dan, all,
>=20
> Thanks for starting the discussion about this parameter.
>=20
> Romascanu, Dan (Dan) wrote:
>> RFC 5101 states the requirement as a SHOULD. Why do you propose to =
turn it into a MUST?=20
>=20
> The SHOULD in RFC 5101 and the MUST in IPFIX-CONFIG refer to different =
things.
>=20
> I see
>  "If the path MTU is unknown, a maximum packet size of 512 octets
>   SHOULD be used" [RFC5101]
> as a recommendation to the user:
> If the user does not know the PMTU, he should configure
> maxPacketSize =3D 512. He may configure maxPacketSize =3D 1400, =
though.
>=20
> I want to specify how the Monitoring Device behaves if the =
maxPacketSize parameter is not configured by the user.
>=20
> There are several possibilities how to do that. Here are three of =
them:
>=20
> 1) We specify a default value in IPFIX-CONFIG.
> 2) The devices configures the parameter value (like other parameters
>   marked "{opt.} in UML).
> 3) We specify that the absence of maxPacketSize instructs the device
>   to use PMTU discovery or 512 bytes.
>=20
> Among the three options, only 3) allows using PMTU discovery.
>=20
> 1) is not good because it disallows PMTU discovery. And the default =
value would have to be 512.
>=20
> 2) means that the device configures maxPacketSize right after the user =
configures the Exporting Process without setting maxPacketSize. =
Thereafter, the device is expected not to change the parameter. Thus, 2) =
is incompatible with PMTU discovery as well because PMTU discovery may =
lead to changing PMTU estimates.
>=20
> I would appreciate to hear more opinions. Maybe somebody has a better =
solution.
>=20
> Apart from the configuration parameter, it could be useful to have a =
state parameter which allows the user to query the PMTU estimate for a =
UDP Transport Session if PMTU discovery is deployed.
>=20
> Regards,
> Gerhard
>=20
>=20
>> Thanks and Regards,
>> Dan
>>=20
>>> -----Original Message-----
>>> From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] On =
Behalf Of Gerhard Muenz
>>> Sent: Friday, March 05, 2010 3:32 PM
>>> To: ipfix@ietf.org
>>> Cc: Daniel Mentz
>>> Subject: [IPFIX] IPFIX-CONFIG: maxPacketSize parameter for UDP =
export
>>>=20
>>>=20
>>> Dear all,
>>>=20
>>> I've missed a configuration issue mentioned in RFC5101:
>>>=20
>>> 10.3.3. MTU
>>>=20
>>>    The maximum size of exported messages MUST be configured such =
that
>>>    the total packet size does not exceed the path MTU.  If the path =
MTU
>>>    is unknown, a maximum packet size of 512 octets SHOULD be used.
>>>=20
>>>=20
>>> Although the text mentions the configuration of the IPFIX Message =
size, I think that it makes more sense to configure the maximum packet =
size (including IP and UDP header).
>>>=20
>>> So, I propose to add a maxPacketSize parameter to the Destination =
class:
>>>=20
>>>=20
>>>    +--------------------------------------------------+
>>>    | Destination                                      |
>>>    +--------------------------------------------------+
>>>    | name                                             |<>-----+
>>>    | exportMemberType =3D parallel                      |       | =
0..1
>>>    | ipfixVersion =3D 10                                |       |
>>>    | transportProtocol                                |   =
+------------+
>>>    | sourceIpAddress[0..*] {[0..1] for UDP and TCP}   |   | =
Transport- |
>>>    | destinationIpAddress[1..*] {[1] for UDP and TCP} |   | Layer-   =
  |
>>>    | destinationPort =3D 4739|4740                      |   | =
Security   |
>>>    | ifName/ifIndex[0..1]                             |   =
+------------+
>>>    | sendBufferSize {opt.}                            |
>>>    | rateLimit[0..1]                                  |
>>>    | timedReliability =3D 0 {SCTP only}                 |
>>>    | numberOfStreams {opt.} {SCTP only}               |
>>> =3D> | maxPacketSize[0..1] {UDP only}                   |
>>>    | templateRefreshTimeout =3D 600 {UDP only}          |
>>>    | optionsTemplateRefreshTimeout =3D 600 {UDP only}   |
>>>    | templateRefreshPacket[0..1] {UDP only}           |
>>>    | optionsTemplateRefreshPacket[0..1] {UDP only}    |
>>>    +--------------------------------------------------+
>>>=20
>>>=20
>>>    maxPacketSize:  This parameter allow the user to specify the =
maximum
>>>       size of packets sent to the Collector.  If configured, the =
user
>>>       MUST ensure that the parameter value does not exceed the path =
MTU.
>>>       If not configured, the Exporting Process MAY derive the =
maximum
>>>       packet size from path MTU discovery mechanisms.  Otherwise, =
the
>>>       packet size MUST not exceed 512 octets as specified in =
[RFC5101].
>>>=20
>>>=20
>>> Please let me know if you like the parameter description.
>>>=20
>>> I have noticed that path MTU discovery for UDP is not yet mentioned =
in any IPFIX documents. Not sure, if it is ok to mention it here.
>>>=20
>>> Gerhard
>>>=20
>=20
> _______________________________________________
> IPFIX mailing list
> IPFIX@ietf.org
> https://www.ietf.org/mailman/listinfo/ipfix


From muenz@net.in.tum.de  Mon Mar  8 01:53:40 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 725A23A68B2 for <ipfix@core3.amsl.com>; Mon,  8 Mar 2010 01:53:40 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[AWL=0.000,  BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id MgIXi1jKiq2U for <ipfix@core3.amsl.com>; Mon,  8 Mar 2010 01:53:39 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 9A44E3A6847 for <ipfix@ietf.org>; Mon,  8 Mar 2010 01:53:38 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id D6E82200A793; Mon,  8 Mar 2010 10:53:40 +0100 (CET)
Message-ID: <4B94C934.9020002@net.in.tum.de>
Date: Mon, 08 Mar 2010 10:53:56 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Brian Trammell <trammell@tik.ee.ethz.ch>
References: <4B9107E1.1050308@net.in.tum.de> <EDC652A26FB23C4EB6384A4584434A0401FC3DE6@307622ANEX5.global.avaya.com> <4B94BE6E.9010307@net.in.tum.de> <E318021F-9CC3-4173-96A2-1D0AD2DB47C1@tik.ee.ethz.ch>
In-Reply-To: <E318021F-9CC3-4173-96A2-1D0AD2DB47C1@tik.ee.ethz.ch>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms090604060309080003000708"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: Daniel Mentz <mentz@in.tum.de>, ipfix@ietf.org
Subject: Re: [IPFIX] IPFIX-CONFIG: maxPacketSize parameter for UDP export
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 08 Mar 2010 09:53:40 -0000

This is a cryptographically signed message in MIME format.

--------------ms090604060309080003000708
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Brian,

Brian Trammell wrote:
> Hi, Gerhard, all,
>=20
> I would propose 3a. The absence of maxPacketSize instructs the device t=
o use PMTU discovery if supported, or the device default if not. The devi=
ce default SHOULD be 512 bytes as per [RFC5101].
>=20
> It has all the advantages of your approach, plus it doesn't supercede a=
 SHOULD with a MUST.
>=20
> Thoughts?

This is possible as well.

The question is how we communicate to the user a) which value is used=20
and b) whether this value is a device-specific default or the result of=20
PMTU discovery.

Here is one possible solution:

    maxPacketSize {opt.} {UDP only}

    maxPacketSize:  This parameter specifies the maximum size of packets
       sent to the Collector.  If set to zero, the Exporting Device MUST
       derive the maximum packet size from path MTU discovery mechanisms.=

       If not configured by the user, the Monitoring Process sets this
       parameter.


So, if the user configures maxPacketSize =3D 0, the device MUST use PMTU =

discovery. If not supported by the device, the configuration fails.

If the user does not configure maxPacketSize, the device MAY set=20
maxPacketSize =3D 0, which means that PMTU discovery is used. Or it sets =
a=20
non-zero value, which is a device default.

Cheers,
Gerhard

>=20
> Cheers,
>=20
> Brian
>=20
> On Mar 8, 2010, at 10:07 AM, Gerhard Muenz wrote:
>=20
>> Dear Dan, all,
>>
>> Thanks for starting the discussion about this parameter.
>>
>> Romascanu, Dan (Dan) wrote:
>>> RFC 5101 states the requirement as a SHOULD. Why do you propose to tu=
rn it into a MUST?=20
>> The SHOULD in RFC 5101 and the MUST in IPFIX-CONFIG refer to different=
 things.
>>
>> I see
>>  "If the path MTU is unknown, a maximum packet size of 512 octets
>>   SHOULD be used" [RFC5101]
>> as a recommendation to the user:
>> If the user does not know the PMTU, he should configure
>> maxPacketSize =3D 512. He may configure maxPacketSize =3D 1400, though=
=2E
>>
>> I want to specify how the Monitoring Device behaves if the maxPacketSi=
ze parameter is not configured by the user.
>>
>> There are several possibilities how to do that. Here are three of them=
:
>>
>> 1) We specify a default value in IPFIX-CONFIG.
>> 2) The devices configures the parameter value (like other parameters
>>   marked "{opt.} in UML).
>> 3) We specify that the absence of maxPacketSize instructs the device
>>   to use PMTU discovery or 512 bytes.
>>
>> Among the three options, only 3) allows using PMTU discovery.
>>
>> 1) is not good because it disallows PMTU discovery. And the default va=
lue would have to be 512.
>>
>> 2) means that the device configures maxPacketSize right after the user=
 configures the Exporting Process without setting maxPacketSize. Thereaft=
er, the device is expected not to change the parameter. Thus, 2) is incom=
patible with PMTU discovery as well because PMTU discovery may lead to ch=
anging PMTU estimates.
>>
>> I would appreciate to hear more opinions. Maybe somebody has a better =
solution.
>>
>> Apart from the configuration parameter, it could be useful to have a s=
tate parameter which allows the user to query the PMTU estimate for a UDP=
 Transport Session if PMTU discovery is deployed.
>>
>> Regards,
>> Gerhard
>>
>>
>>> Thanks and Regards,
>>> Dan
>>>
>>>> -----Original Message-----
>>>> From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] On Beha=
lf Of Gerhard Muenz
>>>> Sent: Friday, March 05, 2010 3:32 PM
>>>> To: ipfix@ietf.org
>>>> Cc: Daniel Mentz
>>>> Subject: [IPFIX] IPFIX-CONFIG: maxPacketSize parameter for UDP expor=
t
>>>>
>>>>
>>>> Dear all,
>>>>
>>>> I've missed a configuration issue mentioned in RFC5101:
>>>>
>>>> 10.3.3. MTU
>>>>
>>>>    The maximum size of exported messages MUST be configured such tha=
t
>>>>    the total packet size does not exceed the path MTU.  If the path =
MTU
>>>>    is unknown, a maximum packet size of 512 octets SHOULD be used.
>>>>
>>>>
>>>> Although the text mentions the configuration of the IPFIX Message si=
ze, I think that it makes more sense to configure the maximum packet size=
 (including IP and UDP header).
>>>>
>>>> So, I propose to add a maxPacketSize parameter to the Destination cl=
ass:
>>>>
>>>>
>>>>    +--------------------------------------------------+
>>>>    | Destination                                      |
>>>>    +--------------------------------------------------+
>>>>    | name                                             |<>-----+
>>>>    | exportMemberType =3D parallel                      |       | 0.=
=2E1
>>>>    | ipfixVersion =3D 10                                |       |
>>>>    | transportProtocol                                |   +---------=
---+
>>>>    | sourceIpAddress[0..*] {[0..1] for UDP and TCP}   |   | Transpor=
t- |
>>>>    | destinationIpAddress[1..*] {[1] for UDP and TCP} |   | Layer-  =
   |
>>>>    | destinationPort =3D 4739|4740                      |   | Securi=
ty   |
>>>>    | ifName/ifIndex[0..1]                             |   +---------=
---+
>>>>    | sendBufferSize {opt.}                            |
>>>>    | rateLimit[0..1]                                  |
>>>>    | timedReliability =3D 0 {SCTP only}                 |
>>>>    | numberOfStreams {opt.} {SCTP only}               |
>>>> =3D> | maxPacketSize[0..1] {UDP only}                   |
>>>>    | templateRefreshTimeout =3D 600 {UDP only}          |
>>>>    | optionsTemplateRefreshTimeout =3D 600 {UDP only}   |
>>>>    | templateRefreshPacket[0..1] {UDP only}           |
>>>>    | optionsTemplateRefreshPacket[0..1] {UDP only}    |
>>>>    +--------------------------------------------------+
>>>>
>>>>
>>>>    maxPacketSize:  This parameter allow the user to specify the maxi=
mum
>>>>       size of packets sent to the Collector.  If configured, the use=
r
>>>>       MUST ensure that the parameter value does not exceed the path =
MTU.
>>>>       If not configured, the Exporting Process MAY derive the maximu=
m
>>>>       packet size from path MTU discovery mechanisms.  Otherwise, th=
e
>>>>       packet size MUST not exceed 512 octets as specified in [RFC510=
1].
>>>>
>>>>
>>>> Please let me know if you like the parameter description.
>>>>
>>>> I have noticed that path MTU discovery for UDP is not yet mentioned =
in any IPFIX documents. Not sure, if it is ok to mention it here.
>>>>
>>>> Gerhard
>>>>
>> _______________________________________________
>> IPFIX mailing list
>> IPFIX@ietf.org
>> https://www.ietf.org/mailman/listinfo/ipfix
>=20

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms090604060309080003000708
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms090604060309080003000708--

From trammell@tik.ee.ethz.ch  Mon Mar  8 01:54:46 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id AADB83A684A for <ipfix@core3.amsl.com>; Mon,  8 Mar 2010 01:54:46 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.599
X-Spam-Level: 
X-Spam-Status: No, score=-6.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id wovYfmI1w87U for <ipfix@core3.amsl.com>; Mon,  8 Mar 2010 01:54:45 -0800 (PST)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id 35AD63A681E for <ipfix@ietf.org>; Mon,  8 Mar 2010 01:54:44 -0800 (PST)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id 9F09CD936F; Mon,  8 Mar 2010 10:54:47 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id pwatxiLBonCH; Mon,  8 Mar 2010 10:54:47 +0100 (MET)
Received: from pb-10072.ethz.ch (pb-10072.ethz.ch [82.130.103.195]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id 58FE6D9341; Mon,  8 Mar 2010 10:54:47 +0100 (MET)
Mime-Version: 1.0 (Apple Message framework v1077)
Content-Type: text/plain; charset=iso-8859-1
From: Brian Trammell <trammell@tik.ee.ethz.ch>
In-Reply-To: <4B94C934.9020002@net.in.tum.de>
Date: Mon, 8 Mar 2010 10:54:46 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <968C7950-B8BD-4C07-B7F7-6B42ADC8C8F8@tik.ee.ethz.ch>
References: <4B9107E1.1050308@net.in.tum.de> <EDC652A26FB23C4EB6384A4584434A0401FC3DE6@307622ANEX5.global.avaya.com> <4B94BE6E.9010307@net.in.tum.de> <E318021F-9CC3-4173-96A2-1D0AD2DB47C1@tik.ee.ethz.ch> <4B94C934.9020002@net.in.tum.de>
To: Gerhard Muenz <muenz@net.in.tum.de>
X-Mailer: Apple Mail (2.1077)
Cc: Daniel Mentz <mentz@in.tum.de>, ipfix@ietf.org
Subject: Re: [IPFIX] IPFIX-CONFIG: maxPacketSize parameter for UDP export
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 08 Mar 2010 09:54:46 -0000

Hi, Gerhard,

Yep, this looks even better.

Cheers,

Brian

On Mar 8, 2010, at 10:53 AM, Gerhard Muenz wrote:

>=20
> Hi Brian,
>=20
> Brian Trammell wrote:
>> Hi, Gerhard, all,
>> I would propose 3a. The absence of maxPacketSize instructs the device =
to use PMTU discovery if supported, or the device default if not. The =
device default SHOULD be 512 bytes as per [RFC5101].
>> It has all the advantages of your approach, plus it doesn't supercede =
a SHOULD with a MUST.
>> Thoughts?
>=20
> This is possible as well.
>=20
> The question is how we communicate to the user a) which value is used =
and b) whether this value is a device-specific default or the result of =
PMTU discovery.
>=20
> Here is one possible solution:
>=20
>   maxPacketSize {opt.} {UDP only}
>=20
>   maxPacketSize:  This parameter specifies the maximum size of packets
>      sent to the Collector.  If set to zero, the Exporting Device MUST
>      derive the maximum packet size from path MTU discovery =
mechanisms.
>      If not configured by the user, the Monitoring Process sets this
>      parameter.
>=20
>=20
> So, if the user configures maxPacketSize =3D 0, the device MUST use =
PMTU discovery. If not supported by the device, the configuration fails.
>=20
> If the user does not configure maxPacketSize, the device MAY set =
maxPacketSize =3D 0, which means that PMTU discovery is used. Or it sets =
a non-zero value, which is a device default.
>=20
> Cheers,
> Gerhard
>=20
>> Cheers,
>> Brian
>> On Mar 8, 2010, at 10:07 AM, Gerhard Muenz wrote:
>>> Dear Dan, all,
>>>=20
>>> Thanks for starting the discussion about this parameter.
>>>=20
>>> Romascanu, Dan (Dan) wrote:
>>>> RFC 5101 states the requirement as a SHOULD. Why do you propose to =
turn it into a MUST?=20
>>> The SHOULD in RFC 5101 and the MUST in IPFIX-CONFIG refer to =
different things.
>>>=20
>>> I see
>>> "If the path MTU is unknown, a maximum packet size of 512 octets
>>>  SHOULD be used" [RFC5101]
>>> as a recommendation to the user:
>>> If the user does not know the PMTU, he should configure
>>> maxPacketSize =3D 512. He may configure maxPacketSize =3D 1400, =
though.
>>>=20
>>> I want to specify how the Monitoring Device behaves if the =
maxPacketSize parameter is not configured by the user.
>>>=20
>>> There are several possibilities how to do that. Here are three of =
them:
>>>=20
>>> 1) We specify a default value in IPFIX-CONFIG.
>>> 2) The devices configures the parameter value (like other parameters
>>>  marked "{opt.} in UML).
>>> 3) We specify that the absence of maxPacketSize instructs the device
>>>  to use PMTU discovery or 512 bytes.
>>>=20
>>> Among the three options, only 3) allows using PMTU discovery.
>>>=20
>>> 1) is not good because it disallows PMTU discovery. And the default =
value would have to be 512.
>>>=20
>>> 2) means that the device configures maxPacketSize right after the =
user configures the Exporting Process without setting maxPacketSize. =
Thereafter, the device is expected not to change the parameter. Thus, 2) =
is incompatible with PMTU discovery as well because PMTU discovery may =
lead to changing PMTU estimates.
>>>=20
>>> I would appreciate to hear more opinions. Maybe somebody has a =
better solution.
>>>=20
>>> Apart from the configuration parameter, it could be useful to have a =
state parameter which allows the user to query the PMTU estimate for a =
UDP Transport Session if PMTU discovery is deployed.
>>>=20
>>> Regards,
>>> Gerhard
>>>=20
>>>=20
>>>> Thanks and Regards,
>>>> Dan
>>>>=20
>>>>> -----Original Message-----
>>>>> From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] On =
Behalf Of Gerhard Muenz
>>>>> Sent: Friday, March 05, 2010 3:32 PM
>>>>> To: ipfix@ietf.org
>>>>> Cc: Daniel Mentz
>>>>> Subject: [IPFIX] IPFIX-CONFIG: maxPacketSize parameter for UDP =
export
>>>>>=20
>>>>>=20
>>>>> Dear all,
>>>>>=20
>>>>> I've missed a configuration issue mentioned in RFC5101:
>>>>>=20
>>>>> 10.3.3. MTU
>>>>>=20
>>>>>   The maximum size of exported messages MUST be configured such =
that
>>>>>   the total packet size does not exceed the path MTU.  If the path =
MTU
>>>>>   is unknown, a maximum packet size of 512 octets SHOULD be used.
>>>>>=20
>>>>>=20
>>>>> Although the text mentions the configuration of the IPFIX Message =
size, I think that it makes more sense to configure the maximum packet =
size (including IP and UDP header).
>>>>>=20
>>>>> So, I propose to add a maxPacketSize parameter to the Destination =
class:
>>>>>=20
>>>>>=20
>>>>>   +--------------------------------------------------+
>>>>>   | Destination                                      |
>>>>>   +--------------------------------------------------+
>>>>>   | name                                             |<>-----+
>>>>>   | exportMemberType =3D parallel                      |       | =
0..1
>>>>>   | ipfixVersion =3D 10                                |       |
>>>>>   | transportProtocol                                |   =
+------------+
>>>>>   | sourceIpAddress[0..*] {[0..1] for UDP and TCP}   |   | =
Transport- |
>>>>>   | destinationIpAddress[1..*] {[1] for UDP and TCP} |   | Layer-  =
   |
>>>>>   | destinationPort =3D 4739|4740                      |   | =
Security   |
>>>>>   | ifName/ifIndex[0..1]                             |   =
+------------+
>>>>>   | sendBufferSize {opt.}                            |
>>>>>   | rateLimit[0..1]                                  |
>>>>>   | timedReliability =3D 0 {SCTP only}                 |
>>>>>   | numberOfStreams {opt.} {SCTP only}               |
>>>>> =3D> | maxPacketSize[0..1] {UDP only}                   |
>>>>>   | templateRefreshTimeout =3D 600 {UDP only}          |
>>>>>   | optionsTemplateRefreshTimeout =3D 600 {UDP only}   |
>>>>>   | templateRefreshPacket[0..1] {UDP only}           |
>>>>>   | optionsTemplateRefreshPacket[0..1] {UDP only}    |
>>>>>   +--------------------------------------------------+
>>>>>=20
>>>>>=20
>>>>>   maxPacketSize:  This parameter allow the user to specify the =
maximum
>>>>>      size of packets sent to the Collector.  If configured, the =
user
>>>>>      MUST ensure that the parameter value does not exceed the path =
MTU.
>>>>>      If not configured, the Exporting Process MAY derive the =
maximum
>>>>>      packet size from path MTU discovery mechanisms.  Otherwise, =
the
>>>>>      packet size MUST not exceed 512 octets as specified in =
[RFC5101].
>>>>>=20
>>>>>=20
>>>>> Please let me know if you like the parameter description.
>>>>>=20
>>>>> I have noticed that path MTU discovery for UDP is not yet =
mentioned in any IPFIX documents. Not sure, if it is ok to mention it =
here.
>>>>>=20
>>>>> Gerhard
>>>>>=20
>>> _______________________________________________
>>> IPFIX mailing list
>>> IPFIX@ietf.org
>>> https://www.ietf.org/mailman/listinfo/ipfix
>=20
> --=20
> Dipl.-Ing. Gerhard M=FCnz
> Chair for Network Architectures and Services (I8)
> Technische Universit=E4t M=FCnchen - Department of Informatics
> Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
> Phone:  +49 89 289-18008       Fax: +49 89 289-18033
> E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz
>=20
>=20


From root@core3.amsl.com  Mon Mar  8 02:45:03 2010
Return-Path: <root@core3.amsl.com>
X-Original-To: ipfix@ietf.org
Delivered-To: ipfix@core3.amsl.com
Received: by core3.amsl.com (Postfix, from userid 0) id 7AEFA3A68C4; Mon,  8 Mar 2010 02:45:01 -0800 (PST)
From: Internet-Drafts@ietf.org
To: i-d-announce@ietf.org
Content-Type: Multipart/Mixed; Boundary="NextPart"
Mime-Version: 1.0
Message-Id: <20100308104502.7AEFA3A68C4@core3.amsl.com>
Date: Mon,  8 Mar 2010 02:45:02 -0800 (PST)
Cc: ipfix@ietf.org
Subject: [IPFIX] I-D Action:draft-ietf-ipfix-configuration-model-05.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 08 Mar 2010 10:45:03 -0000

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts directories.
This draft is a work item of the IP Flow Information Export Working Group of the IETF.


	Title           : Configuration Data Model for IPFIX and PSAMP
	Author(s)       : G. Muenz, et al.
	Filename        : draft-ietf-ipfix-configuration-model-05.txt
	Pages           : 99
	Date            : 2010-03-08

This document specifies a data model for the configuration of
Selection Processes, Caches, Exporting Processes, and Collecting
Processes of IPFIX and PSAMP compliant Monitoring Devices using UML
(Unified Modeling Language) class diagrams.  The configuration data
is encoded in Extensible Markup Language (XML).  The structure of the
data model is specified in a YANG module to ensure compatibility with
the NETCONF protocol.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-ipfix-configuration-model-05.txt

Internet-Drafts are also available by anonymous FTP at:
ftp://ftp.ietf.org/internet-drafts/

Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

--NextPart
Content-Type: Message/External-body;
	name="draft-ietf-ipfix-configuration-model-05.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID: <2010-03-08023339.I-D@ietf.org>


--NextPart--

From muenz@net.in.tum.de  Mon Mar  8 02:49:09 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 112A33A68C4 for <ipfix@core3.amsl.com>; Mon,  8 Mar 2010 02:49:09 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 9akLQ2z2yA7R for <ipfix@core3.amsl.com>; Mon,  8 Mar 2010 02:49:08 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id BE8743A68C1 for <ipfix@ietf.org>; Mon,  8 Mar 2010 02:49:07 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 9E6C0200A793 for <ipfix@ietf.org>; Mon,  8 Mar 2010 11:49:09 +0100 (CET)
Message-ID: <4B94D635.3020200@net.in.tum.de>
Date: Mon, 08 Mar 2010 11:49:25 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: "ipfix@ietf.org" <ipfix@ietf.org>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms080000030106030005060903"
X-Virus-Scanned: ClamAV using ClamSMTP
Subject: [IPFIX] draft-ietf-ipfix-configuration-model-05
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 08 Mar 2010 10:49:09 -0000

This is a cryptographically signed message in MIME format.

--------------ms080000030106030005060903
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit


Dear all,

I've posted draft-ietf-ipfix-configuration-model-05 which shows the 
current status of the draft.

Thanks a lot to the reviewers!

WGLC comments of Atsushi, Lothar, and Brian have been addressed.

Some issues raised by Andrew still need to be solved.

Since I will not go to Anaheim, the mailing list is the right place to 
continue the discussion. I hope that the remaining issues can be solved 
until the IPFIX session. If so, a -06 version will follow very soon.

Regards,
Gerhard






--------------ms080000030106030005060903
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms080000030106030005060903--

From root@core3.amsl.com  Mon Mar  8 09:15:04 2010
Return-Path: <root@core3.amsl.com>
X-Original-To: ipfix@ietf.org
Delivered-To: ipfix@core3.amsl.com
Received: by core3.amsl.com (Postfix, from userid 0) id D42483A6A5E; Mon,  8 Mar 2010 09:15:03 -0800 (PST)
From: Internet-Drafts@ietf.org
To: i-d-announce@ietf.org
Content-Type: Multipart/Mixed; Boundary="NextPart"
Mime-Version: 1.0
Message-Id: <20100308171503.D42483A6A5E@core3.amsl.com>
Date: Mon,  8 Mar 2010 09:15:03 -0800 (PST)
Cc: ipfix@ietf.org
Subject: [IPFIX] I-D Action:draft-ietf-ipfix-mediators-framework-05.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 08 Mar 2010 17:15:04 -0000

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts directories.
This draft is a work item of the IP Flow Information Export Working Group of the IETF.


	Title           : IPFIX Mediation: Framework
	Author(s)       : A. Kobayashi, et al.
	Filename        : draft-ietf-ipfix-mediators-framework-05.txt
	Pages           : 33
	Date            : 2010-03-08

This document describes a framework for IPFIX Mediation.  This
framework extends the IPFIX reference model by defining the IPFIX
Mediator components.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-ipfix-mediators-framework-05.txt

Internet-Drafts are also available by anonymous FTP at:
ftp://ftp.ietf.org/internet-drafts/

Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

--NextPart
Content-Type: Message/External-body;
	name="draft-ietf-ipfix-mediators-framework-05.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID: <2010-03-08091225.I-D@ietf.org>


--NextPart--

From root@core3.amsl.com  Mon Mar  8 10:30:02 2010
Return-Path: <root@core3.amsl.com>
X-Original-To: ipfix@ietf.org
Delivered-To: ipfix@core3.amsl.com
Received: by core3.amsl.com (Postfix, from userid 0) id 7F4E33A6A9E; Mon,  8 Mar 2010 10:30:02 -0800 (PST)
From: Internet-Drafts@ietf.org
To: i-d-announce@ietf.org
Content-Type: Multipart/Mixed; Boundary="NextPart"
Mime-Version: 1.0
Message-Id: <20100308183002.7F4E33A6A9E@core3.amsl.com>
Date: Mon,  8 Mar 2010 10:30:02 -0800 (PST)
Cc: ipfix@ietf.org
Subject: [IPFIX] I-D ACTION:draft-ietf-ipfix-structured-data-01.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 08 Mar 2010 18:30:02 -0000

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts 
directories.
This draft is a work item of the IP Flow Information Export Working Group of the IETF.

	Title		: Export of Structured Data in IPFIX
	Author(s)	: B. Claise, G. Dhandapani, S. Yates, P. Aitken
	Filename	: draft-ietf-ipfix-structured-data-01.txt
	Pages		: 58
	Date		: 2010-3-8
	
This document specifies an extension to IP Flow Information 
        eXport (IPFIX) protocol specification in [RFC5101] and the IPFIX 
        information model specified in [RFC5102] to support hierarchical 
        structured data and lists (sequences) of Information Elements in 
        data records.  This extension allows definition of complex data 
        structures such as variable-length lists and specification of 
        hierarchical containment relationships between Templates.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-ipfix-structured-data-01.txt

Internet-Drafts are also available by anonymous FTP at:
ftp://ftp.ietf.org/internet-drafts/

Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

--NextPart
Content-Type: Message/External-body;
	name="draft-ietf-ipfix-structured-data-01.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID: <2010-3-8102534.I-D@ietf.org>


--NextPart--


From paitken@cisco.com  Mon Mar  8 14:42:59 2010
Return-Path: <paitken@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id AF2343A6A06 for <ipfix@core3.amsl.com>; Mon,  8 Mar 2010 14:42:59 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -3.632
X-Spam-Level: 
X-Spam-Status: No, score=-3.632 tagged_above=-999 required=5 tests=[AWL=-1.633, BAYES_00=-2.599, J_CHICKENPOX_13=0.6]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 2JL6qElEkRug for <ipfix@core3.amsl.com>; Mon,  8 Mar 2010 14:42:54 -0800 (PST)
Received: from ams-iport-2.cisco.com (ams-iport-2.cisco.com [144.254.224.141]) by core3.amsl.com (Postfix) with ESMTP id E4DBC3A68E7 for <ipfix@ietf.org>; Mon,  8 Mar 2010 14:42:53 -0800 (PST)
Authentication-Results: ams-iport-2.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: Au8AALoMlUuQ/uCWe2dsb2JhbACbLxUBAQsLJAYcoWKYNYJUgiQEjjs
X-IronPort-AV: E=Sophos;i="4.49,604,1262563200";  d="scan'208";a="4103549"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-2.cisco.com with ESMTP; 08 Mar 2010 22:09:52 +0000
Received: from [10.61.68.3] (ams3-vpn-dhcp1027.cisco.com [10.61.68.3]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o28MguUF011714; Mon, 8 Mar 2010 22:42:56 GMT
Message-ID: <4B957D70.5020305@cisco.com>
Date: Mon, 08 Mar 2010 22:42:56 +0000
From: Paul Aitken <paitken@cisco.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-GB; rv:1.8.1.22) Gecko/20090605 SeaMonkey/1.1.17 (Ubuntu-1.1.17+nobinonly-0ubuntu0.9.04.1)
MIME-Version: 1.0
To: Nevil Brownlee <n.brownlee@auckland.ac.nz>
References: <AE36820147909644AD2A7CA014B1FB520A042922@xmb-sjc-222.amer.cisco.com>	<20100218135814.C821.1AB7FA03@nttv6.net> <4B945D21.3040804@auckland.ac.nz>
In-Reply-To: <4B945D21.3040804@auckland.ac.nz>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Comments on draft-kashima-ipfix-data-link-layer-monitoring-01.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 08 Mar 2010 22:42:59 -0000

Nevil,

> This draft introuces three new IPFIX Information Elements (IEs)
> that would enable IPFIX to be used for monitoring (parts of) in
> data-link-layer parts of packet headers.  It concentrates on Ethernet,
> especially as used in providing layer-2 VPN services (IEEE 802.1ad
> and 802.1ah).
> 
> Now that I've read it and thought about it, I have three comments:
> 
> 1. The IPFIX/PSAMP architecture standardises the collection of
>    *IP-related* data, this proposal would extend it to cover data from
>    other parts of a packet's headers.  That would be a significant
>    extension.  Is there support for such a generalisation?

Yes, very much so. We have customer need for this. It would be far 
preferable for us to standardise through IPFIX than to go down a 
proprietary route.

Note that there is precedent for this, since the IPFIX info model 
already contains some non-IP fields - eg, MPLS, MAC address, VLAN.


> 2. It's presented as a simple, minimal extension to IPFIX/PSAMP,
>    simply adding three new IEs PSAMP IEs, i.e with numbers above 338.
>    Does it provide enough detail?

Actually, the dataLinkFrameSize and dataLinkFrameSection definitions are 
simply verbatim copies from the old draft-ietf-psamp-info-09.


>    Should there be more than just three?
>    For example, how about having separate IEs for S-VID, C-VID, B-VID
>    and I=SID?  What other IEs are needed for PPP and WiMAX?
> 
> 3. The description of proposed IE dataLinkFrameType seems rather
>    minimal - eventually we'll want to add more data link types.  For
>    example, the ifType-MIB (www.iana.org/assignments/ianaiftype-mib)
>    has a much longer list.  Maybe we have to choose here between a long
>    list of types, each with its own set of fields, or a short list of
>    very general fields?  A short, simple list seems very appealing!

I would propose an extensible IANA sub-registry.

Cheers,
P.

-- 
Paul Aitken
Cisco Systems Ltd, Edinburgh, Scotland.

From kashima@nttv6.net  Tue Mar  9 08:22:19 2010
Return-Path: <kashima@nttv6.net>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 4CDB73A69B2 for <ipfix@core3.amsl.com>; Tue,  9 Mar 2010 08:22:17 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.999
X-Spam-Level: 
X-Spam-Status: No, score=-1.999 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, J_CHICKENPOX_13=0.6]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id OnzoX3UrdtYr for <ipfix@core3.amsl.com>; Tue,  9 Mar 2010 08:22:02 -0800 (PST)
Received: from mail.nttv6.net (mail.nttv6.net [192.16.178.5]) by core3.amsl.com (Postfix) with ESMTP id 37D743A6968 for <ipfix@ietf.org>; Tue,  9 Mar 2010 08:22:02 -0800 (PST)
Received: from [127.0.0.1] (localhost [IPv6:::1]) by mail.nttv6.net (8.14.3/8.14.3) with ESMTP id o29GKUwb074378; Wed, 10 Mar 2010 01:20:45 +0900 (JST) (envelope-from kashima@nttv6.net)
Date: Wed, 10 Mar 2010 01:20:11 +0900
From: Shingo KASHIMA <kashima@nttv6.net>
To: Nevil Brownlee <n.brownlee@auckland.ac.nz>
In-Reply-To: <4B945D21.3040804@auckland.ac.nz>
References: <20100218135814.C821.1AB7FA03@nttv6.net> <4B945D21.3040804@auckland.ac.nz>
Message-Id: <20100309224354.E4EB.1AB7FA03@nttv6.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
X-Mailer: Becky! ver. 2.50.05 [ja]
X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.2.3 (mail.nttv6.net [IPv6:::1]); Wed, 10 Mar 2010 01:20:45 +0900 (JST)
Cc: IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Comments on draft-kashima-ipfix-data-link-layer-monitoring-01.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 09 Mar 2010 16:22:19 -0000

Hi Nevil and all,

Thank you for your comments.

Please see our comments in-line.

> This draft introuces three new IPFIX Information Elements (IEs)
> that would enable IPFIX to be used for monitoring (parts of) in
> data-link-layer parts of packet headers.  It concentrates on Ethernet,
> especially as used in providing layer-2 VPN services (IEEE 802.1ad
> and 802.1ah).

Yes. 

Our interest is not only layer-2 VPN service (Wide-Area Ethernet)
but also cloud service (Data Center Network).
In the newest draft, we added IEEE802.1Qbh as the example of
the extended data link technologies.
http://tools.ietf.org/html/draft-kashima-ipfix-data-link-layer-monitoring-02
"2.2.  Data Center Network Summary"

What we want to say is that this draft can be also applied to
the other data link layer protocols than IEEE 802.1ad/IEEE802.1ah.


> 
> Now that I've read it and thought about it, I have three comments:
> 
> 1. The IPFIX/PSAMP architecture standardises the collection of
>     *IP-related* data, this proposal would extend it to cover data from
>     other parts of a packet's headers.  That would be a significant
>     extension.  Is there support for such a generalisation?

We think the data link layer protocols such as ethernet will be
extended and introduced to various networks, then traffic monitoring
in data link layer becomes more significant.
Though IPFIX/PSAMP is a standard for *IP-related* protocol,
IPFIX/PSAMP architecture is effective and attractive for our large-scale
ethernet networks, e.g. mediation and variouse packet selection.


> 
> 2. It's presented as a simple, minimal extension to IPFIX/PSAMP,
>     simply adding three new IEs PSAMP IEs, i.e with numbers above 338.
>     Does it provide enough detail?  Should there be more than just three?
>     For example, how about having separate IEs for S-VID, C-VID, B-VID
>     and I=SID?  What other IEs are needed for PPP and WiMAX?

I think separated IEs is not bad approach.

In IEEE802.1ah, we need not only {B-VID and I-SID}
but also {B-PCP and I-PCP}*1 at least .
*1{B-PCP and I-PCP}:
  CoS (Class of Service) parameters in {B-TAG and I-TAG}.

In IEEE802.1ad, almost of all header informations have been already
standardized.
- S-VID (=dot1qVlanId)
- S-PCP (=dot1qPriority)
- C-VID (=dot1qCustomerVlanId)
- C-PCP (=dot1qCustomerPriority)


> 
> 3. The description of proposed IE dataLinkFrameType seems rather
>     minimal - eventually we'll want to add more data link types.  For
>     example, the ifType-MIB (www.iana.org/assignments/ianaiftype-mib)
>     has a much longer list.  Maybe we have to choose here between a long
>     list of types, each with its own set of fields, or a short list of
>     very general fields?  A short, simple list seems very appealing!
> 

In the newest draft, we have revised as follows.
"      This Information Element specifies the type of the selected data
       link frame and SHOULD be checked before analyzing higher layer
       protocols.
       The data link layer is defined in [ISO_IEC.7498-1_1994].
       The value matches the value of managed object 'IANAifType' as
       defined in [RFC2863]."

But we have one doubt.
IANAifType means the type of interface, does not mean the type of frame
in a narrow sense.
In particular, IANAifType=ethernetCsmacd(6) means ethernet interfaces,
does not mean ethernet frames.

On the other hand, another registration which differs from IANAifType
takes long time until the standardization finishes.


Best regards.
-- 
Shingo KASHIMA <kashima@nttv6.net>
NTT Information Sharing Platform Lab.



From Quittek@nw.neclab.eu  Tue Mar  9 08:24:18 2010
Return-Path: <Quittek@nw.neclab.eu>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 07E933A6810 for <ipfix@core3.amsl.com>; Tue,  9 Mar 2010 08:24:18 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.699
X-Spam-Level: 
X-Spam-Status: No, score=-0.699 tagged_above=-999 required=5 tests=[AWL=1.900,  BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id GfkdOpESfgoF for <ipfix@core3.amsl.com>; Tue,  9 Mar 2010 08:24:17 -0800 (PST)
Received: from smtp0.neclab.eu (smtp0.neclab.eu [195.37.70.41]) by core3.amsl.com (Postfix) with ESMTP id C62913A6781 for <ipfix@ietf.org>; Tue,  9 Mar 2010 08:24:16 -0800 (PST)
Received: from localhost (localhost.localdomain [127.0.0.1]) by smtp0.neclab.eu (Postfix) with ESMTP id 0DBA12C020505; Tue,  9 Mar 2010 17:24:21 +0100 (CET)
X-Virus-Scanned: Amavisd on Debian GNU/Linux (atlas2.office)
Received: from smtp0.neclab.eu ([127.0.0.1]) by localhost (atlas2.office [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id eiPuUOFdTRnp; Tue,  9 Mar 2010 17:24:20 +0100 (CET)
Received: from VENUS.office (mx1.office [192.168.24.3]) by smtp0.neclab.eu (Postfix) with ESMTP id E1A142C000352; Tue,  9 Mar 2010 17:24:10 +0100 (CET)
X-MimeOLE: Produced By Microsoft Exchange V6.5
Received: from 10.7.0.54 ([10.7.0.54]) by VENUS.office ([192.168.24.102]) with Microsoft Exchange Server HTTP-DAV ; Tue,  9 Mar 2010 16:24:10 +0000
MIME-Version: 1.0
Content-class: urn:content-classes:message
Content-Type: multipart/signed; protocol="application/pkcs7-signature"; micalg=sha1; boundary="B_3351000243_8811297"
Date: Tue, 9 Mar 2010 17:24:03 +0100
Message-ID: <C7BC34B3.77A3F%Quittek@nw.neclab.eu>
In-Reply-To: <4B94D635.3020200@net.in.tum.de>
X-MS-Has-Attach: yes
X-MS-TNEF-Correlator: 
Thread-Topic: [IPFIX] draft-ietf-ipfix-configuration-model-05
Thread-Index: Acq/pO56nmcSeXm5aUq8sBp2Owho2Q==
From: "Juergen Quittek" <Quittek@nw.neclab.eu>
To: "Gerhard Muenz" <muenz@net.in.tum.de>, "IETF IPFIX Working Group" <ipfix@ietf.org>
Subject: Re: [IPFIX] draft-ietf-ipfix-configuration-model-05
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 09 Mar 2010 16:24:18 -0000

> This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

--B_3351000243_8811297
Content-type: text/plain;
	charset="US-ASCII"
Content-transfer-encoding: 7bit

Hi Gerhard,

I appreciate the progress.

Are you still planning a presentation in Anaheim?
Or can we take the draft off the agenda?

Thanks,

    Juergen

On 08.03.10 11:49  "Gerhard Muenz" <muenz@net.in.tum.de> wrote:

> 
> Dear all,
> 
> I've posted draft-ietf-ipfix-configuration-model-05 which shows the
> current status of the draft.
> 
> Thanks a lot to the reviewers!
> 
> WGLC comments of Atsushi, Lothar, and Brian have been addressed.
> 
> Some issues raised by Andrew still need to be solved.
> 
> Since I will not go to Anaheim, the mailing list is the right place to
> continue the discussion. I hope that the remaining issues can be solved
> until the IPFIX session. If so, a -06 version will follow very soon.
> 
> Regards,
> Gerhard
> 
> 
> 
> 
> 
> _______________________________________________
> IPFIX mailing list
> IPFIX@ietf.org
> https://www.ietf.org/mailman/listinfo/ipfix

--B_3351000243_8811297
Content-type: application/pkcs7-signature; name="smime.p7s"
Content-transfer-encoding: base64
Content-disposition: attachment;
	filename="smime.p7s"
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--B_3351000243_8811297--


From n.brownlee@auckland.ac.nz  Tue Mar  9 13:26:10 2010
Return-Path: <n.brownlee@auckland.ac.nz>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 791743A69C2 for <ipfix@core3.amsl.com>; Tue,  9 Mar 2010 13:26:10 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.524
X-Spam-Level: 
X-Spam-Status: No, score=-6.524 tagged_above=-999 required=5 tests=[AWL=0.075,  BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id ZBoPtWqq2qPj for <ipfix@core3.amsl.com>; Tue,  9 Mar 2010 13:26:09 -0800 (PST)
Received: from mailhost.auckland.ac.nz (larry.its.auckland.ac.nz [130.216.12.34]) by core3.amsl.com (Postfix) with ESMTP id 9D8FD3A69B5 for <ipfix@ietf.org>; Tue,  9 Mar 2010 13:26:09 -0800 (PST)
Received: from localhost (localhost.localdomain [127.0.0.1]) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 36E291B8AA for <ipfix@ietf.org>; Wed, 10 Mar 2010 10:26:10 +1300 (NZDT)
X-Virus-Scanned: by amavisd-new at mailhost.auckland.ac.nz
Received: from mailhost.auckland.ac.nz ([127.0.0.1]) by localhost (larry.its.auckland.ac.nz [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id X9dbYU8YSBEX for <ipfix@ietf.org>; Wed, 10 Mar 2010 10:26:10 +1300 (NZDT)
Received: from nevil-laptop.sfac.auckland.ac.nz (nevil-laptop.sfac.auckland.ac.nz [130.216.38.130]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 184481B8A7 for <ipfix@ietf.org>; Wed, 10 Mar 2010 10:26:07 +1300 (NZDT)
Message-ID: <4B96BCEF.3020906@auckland.ac.nz>
Date: Wed, 10 Mar 2010 10:26:07 +1300
From: Nevil Brownlee <n.brownlee@auckland.ac.nz>
User-Agent: Thunderbird 2.0.0.12 (Macintosh/20080213)
MIME-Version: 1.0
To: IPFIX Working Group <ipfix@ietf.org>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Subject: [IPFIX] Agenda for IPFIX meeting at Anaheim IETF
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 09 Mar 2010 21:26:10 -0000

Hi all:

The latest update of our meeting agenda is now at
   http://www.ietf.org/proceedings/10mar/agenda/ipfix.txt

Would everyone who's presenting please email me your slides before
Monday evening, 22 March, so that I can get them on the Meeting
Materials web site.

See you in Anaheim, Nevil

-- 
---------------------------------------------------------------------
  Nevil Brownlee                    Computer Science Department | ITS
  Phone: +64 9 373 7599 x88941             The University of Auckland
  FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand

From dharani@cisco.com  Tue Mar  9 14:17:27 2010
Return-Path: <dharani@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 852C13A6AA9 for <ipfix@core3.amsl.com>; Tue,  9 Mar 2010 14:17:27 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -9.999
X-Spam-Level: 
X-Spam-Status: No, score=-9.999 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, J_CHICKENPOX_13=0.6, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id CUFuJtlvvUd6 for <ipfix@core3.amsl.com>; Tue,  9 Mar 2010 14:17:25 -0800 (PST)
Received: from sj-iport-2.cisco.com (sj-iport-2.cisco.com [171.71.176.71]) by core3.amsl.com (Postfix) with ESMTP id 6C6E33A6AA3 for <ipfix@ietf.org>; Tue,  9 Mar 2010 14:17:24 -0800 (PST)
Authentication-Results: sj-iport-2.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-AV: E=Sophos;i="4.49,610,1262563200"; d="scan'208";a="245496738"
Received: from sj-core-5.cisco.com ([171.71.177.238]) by sj-iport-2.cisco.com with ESMTP; 09 Mar 2010 22:17:28 +0000
Received: from xbh-sjc-231.amer.cisco.com (xbh-sjc-231.cisco.com [128.107.191.100]) by sj-core-5.cisco.com (8.13.8/8.14.3) with ESMTP id o29MHO32028847; Tue, 9 Mar 2010 22:17:25 GMT
Received: from xmb-sjc-222.amer.cisco.com ([128.107.191.106]) by xbh-sjc-231.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.3959);  Tue, 9 Mar 2010 14:17:24 -0800
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Date: Tue, 9 Mar 2010 14:17:18 -0800
Message-ID: <AE36820147909644AD2A7CA014B1FB520A3206E4@xmb-sjc-222.amer.cisco.com>
In-Reply-To: <4B957D70.5020305@cisco.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [IPFIX] Comments ondraft-kashima-ipfix-data-link-layer-monitoring-01.txt
Thread-Index: Acq/ELnNk/Po3KhJTQupe5n5jCeEJwAxIqDQ
References: <AE36820147909644AD2A7CA014B1FB520A042922@xmb-sjc-222.amer.cisco.com>	<20100218135814.C821.1AB7FA03@nttv6.net><4B945D21.3040804@auckland.ac.nz> <4B957D70.5020305@cisco.com>
From: "Dharani Vilwanathan (dharani)" <dharani@cisco.com>
To: "Paul Aitken (paitken)" <paitken@cisco.com>, "Nevil Brownlee" <n.brownlee@auckland.ac.nz>
X-OriginalArrivalTime: 09 Mar 2010 22:17:24.0780 (UTC) FILETIME=[4BB8F6C0:01CABFD6]
Cc: IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Comments ondraft-kashima-ipfix-data-link-layer-monitoring-01.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 09 Mar 2010 22:17:27 -0000

Hi Nevil/Paul,

I too would like to convey that there is a need for obtaining layer 2
info in IPFIX records along with rest of the IP fields of a flow.

We would like to make sure this will be of interest to IPFIX working
group and community, because Yaonan, Laxmi and I are working on various
ethernet types and see if we can address missing fields. We would like
to bring this to IPFIX forum once we have the details.=20

Thanks
dharani

-----Original Message-----
From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] On Behalf
Of Paul Aitken (paitken)
Sent: Monday, March 08, 2010 2:43 PM
To: Nevil Brownlee
Cc: IPFIX Working Group
Subject: Re: [IPFIX] Comments
ondraft-kashima-ipfix-data-link-layer-monitoring-01.txt

Nevil,

> This draft introuces three new IPFIX Information Elements (IEs) that=20
> would enable IPFIX to be used for monitoring (parts of) in=20
> data-link-layer parts of packet headers.  It concentrates on Ethernet,

> especially as used in providing layer-2 VPN services (IEEE 802.1ad and

> 802.1ah).
>=20
> Now that I've read it and thought about it, I have three comments:
>=20
> 1. The IPFIX/PSAMP architecture standardises the collection of
>    *IP-related* data, this proposal would extend it to cover data from
>    other parts of a packet's headers.  That would be a significant
>    extension.  Is there support for such a generalisation?

Yes, very much so. We have customer need for this. It would be far
preferable for us to standardise through IPFIX than to go down a
proprietary route.

Note that there is precedent for this, since the IPFIX info model
already contains some non-IP fields - eg, MPLS, MAC address, VLAN.


> 2. It's presented as a simple, minimal extension to IPFIX/PSAMP,
>    simply adding three new IEs PSAMP IEs, i.e with numbers above 338.
>    Does it provide enough detail?

Actually, the dataLinkFrameSize and dataLinkFrameSection definitions are

simply verbatim copies from the old draft-ietf-psamp-info-09.


>    Should there be more than just three?
>    For example, how about having separate IEs for S-VID, C-VID, B-VID
>    and I=3DSID?  What other IEs are needed for PPP and WiMAX?
>=20
> 3. The description of proposed IE dataLinkFrameType seems rather
>    minimal - eventually we'll want to add more data link types.  For
>    example, the ifType-MIB (www.iana.org/assignments/ianaiftype-mib)
>    has a much longer list.  Maybe we have to choose here between a
long
>    list of types, each with its own set of fields, or a short list of
>    very general fields?  A short, simple list seems very appealing!

I would propose an extensible IANA sub-registry.

Cheers,
P.

--=20
Paul Aitken
Cisco Systems Ltd, Edinburgh, Scotland.
_______________________________________________
IPFIX mailing list
IPFIX@ietf.org
https://www.ietf.org/mailman/listinfo/ipfix

From n.brownlee@auckland.ac.nz  Tue Mar  9 14:24:13 2010
Return-Path: <n.brownlee@auckland.ac.nz>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 2C5953A6AAF for <ipfix@core3.amsl.com>; Tue,  9 Mar 2010 14:24:13 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.239
X-Spam-Level: 
X-Spam-Status: No, score=-6.239 tagged_above=-999 required=5 tests=[AWL=-0.240, BAYES_00=-2.599, J_CHICKENPOX_13=0.6, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 23apyf8L036X for <ipfix@core3.amsl.com>; Tue,  9 Mar 2010 14:24:11 -0800 (PST)
Received: from mailhost.auckland.ac.nz (larry.its.auckland.ac.nz [130.216.12.34]) by core3.amsl.com (Postfix) with ESMTP id 52CFB3A6A99 for <ipfix@ietf.org>; Tue,  9 Mar 2010 14:24:11 -0800 (PST)
Received: from localhost (localhost.localdomain [127.0.0.1]) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 7CCD5416B26; Wed, 10 Mar 2010 11:24:14 +1300 (NZDT)
X-Virus-Scanned: by amavisd-new at mailhost.auckland.ac.nz
Received: from mailhost.auckland.ac.nz ([127.0.0.1]) by localhost (larry.its.auckland.ac.nz [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id RBH1pGH+AgmV; Wed, 10 Mar 2010 11:24:14 +1300 (NZDT)
Received: from nevil-laptop.sfac.auckland.ac.nz (nevil-laptop.sfac.auckland.ac.nz [130.216.38.130]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 12DD94143F3; Wed, 10 Mar 2010 11:24:14 +1300 (NZDT)
Message-ID: <4B96CA8E.6060700@auckland.ac.nz>
Date: Wed, 10 Mar 2010 11:24:14 +1300
From: Nevil Brownlee <n.brownlee@auckland.ac.nz>
User-Agent: Thunderbird 2.0.0.12 (Macintosh/20080213)
MIME-Version: 1.0
To: "Dharani Vilwanathan (dharani)" <dharani@cisco.com>
References: <AE36820147909644AD2A7CA014B1FB520A042922@xmb-sjc-222.amer.cisco.com>	<20100218135814.C821.1AB7FA03@nttv6.net><4B945D21.3040804@auckland.ac.nz> <4B957D70.5020305@cisco.com> <AE36820147909644AD2A7CA014B1FB520A3206E4@xmb-sjc-222.amer.cisco.com>
In-Reply-To: <AE36820147909644AD2A7CA014B1FB520A3206E4@xmb-sjc-222.amer.cisco.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Comments ondraft-kashima-ipfix-data-link-layer-monitoring-01.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 09 Mar 2010 22:24:13 -0000

Hi Dharani:

I agree that being able to report on Layer 2 is important and useful -
we just need to work out the best strategy to document/standardise it.

Cheers, Nevil


Dharani Vilwanathan (dharani) wrote:
> Hi Nevil/Paul,
> 
> I too would like to convey that there is a need for obtaining layer 2
> info in IPFIX records along with rest of the IP fields of a flow.
> 
> We would like to make sure this will be of interest to IPFIX working
> group and community, because Yaonan, Laxmi and I are working on various
> ethernet types and see if we can address missing fields. We would like
> to bring this to IPFIX forum once we have the details. 
> 
> Thanks
> dharani
> 
> -----Original Message-----
> From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] On Behalf
> Of Paul Aitken (paitken)
> Sent: Monday, March 08, 2010 2:43 PM
> To: Nevil Brownlee
> Cc: IPFIX Working Group
> Subject: Re: [IPFIX] Comments
> ondraft-kashima-ipfix-data-link-layer-monitoring-01.txt
> 
> Nevil,
> 
>> This draft introuces three new IPFIX Information Elements (IEs) that 
>> would enable IPFIX to be used for monitoring (parts of) in 
>> data-link-layer parts of packet headers.  It concentrates on Ethernet,
> 
>> especially as used in providing layer-2 VPN services (IEEE 802.1ad and
> 
>> 802.1ah).
>>
>> Now that I've read it and thought about it, I have three comments:
>>
>> 1. The IPFIX/PSAMP architecture standardises the collection of
>>    *IP-related* data, this proposal would extend it to cover data from
>>    other parts of a packet's headers.  That would be a significant
>>    extension.  Is there support for such a generalisation?
> 
> Yes, very much so. We have customer need for this. It would be far
> preferable for us to standardise through IPFIX than to go down a
> proprietary route.
> 
> Note that there is precedent for this, since the IPFIX info model
> already contains some non-IP fields - eg, MPLS, MAC address, VLAN.
> 
> 
>> 2. It's presented as a simple, minimal extension to IPFIX/PSAMP,
>>    simply adding three new IEs PSAMP IEs, i.e with numbers above 338.
>>    Does it provide enough detail?
> 
> Actually, the dataLinkFrameSize and dataLinkFrameSection definitions are
> 
> simply verbatim copies from the old draft-ietf-psamp-info-09.
> 
> 
>>    Should there be more than just three?
>>    For example, how about having separate IEs for S-VID, C-VID, B-VID
>>    and I=SID?  What other IEs are needed for PPP and WiMAX?
>>
>> 3. The description of proposed IE dataLinkFrameType seems rather
>>    minimal - eventually we'll want to add more data link types.  For
>>    example, the ifType-MIB (www.iana.org/assignments/ianaiftype-mib)
>>    has a much longer list.  Maybe we have to choose here between a
> long
>>    list of types, each with its own set of fields, or a short list of
>>    very general fields?  A short, simple list seems very appealing!
> 
> I would propose an extensible IANA sub-registry.
> 
> Cheers,
> P.
> 


-- 
---------------------------------------------------------------------
  Nevil Brownlee                    Computer Science Department | ITS
  Phone: +64 9 373 7599 x88941             The University of Auckland
  FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand

From dharani@cisco.com  Tue Mar  9 14:39:46 2010
Return-Path: <dharani@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 339043A6AC5 for <ipfix@core3.amsl.com>; Tue,  9 Mar 2010 14:39:46 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -9.999
X-Spam-Level: 
X-Spam-Status: No, score=-9.999 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, J_CHICKENPOX_13=0.6, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id nhttSsmVukJZ for <ipfix@core3.amsl.com>; Tue,  9 Mar 2010 14:39:45 -0800 (PST)
Received: from sj-iport-5.cisco.com (sj-iport-5.cisco.com [171.68.10.87]) by core3.amsl.com (Postfix) with ESMTP id 358203A6AB6 for <ipfix@ietf.org>; Tue,  9 Mar 2010 14:39:45 -0800 (PST)
Authentication-Results: sj-iport-5.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-AV: E=Sophos;i="4.49,610,1262563200"; d="scan'208";a="163179640"
Received: from sj-core-4.cisco.com ([171.68.223.138]) by sj-iport-5.cisco.com with ESMTP; 09 Mar 2010 22:39:50 +0000
Received: from xbh-sjc-231.amer.cisco.com (xbh-sjc-231.cisco.com [128.107.191.100]) by sj-core-4.cisco.com (8.13.8/8.14.3) with ESMTP id o29MdoUV001963; Tue, 9 Mar 2010 22:39:50 GMT
Received: from xmb-sjc-222.amer.cisco.com ([128.107.191.106]) by xbh-sjc-231.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.3959);  Tue, 9 Mar 2010 14:39:49 -0800
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Date: Tue, 9 Mar 2010 14:39:28 -0800
Message-ID: <AE36820147909644AD2A7CA014B1FB520A320714@xmb-sjc-222.amer.cisco.com>
In-Reply-To: <4B96CA8E.6060700@auckland.ac.nz>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [IPFIX] Comments ondraft-kashima-ipfix-data-link-layer-monitoring-01.txt
Thread-Index: Acq/10l4REGhwuvuRkWDcYXYn3rgngAAfPUw
References: <AE36820147909644AD2A7CA014B1FB520A042922@xmb-sjc-222.amer.cisco.com>	<20100218135814.C821.1AB7FA03@nttv6.net><4B945D21.3040804@auckland.ac.nz> <4B957D70.5020305@cisco.com> <AE36820147909644AD2A7CA014B1FB520A3206E4@xmb-sjc-222.amer.cisco.com> <4B96CA8E.6060700@auckland.ac.nz>
From: "Dharani Vilwanathan (dharani)" <dharani@cisco.com>
To: "Nevil Brownlee" <n.brownlee@auckland.ac.nz>
X-OriginalArrivalTime: 09 Mar 2010 22:39:49.0997 (UTC) FILETIME=[6D88C9D0:01CABFD9]
Cc: IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Comments ondraft-kashima-ipfix-data-link-layer-monitoring-01.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 09 Mar 2010 22:39:46 -0000

Hi Nevil,

Thanks for the quick response and thanks for sharing your thoughts.

Regards
dharani=20

-----Original Message-----
From: Nevil Brownlee [mailto:n.brownlee@auckland.ac.nz]=20
Sent: Tuesday, March 09, 2010 2:24 PM
To: Dharani Vilwanathan (dharani)
Cc: Paul Aitken (paitken); IPFIX Working Group
Subject: Re: [IPFIX] Comments
ondraft-kashima-ipfix-data-link-layer-monitoring-01.txt


Hi Dharani:

I agree that being able to report on Layer 2 is important and useful -
we just need to work out the best strategy to document/standardise it.

Cheers, Nevil


Dharani Vilwanathan (dharani) wrote:
> Hi Nevil/Paul,
>=20
> I too would like to convey that there is a need for obtaining layer 2=20
> info in IPFIX records along with rest of the IP fields of a flow.
>=20
> We would like to make sure this will be of interest to IPFIX working=20
> group and community, because Yaonan, Laxmi and I are working on=20
> various ethernet types and see if we can address missing fields. We=20
> would like to bring this to IPFIX forum once we have the details.
>=20
> Thanks
> dharani
>=20
> -----Original Message-----
> From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] On Behalf

> Of Paul Aitken (paitken)
> Sent: Monday, March 08, 2010 2:43 PM
> To: Nevil Brownlee
> Cc: IPFIX Working Group
> Subject: Re: [IPFIX] Comments
> ondraft-kashima-ipfix-data-link-layer-monitoring-01.txt
>=20
> Nevil,
>=20
>> This draft introuces three new IPFIX Information Elements (IEs) that=20
>> would enable IPFIX to be used for monitoring (parts of) in=20
>> data-link-layer parts of packet headers.  It concentrates on=20
>> Ethernet,
>=20
>> especially as used in providing layer-2 VPN services (IEEE 802.1ad=20
>> and
>=20
>> 802.1ah).
>>
>> Now that I've read it and thought about it, I have three comments:
>>
>> 1. The IPFIX/PSAMP architecture standardises the collection of
>>    *IP-related* data, this proposal would extend it to cover data
from
>>    other parts of a packet's headers.  That would be a significant
>>    extension.  Is there support for such a generalisation?
>=20
> Yes, very much so. We have customer need for this. It would be far=20
> preferable for us to standardise through IPFIX than to go down a=20
> proprietary route.
>=20
> Note that there is precedent for this, since the IPFIX info model=20
> already contains some non-IP fields - eg, MPLS, MAC address, VLAN.
>=20
>=20
>> 2. It's presented as a simple, minimal extension to IPFIX/PSAMP,
>>    simply adding three new IEs PSAMP IEs, i.e with numbers above 338.
>>    Does it provide enough detail?
>=20
> Actually, the dataLinkFrameSize and dataLinkFrameSection definitions=20
> are
>=20
> simply verbatim copies from the old draft-ietf-psamp-info-09.
>=20
>=20
>>    Should there be more than just three?
>>    For example, how about having separate IEs for S-VID, C-VID, B-VID
>>    and I=3DSID?  What other IEs are needed for PPP and WiMAX?
>>
>> 3. The description of proposed IE dataLinkFrameType seems rather
>>    minimal - eventually we'll want to add more data link types.  For
>>    example, the ifType-MIB (www.iana.org/assignments/ianaiftype-mib)
>>    has a much longer list.  Maybe we have to choose here between a
> long
>>    list of types, each with its own set of fields, or a short list of
>>    very general fields?  A short, simple list seems very appealing!
>=20
> I would propose an extensible IANA sub-registry.
>=20
> Cheers,
> P.
>=20


--
---------------------------------------------------------------------
  Nevil Brownlee                    Computer Science Department | ITS
  Phone: +64 9 373 7599 x88941             The University of Auckland
  FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand

From muenz@net.in.tum.de  Wed Mar 10 02:14:47 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 9833B3A6851 for <ipfix@core3.amsl.com>; Wed, 10 Mar 2010 02:14:47 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id XDH-RquzqAyb for <ipfix@core3.amsl.com>; Wed, 10 Mar 2010 02:14:46 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 29CE43A67A5 for <ipfix@ietf.org>; Wed, 10 Mar 2010 02:14:45 -0800 (PST)
Received: from [131.159.14.149] (habanero.net.in.tum.de [131.159.14.149]) by mail.net.in.tum.de (Postfix) with ESMTPSA id CB1092057C6F; Wed, 10 Mar 2010 11:14:46 +0100 (CET)
Message-ID: <4B977127.2000903@net.in.tum.de>
Date: Wed, 10 Mar 2010 11:15:03 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Juergen Quittek <Quittek@nw.neclab.eu>
References: <C7BC34B3.77A3F%Quittek@nw.neclab.eu>
In-Reply-To: <C7BC34B3.77A3F%Quittek@nw.neclab.eu>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms090703090909060301090400"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] draft-ietf-ipfix-configuration-model-05
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 10 Mar 2010 10:14:47 -0000

This is a cryptographically signed message in MIME format.

--------------ms090703090909060301090400
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi J=FCrgen,

I assume that Benoit can give a short presentation as usual.

Regards,
Gerhard

Juergen Quittek wrote:
> Hi Gerhard,
>=20
> I appreciate the progress.
>=20
> Are you still planning a presentation in Anaheim?
> Or can we take the draft off the agenda?
>=20
> Thanks,
>=20
>     Juergen
>=20
> On 08.03.10 11:49  "Gerhard Muenz" <muenz@net.in.tum.de> wrote:
>=20
>> Dear all,
>>
>> I've posted draft-ietf-ipfix-configuration-model-05 which shows the
>> current status of the draft.
>>
>> Thanks a lot to the reviewers!
>>
>> WGLC comments of Atsushi, Lothar, and Brian have been addressed.
>>
>> Some issues raised by Andrew still need to be solved.
>>
>> Since I will not go to Anaheim, the mailing list is the right place to=

>> continue the discussion. I hope that the remaining issues can be solve=
d
>> until the IPFIX session. If so, a -06 version will follow very soon.
>>
>> Regards,
>> Gerhard
>>
>>
>>
>>
>>
>> _______________________________________________
>> IPFIX mailing list
>> IPFIX@ietf.org
>> https://www.ietf.org/mailman/listinfo/ipfix

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms090703090909060301090400
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms090703090909060301090400--

From muenz@net.in.tum.de  Thu Mar 11 02:28:57 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 8ED673A6A0E for <ipfix@core3.amsl.com>; Thu, 11 Mar 2010 02:28:57 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id t2wtfCrnxqFC for <ipfix@core3.amsl.com>; Thu, 11 Mar 2010 02:28:55 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 8C16B3A67B3 for <ipfix@ietf.org>; Thu, 11 Mar 2010 02:28:55 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 34DF220C3CB0; Thu, 11 Mar 2010 11:28:58 +0100 (CET)
Message-ID: <4B98C5FC.5030909@net.in.tum.de>
Date: Thu, 11 Mar 2010 11:29:16 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Andrew Johnson <andrjohn@cisco.com>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com> <4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com>
In-Reply-To: <4B92BA5F.1060601@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms080505070708060207060803"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 11 Mar 2010 10:28:57 -0000

This is a cryptographically signed message in MIME format.

--------------ms080505070708060207060803
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Andrew, all,

Please find my answer at the end of this e-mail.
There are text revisions to approve and a question from my side.

Andrew Johnson wrote:
> Gerhard Muenz wrote:
>> Andrew Johnson wrote:
>>> Gerhard Muenz wrote:
>>>>> I can see why it would be convenient to configure multiple caches t=
o
>>>>> be populated from a single Selection Process, and it could even be
>>>>> implemented that way (to some extent), but that is not the PSAMP mo=
del.
>>>>> The configuration model doesn't have to exactly match this model, b=
ut
>>>>> you should ensure that the differences are made clear.
>>>> The PSAMP model has no notion of Caches.
>>>>
>>>> Could you please show me the paragraphs in the PSAMP documents which=

>>>> would indirectly prevent the existence of more than one Cache per
>>>> Metering Process?
>>> I didn't say that you were PSAMP restricted you to one cache per
>>> Metering Process.  My point was about Selection Processes and my conc=
ern
>>> is over the statistics that PSAMP requires exported for each Selectio=
n
>>> Process.
>>>
>>> For example, imagine a simple selection process configuration that
>>> selects 1 in 10 packets.  That configuration could first be used with=

>>> Cache A:
>>>
>>>     Selection Process A1:
>>>       Observed packet stream count: 100
>>>       Selected packet stream count:  10
>>>
>>>     Cache A:
>>>       Total packets in: 10
>>>
>>>
>>> Now that same configuration can be used to populate another Cache lat=
er,
>>> but the statistics need kept separate for each Metering Process:
>>>
>>>     Selection Process A1:
>>>       Observed packet stream count: 130
>>>       Selected packet stream count:  13
>>>
>>>     Cache A:
>>>       Total packets in: 10
>>>
>>>
>>>     Selection Process B1:
>>>       Observed packet stream count:  30
>>>       Selected packet stream count:   3
>>>
>>>     Cache B:
>>>       Total packets in:  3
>>>
>>> PSAMP requires that these stats are kept and exported, which means th=
at
>>> the stats for A1 and B1 need to be exported under different Selection=

>>> Process IDs.  I guess the real restriction is that a new cache cannot=
 be
>>> populated from an existing Selection Process.  While a template
>>> configuration can be used it needs to spawn a separate instances so t=
hat
>>> the stats can kept separate.
>> This is a good example. I see it differently, though.
>>
>> In order to forward the output of the Selection Process to a second
>> Cache, you need to change the configuration of the Selection Process.
>> See how it looks in XML:
>>
>> On Cache:
>>
>> SP A1 -> Cache A
>>
>>   <selectionProcess>
>>     <name>A1</name>
>>     <selector>
>>       ...
>>     </selector>
>>     <cache>A</cache>
>>   </selectionProcess>
>>
>> After adding the second Cache:
>>
>>        +-> Cache A
>> SP A1 -+
>>        +-> Cache B
>>
>>   <selectionProcess>
>>     <name>A1</name>
>>     <selector>
>>       ...
>>     </selector>
>>     <cache>A</cache>
>>     <cache>B</cache>
>>   </selectionProcess>
>>
>> So, switching from the first configuration to the second one involves =
a
>> reconfiguration of the Selection Process A1.
>>
>> Regardless of how it is realized, the counters need to be reset becaus=
e
>> the configuration of the SP has changed.
>=20
> Well, it might not be clear that the stats need cleared because of the =

> nature of the change doesn't effect the selection behaviour of the=20
> Selection Process.
>=20
> What about this case:
>=20
>      SP A1 -> SP B1 -> Cache A
>=20
>    becomes
>=20
>                      +-> Cache A
>      SP A1 -> SP B1 -+
>                      +-> Cache B
>=20
> In this case SP A1 didn't have its configuration changed at all, yet it=
=20
> must have its stats reset for Cache B.  In fact, if you only have one=20
> set of stats for each SP, then adding a new Cache means resetting the=20
> stats of the entire tree.  i.e. if the above was part of a tree like=20
> this, then all every SP in the tree would need its stats reset when=20
> Cache B was added:
>=20
>                            +-> Cache A
>                  +-> SP Y -+
>                  |         +-> Cache C
>     OP1 -> SP Z -+
>                  |                   +-> Cache A
>                  +-> SP A1 -> SP B1 -+
>                                      +-> Cache B
>=20
> Since Cache B needs fresh stats, B1 needs reset, therefore A1 needs=20
> reset, therefore Z needs reset, which then means all its children need =

> reset.
>=20
>=20
>> Before the configuration of the SP is changed, it makes sense to expor=
t
>> all old counters. Otherwise, they will be lost.
>>
>> Probably, the cacheDataRecords counter of the Cache should also be res=
et
>> if the Cache is reconfigured.
>=20
> Is the cache being reconfigured or just populated from a different sour=
ce?
>=20
> You seem to be making assumptions here about how I need to manage my=20
> stats.  I can conceive of a PSAMP compliant implementation that can be =

> configured using your YANG model, that doesn't need to reset any stats =

> when a new cache is added to a Selection Sequence.
>=20
> It would be acceptable for each selection process in the chain to keep =

> separate stats for each cache that is feeds.  In fact the configuration=
=20
> would act as a template and a new instance of each Selector would be=20
> created for each use.
>=20
>=20
>> Should I add a sentence which says that packetsObserved, packetsDroppe=
d,
>> cacheDataRecords need to be reset upon reconfiguration?
>=20
> That would make sense.  However, my concern was over the stats that are=
=20
> exported by PSAMP, which is a different story.
>=20
>=20
>> BTW: There are several ways to implement a switch from one configurati=
on
>> to a new one. If your processes do not support reconfiguration at
>> runtime, you can replace them by newly instantiated processes. In this=

>> case, the counters will be reset, anyway.

So, your concerns are not related to the counters included in the=20
configuration data model, but about how PSAMP statistics would be=20
exported in Selection Sequence Statistics Reports.

We have discussed two possibilities to maintain the necessary statistics:=

- reset all statistics with every reconfiguration of the device
- maintain serapate statistics for every cache
We can conclude that the configuration data model does not collide with=20
RFC5476, right?

Nevertheless, I reworked the corresponding parts of the text with the=20
following objectives:
- no reference to parameter descriptions in the MIB (as proposed by
   Brian)
- make adaptations to the state parameters adopted from the MIB where it
   is useful
- recommend re-initialization of counters at reconfiguration

As an example, there exists YANG counter types which are set to zero at=20
initialization (in contrast to MIB counters which start with an=20
arbitrary value). I propose to use these counter types.

So, please comment the current text:


4.2.1.  Selector Class

     +---------------------------------------+
     | Selector                              |
     +---------------------------------------+      1 +-----------------+=

     | name                                  |<>------+ SelectAll/      |=

     | selectorId {readOnly}                 |        | SampCountBased/ |=

     | packetsObserved {readOnly}            |        | SampTimeBased/  |=

     | packetsDropped {readOnly}             |        | SampRandOutOfN/ |=

     | selectorInitializationTime {readOnly} |        | SampUniProb/    |=

     |                                       |        | FilterMatch/    |=

     |                                       |        | FilterHash/     |=

     +---------------------------------------+        +-----------------+=


                          Figure 8: Selector class

[snip]

    As state parameters, the Selector class also contains the Selector
    statistics packetsObserved and packetsDropped as well as
    selectorInitializationTime:

    packetsObserved:  The number of packets observed at the input of the
       Selector since the point in time indicated by
       selectorDiscontinuityTime.  In contrast to
       ipfixSelectionProcessStatsPacketsObserved in the IPFIX MIB module
       [I-D.ietf-ipfix-mib], the counter is initialized with zero.

    packetsDropped:  The number of packets discarded by the Selector
       since the point in time indicated by selectorDiscontinuityTime.
       In contrast to ipfixSelectionProcessStatsPacketsDropped in the
       IPFIX MIB module [I-D.ietf-ipfix-mib], the counter is initialized
       with zero.

    selectorInitializationTime:  The value of sysUpTime at the most
       recent occasion at which the Selector was re-initialized and at
       which the counters packetsObserved and packektsDropped were reset
       to zero.  In contrast to
       ipfixSelectionProcessStatsDiscontinuityTime in the IPFIX MIB
       module [I-D.ietf-ipfix-mib], the zero value does not have any
       specific meaning.

    The counter values of packetsObserved and packetsDropped MUST both be=

    set to zero at the time indicated by selectorInitializationTime.  It
    is RECOMMENDED to reset the counters after every reconfiguration of
    the Selector.

[snip]

4.3.  Cache Class

      +-------------------------------------+
      | Cache                               |
      +-------------------------------------+         1 +-------------+
      | name                                |<>---------| CacheLayout |
      | cacheMode                           |           +-------------+
      | maxRecords {opt.}                   |
      | activeTimeout {opt.} {Cache Modes   | 0..*
      |   "timeout" and "natural" only}     |---------------+
      | inactiveTimeout {opt.} {Cache Modes |               | 0..*
      |   "timeout" and "natural" only}     |               V
      | exportInterval {opt.} {Cache Mode   |      +------------------+
      |   "permanent" only}                 |      | ExportingProcess |
      | activeFlows {readOnly}              |      +------------------+
      | inactiveFlows {readOnly}            |
      | cacheDataRecords {readOnly}         |
      | cacheInitializationTime {readOnly}  |
      +-------------------------------------+

                           Figure 11: Cache class

[snip]

    As state parameters, the Cache class contains the Metering Process
    statistics activeFlows, inactiveFlows, and cacheDataRecords, as well
    as cacheInitializationTime:

    activeFlows:  The number of Flows currently active in this cache.
       This value corresponds to ipfixMeteringProcessCacheActiveFlows in
       the IPFIX MIB module [I-D.ietf-ipfix-mib].

    inactiveFlows:  The number of Flows currently inactive in this cache.=

       This value corresponds to ipfixMeteringProcessCacheInactiveFlows
       in the IPFIX MIB module [I-D.ietf-ipfix-mib].

    cacheDataRecords:  The number of Data Records generated by this Cache=

       since the point in time indicated by cacheIntitializationTime.
       In contrast to ipfixMeteringProcessDataRecords in the IPFIX MIB
       module [I-D.ietf-ipfix-mib], the counter is initialized with zero.=


    cacheInitializationTime:  The value of sysUpTime at the most recent
       occasion at which the Cache was re-initialized and at which the
       counter cacheDataRecords was reset to zero.  In contrast to
       ipfixMeteringProcessDiscontinuityTime in the IPFIX MIB module
       [I-D.ietf-ipfix-mib], the zero value does not have any specific
       meaning.

    The counter value of cacheDataRecords MUST be set to zero at the time=

    indicated by cacheInitializationTime.  It is RECOMMENDED to reset the=

    counter after every reconfiguration of the Cache.


Question:
Is it clear what inactive Flows are?
I can imagine what this number is supposed to by, but I'm not sure.

Regards,
Gerhard


--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms080505070708060207060803
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature

MIAGCSqGSIb3DQEHAqCAMIACAQExCzAJBgUrDgMCGgUAMIAGCSqGSIb3DQEHAQAAoIIO5jCC
BEYwggOvoAMCAQICEGb9R+PCGeToms2Z3fU6yyQwDQYJKoZIhvcNAQEFBQAwXzELMAkGA1UE
BhMCVVMxFzAVBgNVBAoTDlZlcmlTaWduLCBJbmMuMTcwNQYDVQQLEy5DbGFzcyAxIFB1Ymxp
YyBQcmltYXJ5IENlcnRpZmljYXRpb24gQXV0aG9yaXR5MB4XDTA1MTAyODAwMDAwMFoXDTE1
MTAyNzIzNTk1OVowgd0xCzAJBgNVBAYTAlVTMRcwFQYDVQQKEw5WZXJpU2lnbiwgSW5jLjEf
MB0GA1UECxMWVmVyaVNpZ24gVHJ1c3QgTmV0d29yazE7MDkGA1UECxMyVGVybXMgb2YgdXNl
IGF0IGh0dHBzOi8vd3d3LnZlcmlzaWduLmNvbS9ycGEgKGMpMDUxHjAcBgNVBAsTFVBlcnNv
bmEgTm90IFZhbGlkYXRlZDE3MDUGA1UEAxMuVmVyaVNpZ24gQ2xhc3MgMSBJbmRpdmlkdWFs
IFN1YnNjcmliZXIgQ0EgLSBHMjCCASIwDQYJKoZIhvcNAQEBBQADggEPADCCAQoCggEBAMnf
rOfq+PgDFMQAktXBfjbCPO98chXLwKuMPRyVzm8eECw/AO2XJua2x+atQx0/pIdHR0w+VPhs
+Mf8sZ69MHC8l7EDBeqV8a1AxUR6SwWi8mD81zplYu//EHuiVrvFTnAt1qIfPO2wQuhejVch
rKaZ2RHp0hoHwHRHQgv8xTTq/ea6JNEdCBU3otdzzwFBL2OyOj++pRpu9MlKWz2VphW7NQIZ
+dTvvI8OcXZZu0u2Ptb8Whb01g6J8kn+bAztFenZiHWcec5gJ925rXXOL3OVekA6hXVJsLjf
aLyrzROChRFQo+A8C67AClPN1zBvhTJGG+RJEMJs4q8fef/btLUCAwEAAaOB/zCB/DASBgNV
HRMBAf8ECDAGAQH/AgEAMEQGA1UdIAQ9MDswOQYLYIZIAYb4RQEHFwEwKjAoBggrBgEFBQcC
ARYcaHR0cHM6Ly93d3cudmVyaXNpZ24uY29tL3JwYTALBgNVHQ8EBAMCAQYwEQYJYIZIAYb4
QgEBBAQDAgEGMC4GA1UdEQQnMCWkIzAhMR8wHQYDVQQDExZQcml2YXRlTGFiZWwzLTIwNDgt
MTU1MB0GA1UdDgQWBBQRfV4ZfTwE32ps1qKKGj8x2DuUUjAxBgNVHR8EKjAoMCagJKAihiBo
dHRwOi8vY3JsLnZlcmlzaWduLmNvbS9wY2ExLmNybDANBgkqhkiG9w0BAQUFAAOBgQA8o9oC
YzrEk6qrctPcrVA4HgyeFkqIt+7r2f8PjZWg1rv6aguuYYTYaEeJ70+ssh9JQZtJM3aTi55u
uUMcYL3C3Ioth8FFwBFyBBprJCpsb+f8BxMp0Hc6I+f1wYVoGb/GAVQgGa41gsxiPGEJxvTV
67APpp8zhZrTcY5Qj5ndYjCCBUowggQyoAMCAQICEF0kYdlVNx5E1qcmxun5IrswDQYJKoZI
hvcNAQEFBQAwgd0xCzAJBgNVBAYTAlVTMRcwFQYDVQQKEw5WZXJpU2lnbiwgSW5jLjEfMB0G
A1UECxMWVmVyaVNpZ24gVHJ1c3QgTmV0d29yazE7MDkGA1UECxMyVGVybXMgb2YgdXNlIGF0
IGh0dHBzOi8vd3d3LnZlcmlzaWduLmNvbS9ycGEgKGMpMDUxHjAcBgNVBAsTFVBlcnNvbmEg
Tm90IFZhbGlkYXRlZDE3MDUGA1UEAxMuVmVyaVNpZ24gQ2xhc3MgMSBJbmRpdmlkdWFsIFN1
YnNjcmliZXIgQ0EgLSBHMjAeFw0xMDAxMDIwMDAwMDBaFw0xMTAxMDIyMzU5NTlaMIIBEzEX
MBUGA1UEChMOVmVyaVNpZ24sIEluYy4xHzAdBgNVBAsTFlZlcmlTaWduIFRydXN0IE5ldHdv
cmsxRjBEBgNVBAsTPXd3dy52ZXJpc2lnbi5jb20vcmVwb3NpdG9yeS9SUEEgSW5jb3JwLiBi
eSBSZWYuLExJQUIuTFREKGMpOTgxHjAcBgNVBAsTFVBlcnNvbmEgTm90IFZhbGlkYXRlZDEz
MDEGA1UECxMqRGlnaXRhbCBJRCBDbGFzcyAxIC0gTmV0c2NhcGUgRnVsbCBTZXJ2aWNlMRYw
FAYDVQQDFA1HZXJoYXJkIE11ZW56MSIwIAYJKoZIhvcNAQkBFhNtdWVuekBuZXQuaW4udHVt
LmRlMIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAl/YDW9cnAXJjesGhwtGwj9JN
IHWXQ7QA18s+wGZNsQY0k4eKBeLTk/RAiC1hP+uQAXLoGQHQF/ljjQCQeRmnsN6w04iAzKVo
Ru+NP7Y6ipn8uztx+z0Y3w43CO+LWUudp3PIF3+Fm2zwsAFXj/IYy/8Kq6m9+nQgYO/D4HoN
DDR3//GepXghK0x93gR0GCkm8pgKINHNBD47/IZbDe/cIpAC9HaoxCeLiB1QaCDIYPIrJ0Vx
kk0crAA0tPaMDjdSQTmNRAn2P+7Wg7K+DLvauBHfxLs8PrsolI+XdwiBu6henK9nPxvtoHmm
mfxcdslKLbS9Xz2rHxi6PVVgv1ubWwIDAQABo4HMMIHJMAkGA1UdEwQCMAAwRAYDVR0gBD0w
OzA5BgtghkgBhvhFAQcXATAqMCgGCCsGAQUFBwIBFhxodHRwczovL3d3dy52ZXJpc2lnbi5j
b20vcnBhMAsGA1UdDwQEAwIFoDAdBgNVHSUEFjAUBggrBgEFBQcDBAYIKwYBBQUHAwIwSgYD
VR0fBEMwQTA/oD2gO4Y5aHR0cDovL0luZEMxRGlnaXRhbElELWNybC52ZXJpc2lnbi5jb20v
SW5kQzFEaWdpdGFsSUQuY3JsMA0GCSqGSIb3DQEBBQUAA4IBAQBPU7OeWNo2LRzsdu/RuuAk
L5cj/RzeZ0xv9q4K/eosy9dRu45XACpPvb0fjc3u/dsY4m3jDqzUsB02IrrnF9jpEo10BJPw
b/RZ7NaU7jC/gEknL3bSXiD2PVwL4ZI5ChH5+TJUboWhVh+FpOwOfP/yKk7wQM/46iXXzRIz
enk0rcJp71k4aQgQTML3/8MDeD6TUipuQwQMY9n6T6DsNd/6JpGa+x9ZX4HIEG6+EWwQ85G5
vGfDBaq3dFrksHE75u7b7pf75wm9HIFU2SDHP2tymTHJt/XM9sqP1URIC57s+uZV0/d5KRc1
uPcJ0FAnAiC9o7ybPIh27shiq0FvOEwfMIIFSjCCBDKgAwIBAgIQXSRh2VU3HkTWpybG6fki
uzANBgkqhkiG9w0BAQUFADCB3TELMAkGA1UEBhMCVVMxFzAVBgNVBAoTDlZlcmlTaWduLCBJ
bmMuMR8wHQYDVQQLExZWZXJpU2lnbiBUcnVzdCBOZXR3b3JrMTswOQYDVQQLEzJUZXJtcyBv
ZiB1c2UgYXQgaHR0cHM6Ly93d3cudmVyaXNpZ24uY29tL3JwYSAoYykwNTEeMBwGA1UECxMV
UGVyc29uYSBOb3QgVmFsaWRhdGVkMTcwNQYDVQQDEy5WZXJpU2lnbiBDbGFzcyAxIEluZGl2
aWR1YWwgU3Vic2NyaWJlciBDQSAtIEcyMB4XDTEwMDEwMjAwMDAwMFoXDTExMDEwMjIzNTk1
OVowggETMRcwFQYDVQQKEw5WZXJpU2lnbiwgSW5jLjEfMB0GA1UECxMWVmVyaVNpZ24gVHJ1
c3QgTmV0d29yazFGMEQGA1UECxM9d3d3LnZlcmlzaWduLmNvbS9yZXBvc2l0b3J5L1JQQSBJ
bmNvcnAuIGJ5IFJlZi4sTElBQi5MVEQoYyk5ODEeMBwGA1UECxMVUGVyc29uYSBOb3QgVmFs
aWRhdGVkMTMwMQYDVQQLEypEaWdpdGFsIElEIENsYXNzIDEgLSBOZXRzY2FwZSBGdWxsIFNl
cnZpY2UxFjAUBgNVBAMUDUdlcmhhcmQgTXVlbnoxIjAgBgkqhkiG9w0BCQEWE211ZW56QG5l
dC5pbi50dW0uZGUwggEiMA0GCSqGSIb3DQEBAQUAA4IBDwAwggEKAoIBAQCX9gNb1ycBcmN6
waHC0bCP0k0gdZdDtADXyz7AZk2xBjSTh4oF4tOT9ECILWE/65ABcugZAdAX+WONAJB5Gaew
3rDTiIDMpWhG740/tjqKmfy7O3H7PRjfDjcI74tZS52nc8gXf4WbbPCwAVeP8hjL/wqrqb36
dCBg78Pgeg0MNHf/8Z6leCErTH3eBHQYKSbymAog0c0EPjv8hlsN79wikAL0dqjEJ4uIHVBo
IMhg8isnRXGSTRysADS09owON1JBOY1ECfY/7taDsr4Mu9q4Ed/Euzw+uyiUj5d3CIG7qF6c
r2c/G+2geaaZ/Fx2yUottL1fPasfGLo9VWC/W5tbAgMBAAGjgcwwgckwCQYDVR0TBAIwADBE
BgNVHSAEPTA7MDkGC2CGSAGG+EUBBxcBMCowKAYIKwYBBQUHAgEWHGh0dHBzOi8vd3d3LnZl
cmlzaWduLmNvbS9ycGEwCwYDVR0PBAQDAgWgMB0GA1UdJQQWMBQGCCsGAQUFBwMEBggrBgEF
BQcDAjBKBgNVHR8EQzBBMD+gPaA7hjlodHRwOi8vSW5kQzFEaWdpdGFsSUQtY3JsLnZlcmlz
aWduLmNvbS9JbmRDMURpZ2l0YWxJRC5jcmwwDQYJKoZIhvcNAQEFBQADggEBAE9Ts55Y2jYt
HOx279G64CQvlyP9HN5nTG/2rgr96izL11G7jlcAKk+9vR+Nze792xjibeMOrNSwHTYiuucX
2OkSjXQEk/Bv9Fns1pTuML+ASScvdtJeIPY9XAvhkjkKEfn5MlRuhaFWH4Wk7A58//IqTvBA
z/jqJdfNEjN6eTStwmnvWThpCBBMwvf/wwN4PpNSKm5DBAxj2fpPoOw13/omkZr7H1lfgcgQ
br4RbBDzkbm8Z8MFqrd0WuSwcTvm7tvul/vnCb0cgVTZIMc/a3KZMcm39cz2yo/VREgLnuz6
5lXT93kpFzW49wnQUCcCIL2jvJs8iHbuyGKrQW84TB8xggTsMIIE6AIBATCB8jCB3TELMAkG
A1UEBhMCVVMxFzAVBgNVBAoTDlZlcmlTaWduLCBJbmMuMR8wHQYDVQQLExZWZXJpU2lnbiBU
cnVzdCBOZXR3b3JrMTswOQYDVQQLEzJUZXJtcyBvZiB1c2UgYXQgaHR0cHM6Ly93d3cudmVy
aXNpZ24uY29tL3JwYSAoYykwNTEeMBwGA1UECxMVUGVyc29uYSBOb3QgVmFsaWRhdGVkMTcw
NQYDVQQDEy5WZXJpU2lnbiBDbGFzcyAxIEluZGl2aWR1YWwgU3Vic2NyaWJlciBDQSAtIEcy
AhBdJGHZVTceRNanJsbp+SK7MAkGBSsOAwIaBQCgggLOMBgGCSqGSIb3DQEJAzELBgkqhkiG
9w0BBwEwHAYJKoZIhvcNAQkFMQ8XDTEwMDMxMTEwMjkxNlowIwYJKoZIhvcNAQkEMRYEFKbD
Dol7+cuiOeUNWM+UI/IBbz+QMF8GCSqGSIb3DQEJDzFSMFAwCwYJYIZIAWUDBAECMAoGCCqG
SIb3DQMHMA4GCCqGSIb3DQMCAgIAgDANBggqhkiG9w0DAgIBQDAHBgUrDgMCBzANBggqhkiG
9w0DAgIBKDCCAQMGCSsGAQQBgjcQBDGB9TCB8jCB3TELMAkGA1UEBhMCVVMxFzAVBgNVBAoT
DlZlcmlTaWduLCBJbmMuMR8wHQYDVQQLExZWZXJpU2lnbiBUcnVzdCBOZXR3b3JrMTswOQYD
VQQLEzJUZXJtcyBvZiB1c2UgYXQgaHR0cHM6Ly93d3cudmVyaXNpZ24uY29tL3JwYSAoYykw
NTEeMBwGA1UECxMVUGVyc29uYSBOb3QgVmFsaWRhdGVkMTcwNQYDVQQDEy5WZXJpU2lnbiBD
bGFzcyAxIEluZGl2aWR1YWwgU3Vic2NyaWJlciBDQSAtIEcyAhBdJGHZVTceRNanJsbp+SK7
MIIBBQYLKoZIhvcNAQkQAgsxgfWggfIwgd0xCzAJBgNVBAYTAlVTMRcwFQYDVQQKEw5WZXJp
U2lnbiwgSW5jLjEfMB0GA1UECxMWVmVyaVNpZ24gVHJ1c3QgTmV0d29yazE7MDkGA1UECxMy
VGVybXMgb2YgdXNlIGF0IGh0dHBzOi8vd3d3LnZlcmlzaWduLmNvbS9ycGEgKGMpMDUxHjAc
BgNVBAsTFVBlcnNvbmEgTm90IFZhbGlkYXRlZDE3MDUGA1UEAxMuVmVyaVNpZ24gQ2xhc3Mg
MSBJbmRpdmlkdWFsIFN1YnNjcmliZXIgQ0EgLSBHMgIQXSRh2VU3HkTWpybG6fkiuzANBgkq
hkiG9w0BAQEFAASCAQA8/9EB896YB8KgtqcF761j/IPetbVryP/Mcp3TsV3gBPNLEz558qoj
V7b1OYvbwgM3MVPasWm5rNVmnOFZYTslq5U3fnC17GmJLU6isjeNUghHerLYsdXKp9MNBGXm
2mW7/Do1c4nobA44SSZn+H7s0O81xSNPPD8mEYotNOiFvyfTcvguUzNXmEvHV4o33q3gOgqD
fp0WMFsc1ASsSsfKg5gyA2xoVDqrxdex1Fk/T7yBOt9cH95KTdhbjaT7UwGp/Te5rYkqN4BX
k44AwCRSdPqPmvrKCe1SRnJpYXRBanu3yw9Or7x8cmmS+2kMGJQsqvc+tfHCqpkg36i/bPxl
AAAAAAAA
--------------ms080505070708060207060803--

From muenz@net.in.tum.de  Thu Mar 11 04:30:52 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 3A2723A6B47 for <ipfix@core3.amsl.com>; Thu, 11 Mar 2010 04:30:52 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.042
X-Spam-Level: 
X-Spam-Status: No, score=-1.042 tagged_above=-999 required=5 tests=[AWL=-1.207, BAYES_40=-0.185, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id FYjjL1BibAWz for <ipfix@core3.amsl.com>; Thu, 11 Mar 2010 04:30:50 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 7D7A33A6AFE for <ipfix@ietf.org>; Thu, 11 Mar 2010 04:30:49 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 81E5720C3BEA; Thu, 11 Mar 2010 13:30:46 +0100 (CET)
Message-ID: <4B98E288.4040204@net.in.tum.de>
Date: Thu, 11 Mar 2010 13:31:04 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Andrew Johnson <andrjohn@cisco.com>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B9257CE.6040409@cisco.com>
In-Reply-To: <4B9257CE.6040409@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms020703090909090707020407"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 11 Mar 2010 12:30:52 -0000

This is a cryptographically signed message in MIME format.

--------------ms020703090909090707020407
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Andrew, all,

>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>> Section 3:
>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>> If a Selection Process is part of a Metering Process and a single
>>> Selection Process can feed into more than one Cache, then it is
>>> implied that a single Metering Process can have more than once cache.=

>> Note: The text has slightly changed after Atsushi's review. Now, it=20
>> explicitly says that a Metering Process may contain multiple Selection=
=20
>> Processes and multiple Caches.
>=20
> If a Metering Process can have multiple caches and an IPFIX device can =

> have multiple Metering Processes, how do we associate the caches with=20
> Metering Processes?  Can this configuration system only be used to=20
> configure a single Metering Process within a device?

The model does not group Selection Processes and Caches into Metering=20
Processes. This is done on purpose because there is not a single correct =

view.

All Selection Process and all Caches of a Monitoring Device can be=20
regarded as one Metering Process.

Alternatively, Selection Processes and Caches which are not directly or=20
indirectly linked to each other can be seen as belonging to different=20
Metering Processes.

>> Furthermore, it says that only packets observed at a single Observatio=
n=20
>> Point may enter a Selection Process. IMO, this results from the PSAMP =

>> definitions of Observed Packet Stream and Selection Process. Please te=
ll=20
>> me if this conclusion is wrong.
>=20
> A Selection process should only have a single feed, but can have=20
> multiple observers.  This results in the number of possible chains=20
> (Selection Sequences) being equal to the number of Selectors that=20
> directly feed a cache.

Ok. Single feed means single Observation Point as specified in the draft.=


>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>> From section 3.1:
>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>> Why have so many SCTP or UDP only options?  Can't you use UML
>>> inheritance to do something like:
>>>
>>>     +------------------------------------------------+
>>>     | Destination                                    |
>>>     +------------------------------------------------+
>>>     | name                                           |
>>>     | exportMemberType =3D parallel                    |
>>>     | ipfixVersion =3D 10                              |
>>>     | transportProtocol                              |
>>>     | destinationIpAddress                           |
>>>     | destinationPort =3D 4739|4740                    |
>>>     | sendBufferSize {opt.}                          |
>>>     | rateLimit[0..1]                                |
>>>     +------------------------------------------------+
>>>           ^
>>>           | 1
>>>      +----+-------------------+
>>>      |                        |
>>>      |    +------------------------------------------------+
>>>      |    | SCTP Destination                               |
>>>      |    +------------------------------------------------+
>>>      |    | localIpAddress[0..*] {SCTP only}              =20
>>> |<>-----+                |    | timedReliability =3D 0 {SCTP=20
>>> only}               |       | 0..1         |    | numberOfStreams=20
>>> {opt.} {SCTP only}             |       |                |    |=20
>>> sourceIpAddress[0..1] {UDP only}               |   +------------+
>>>      |    | templateRefreshTimeout =3D 600 {UDP only}        |   |=20
>>> Transport- |
>>>      |    | optionsTemplateRefreshTimeout =3D 600 {UDP only} |   |=20
>>> Layer-     |
>>>      |    | templateRefreshPacket[0..1] {UDP only}         |   |=20
>>> Security   |
>>>      |    | optionsTemplateRefreshPacket[0..1] {UDP only}  |  =20
>>> +------------+
>>>      |    +------------------------------------------------+
>>>      |
>>>      +---------------------+
>>>                            |
>>>     +------------------------------------------------+
>>>     | UDP Destination                                |
>>>     +------------------------------------------------+
>>>     | sourceIpAddress[0..1] {UDP only}              =20
>>> |<>-----+               | templateRefreshTimeout =3D 600 {UDP=20
>>> only}        |       | 0..1        | optionsTemplateRefreshTimeout =3D=
=20
>>> 600 {UDP only} |       |               | templateRefreshPacket[0..1] =

>>> {UDP only}         |   +------------+
>>>     | optionsTemplateRefreshPacket[0..1] {UDP only}  |   | Transport-=
 |
>>>     +------------------------------------------------+   | Layer-    =
 |
>>>                                    | Security   |
>>>                                                          +-----------=
-+
>> This looks nice. Nevertheless, there are some drawbacks:
>>
>> Since YANG does not support inheritance, it cannot be mapped to YANG=20
>> directly.
>=20
> I'm not that familiar with YANG, perhaps it deals with the situation=20
> more gracefully than the UML diagram shows.  Can you point the=20
> Destination to one of a choice of protocol specific objects?

Agreed. It is easier to model and describe protocol-specific Destination =

classes. To be consistent, I propose to have protocol-specific Receiver=20
classes in the CollectingProcess class as well.


4.4.  ExportingProcess Class

              +--------------------+
              | ExportingProcess   |
              +--------------------+   0..* +------------------+
              | name               |<>------| SctpDestination  |
              |                    |        +------------------+
              |                    |
              |                    |   0..* +------------------+
              |                    |<>------| UdpDestination   |
              |                    |        +------------------+
              |                    |
              |                    |   0..* +------------------+
              |                    |<>------| TcpDestination   |
              |                    |        +------------------+
              |                    |
              |                    |   0..* +------------------+
              |                    |<>------| FileWriter       |
              |                    |        +------------------+
              |                    |
              |                    |   0..* +------------------+
              |                    |<>------| Options          |
              |                    |        +------------------+
              |                    |
              |                    |   0..* +------------------+
              |                    |<>------| TransportSession |
              +--------------------+        +------------------+


4.5.  CollectingProcess Class

            +-------------------+
            | CollectingProcess |
            +-------------------+
            | name              |       0..* +------------------+
            |                   |<>----------| SctpReceiver     |
            |                   |            +------------------+
            |                   |
            |                   |       0..* +------------------+
            |                   |<>----------| UdpReceiver      |
            |                   |            +------------------+
            |                   |
            |                   |       0..* +------------------+
            |                   |<>----------| TcpReceiver      |
            |                   |            +------------------+
            |                   |
            |                   |       0..* +------------------+
            |                   |<>----------| FileReader       |
            |                   |            +------------------+
            |                   |
            |                   | 0..*  0..* +------------------+
            |                   |----------->| ExportingProcess |
            |                   |            +------------------+
            |                   |
            |                   |       0..* +------------------+
            |                   |<>----------| TransportSession |
            +-------------------+            +------------------+


>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>> From section 4.1:
>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>> Doesn't the direction belong in the Observation Point class, not the
>>> Interface or LC class.
>> If the current position of the direction parameter bothers you, we can=
=20
>> move it.
>>
>> However, my reasoning is the following:
>> - If it appears in the ObservationPoint class, it looks like the=20
>> Observation Point is doing some filtering.
>> - If it appears in the Interface/Linecard class, it looks like the=20
>> filtering is already done at the interface or linecard.
>>
>> I thought that the interface/linecard is doing the filtering.
>=20
> I don't see it as filtering, just part of the definition of the=20
> Observation Point.  The Interface is most likely able to send and=20
> receive traffic, but the observation point could be on only the send or=
=20
> receive channel.  i.e. It's not a input only interface, but the=20
> Observation Point could only see input packets.
>=20
> Currently you have a 1:1 mapping of OP to interface.  If the direction =

> is moved to the Observation Point then you could have two observation=20
> points on the same interface, one on the input side and one on the outp=
ut.

With the current model, there is no problem to configure two OPs on the=20
same interface:

      <observationPoint>
        <name>interface ingress</name>
        <interface>
          <ifIndex>3</ifIndex>
          <direction>ingress</direction>
        </interface>
        ...
      </observationPoint>

      <observationPoint>
        <name>interface egress</name>
        <interface>
          <ifIndex>3</ifIndex>
          <direction>egress</direction>
        </interface>
        ...
      </observationPoint>

>> There is another small advantage: The current position of the paramete=
r=20
>> makes it easier to add vendor or implementation-specific packet source=
s=20
>> for which the direction parameter has no meaning, for example:
>>
>> +------------------+
>> | PcapFile         |
>> +------------------+
>> | filename         |
>> +------------------+
>>
>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>> From Section 4.2.1
>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>> Why packetsDropped and not packetsSelected, when PSAMP talks about
>>> packetsSelected in the stats it needs to export?
>>
>> All state parameters are adopted from the IPFIX-MIB. There, we have:
>>
>> +--ipfixSelectionProcessStatsTable(4)
>>    |
>>    +--ipfixSelectionProcessStatsEntry(1)=20
>> [ipfixMeteringProcessCacheId,ipfixSelectionProcessIndex,ipfixSelection=
ProcessSelectorIndex]=20
>>
>>       |
>>       +-- r-n Counter64 ipfixSelectionProcessStatsPacketsObserved(1)
>>       +-- r-n Counter64 ipfixSelectionProcessStatsPacketsDropped(2)
>>       +-- r-n TimeStamp ipfixSelectionProcessStatsDiscontinuityTime(3)=

>>
>>
>> Probably, ipfixSelectionProcessStatsPacketsSelected would make more=20
>> sense in the MIB.
>>
>> I assume that the MIB standardization cannot be stopped any more, righ=
t?
>>
>> So, what do you suggest?
>=20
> I don't know.  Why follow the MIB exactly?  Any idea about the history =

> of this choice or dropped rather than selected?

No idea.

IMO, it does not make sense to have three counters (packetsDropped,=20
packetsSelected, packetsObserved) because two of them are sufficient.

For the moment, I'll keep packetsObserved and packetsDropped like in the =

MIB.

Anyway, as concluded in the other e-mail thread, these counters are not=20
necessarily identical to those exported in the PSAMP statistic reports.

>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>> From Section 4.3.1
>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>> Isn't is possible to generate a cache without any keys?
>>> This would generate a cache with a single flow?  Do we allow this?
>> A the moment, there is no restriction which says that a flow key field=
=20
>> must exist.
>>
>> Do you want to add such a restriction?
>=20
> A cache without keys will result in a single flow, which seems a bit=20
> daft.  I guess there is no pressing reason to add this restriction thou=
gh.
 >
>>> Would it be useful to have two lists, one for keys and one for
>>> non-keys?
>> There is an argument for having a single list:
>>
>> A Cache implementation may generate records using the specified field =

>> order. So, the fields occur in the same order as specified in the=20
>> configuration.
>=20
> If we're worried about implementation specifics can't they extend the=20
> YANG model?

Yes, they can.

>> With separate lists for flow key fields and non-key fields, we cannot =

>> arbitrarily mix these two types of fields.
>>
>> Is there any advantage of having separate lists?
>=20
> I was just thinking about how you'd specify that at least one key was=20
> required.

Caches of type "immediate" do not have Flow Keys. So, the restrictions=20
would not apply to such Caches.

>>> For a permanent cache, update counters should be used and for a
>>> timeout cache the delta counters should be used.  Where is this
>>> sort of thing enforced?
>> This is not enforced. I would not strictly enforce it because strange =

>> configurations might make sense in specific use cases.
>=20
> OK.
>=20
>> However, we could add a note that totalCounters are typically used for=
=20
>> "permanent" and deltaCounters for "timeout".
>>
>> Could you provide a sentence or two?
>=20
> Perhaps a more general note that not all fields will be appropriate for=
=20
> all configurations. Is the Exporting Device free to change the fields
 > if it needs total counters for a permanent cache?

No.

I added a new paragraph at the end of 4.3.1. (CacheLayout Class):

    Note that the use of Information Elements can be restricted to
    certain Cache Modes or to Flow Key or non-key fields.  Such
    restrictions may result from Information Element definitions or from
    device-specific constraints.  According to Section 5, the Monitoring
    Device MUST notify the user if a Cache field cannot be configured
    with the given Information Element.

Ok?

>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>> From Section 4.4.1
>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>>    exportMemberType:  Configures the export member type that correspo=
nds
>>>       to the ipfixTransportSessionGroupMemberType object in
>>>       [I-D.ietf-ipfix-mib].  The following parameter values are
>>>       specified by the configuration data model:
>>>       *  primary: primary target of the Exporting Process
>>>       *  secondary: secondary target of the Exporting Process used wh=
en
>>>          the primary target is not reachable
>>>       *  parallel: parallel exporting to all destinations and files o=
f
>>>          the Exporting Process
>>>       *  loadBalancing: load-balancing between different destinations=

>>>          and files of the Exporting Process
>>>       "parallel" is the default value if this parameter is not
>>>       configured.  If one destination or file is configured as "prima=
ry"
>>>
>>> ---------------------------------------------------------------------=
---
>>> Comments
>>> ---------------------------------------------------------------------=
---
>>> It's not clear to me that this parameter belongs to the Destination
>>> class.  The number of restrictions on when you can use the various
>>> types suggests that we could simplify things.
>>>
>>> A Cache can pass flow records to multiple Exporting Processes, so we
>>> don't really need a parallel option.  This would mean that each
>>> Exporting Process is responsible for sending one complete set of
>>> records.  i.e. one record in, one record out.  It could do that, eith=
er
>>> using a redundancy model or with a load-balancing one.
>>>
>>> In other words, the exportType would be an Exporting Process attribut=
e
>>> which would indicate whether the list of Destinations should be load-=

>>> balanced across or treated as Primary, Seondary, Tertiary, etc.
>> At the moment, the parameter is inline with the IPFIX-MIB:
>>
>> ipfixExportMemberType OBJECT-TYPE
>>     SYNTAX      INTEGER {
>>                     unknown(0),
>>                     primary(1),
>>                     secondary(2),
>>                     parallel(3),
>>                     loadBalancing(4)
>>                 }
>>     MAX-ACCESS  read-only
>>     STATUS      current
>>     DESCRIPTION
>>         "The type of a member Transport Session in a Transport
>>         Session group (identified by the value of ipfixExportIndex,
>>         ipfixObservationDomainId and ipfixMeteringProcessCacheId).
>>         The following values are valid:
>>
>>         unknown(0)
>>             This value MUST be used if the status of the group
>>             membership cannot be detected by the equipment. This
>>             value should be avoided as far as possible.
>>
>>         primary(1)
>>             This value is used for a group member that is used as
>>             the primary target of an Exporter. Other group members
>>             (with the same ipfixExportIndex and
>>             ipfixMeteringProcessCacheId) MUST NOT have the value
>>             primary(1) but MUST have the value secondary(2).
>>             This value MUST also be specified if the Exporter does
>>             not support Transport Session grouping. In this case
>>             the group contains only one Transport Session.
>>
>>         secondary(2)
>>             This value is used for a group member that is used as a
>>             secondary target of an Exporter. The Exporter will use
>>             one of the targets specified as secondary(2) within the
>>             same Transport Session group when the primary target is
>>             not reachable.
>>
>>         parallel(3)
>>             This value is used for a group member that is used for
>>             duplicate exporting i.e., all group members identified
>>             by the ipfixExportIndex are exporting the same Records
>>             in parallel. This implies that all group members MUST
>>             have the same membertype parallel(3).
>>
>>         loadBalancing(4)
>>             This value is used for a group member that is used
>>             as one target for load-balancing. This means that a
>>             Record is sent to one of the group members in this
>>             group identified by ipfixExportIndex.
>>             This implies that all group members MUST have the same
>>             membertype loadBalancing(4)."
>>     ::=3D { ipfixExportEntry 2 }
>>
>>
>> I'm open to better proposals.
>>
>> "parallel" and "loadBalancing" can be configured once with a single=20
>> ExportingProcess class parameter.
>>
>> How would you configure "primary" and "secondary with a single Exporti=
ng=20
>> Process parameter?

This issue is on my todo list.

>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>> From Section 4.7: TransportSession class
>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>> I think you want to add a timestamp for when the session came up
>>> and when it went down (unless it's still active).
>> Currently, there is no such parameter in ipfixTransportSessionTable.
>>
>> Hm, it seems that we should have asked you to review the IPFIX-MIB :)
>>
>> Shall we go beyond IPFIX-MIB and add such state parameters?
>=20
> I don't see why not.  Is it possible to add to MIB after the initial=20
> definition?  Not necessarily now, but perhaps after it's been in use a =

> while and certain additions are in demand.
>=20
>=20
>>> There's no association with the relevant destination.  It seems to me=

>>> that your really want this as part of each Destination or FileReader.=

>>> Specifically, I think you want to separate the Transport Session from=

>>> the File Reader stats.  Potentially there could be a one to many
>>> relationship of Destination and Transport Sessions, each with their o=
wn
>>> timestamp for coming up and going down.
>> Currently, TransportSession is a mere copy of ipfixTransportSessionTab=
le=20
>> which exists for both, Exporters and Collectors.
>>
>> In the case of the _Exporting Process_ , we could move the state=20
>> parameters to the _Destination_ and _FileWriter_ class.
>>
>> I think that there can be at most one Transport Session per Destinatio=
n=20
>> and per FileWriter object. At least at the same point in time.
>> However, it may happen that no Transport Session exists, for example i=
n=20
>> the case of a secondary export destination.
>=20
> Well let's imagine and Exporting process with a Primary Destination and=
=20
> a Secondary Destination.  The Primary Destination starts a Transport=20
> Session, which works for a while and then is lost.  The Secondary=20
> Destination then opens a Transport Session, which works fine.  The=20
> Exporting Process then manages to reconnect with Primary Destination an=
d=20
> opens a new Transport Session with it and closes the Secondary=20
> Destination's Transport Session.
>=20
> So, we have seen three Transport Sessions and if the Exporting Process =

> kept a full history of the statistics it could show:
>=20
>    Primary Destination (10.2.100.2 port 4739)
>      - Transport Session (Open)
>          Start Time: 10:01.19
>          End Time: --:--.--
>          Records sent: 5345
>          ...
>     -  Transport Session (Closed)
>          Start Time: 08:00.00
>          End Time: 09:45.14
>          Records sent: 123210
>=20
>    Secondary Destination (10.3.0.15 port 4739)
>     - Transport Session (Closed)
>         Start Time: 09:45.16
>         End Time: 10:01.20
>         Records sent: 14022
>=20
> I guess it would be up to the implementation to control how much histor=
y=20
> was kept.

But what happens if you remove the primary destination from the=20
configuration?

If the transport session history is linked to the destination, it will=20
be lost.

If you want to keep a history of Transport Sessions, it is probably=20
better to keep the information at a distinct place at it is now.

>> So, we could move a single set of Transport Session state parameters t=
o=20
>> the Destination and FileWriter classes. The parameters would appear or=
=20
>> not, depending on whether the Transport Session exists.
>=20
> I think this makes sense, otherwise you have to match the Destination=20
> configuration to the Transport Session statistics by matching the=20
> various address and port field.
 >
>> In the case of the _Collecting Process_ , we can also move the Transpo=
rt=20
>> Session state parameters to the _Receiver_ and _FileReader_ classes. A=
s=20
>> you say, there can be multiple Transport Sessions per Receiver.
>=20
> Well it seems that every field in the Destination Class is present in=20
> the Transport Session Class.  In fact, the Transport Session seems to b=
e=20
> a copy of the Destination Class plus some statistics.  This is not true=
=20
> of the Receiver class.

There is an important difference between configuration parameter and=20
state parameters:

- configuration parameters are not supposed to change dynamically, only
   after the (re)configuration
- state parameters may change at any time

For example, if the user does not configure a source IP address in the=20
Destination class, the source IP address in the IPFIX packets changes if =

the outgoing interface changes.

> Perhaps the Destination Class could be replaced by a Transport Session =

> class that has some readonly stats and the Receiver Class can then=20
> contain multiple Transport Sessions.

IMO, it is wrong to think that Transport Sessions are configurable.
Transport Sessions represent states which may change dynamically.

If you cut a link, the Transport Session does not exist any more, but=20
the configuration of the Exporting Process is still there.

Similarly, if you configure a backup destination, the Transport Session=20
won't exist until the backup is needed.

What we can do is having one Transport Session per Destination:

     +------------------------------+
     | SctpDestination              |
     +------------------------------+    0..1 +------------------------+
     | name                         |<>-------| TransportLayerSecurity |
     | exportMemberType =3D parallel  |         +------------------------=
+
     | ipfixVersion =3D 10            |
     | sourceIpAddress[0..*]        |    0..1 +------------------------+
     | destinationIpAddress[1..*]   |<>-------+ TransportSession       |
     | destinationPort =3D 4739|4740  |         +------------------------=
+
     | ifName/ifIndex[0..1]         |
     | sendBufferSize {opt.}        |
     | rateLimit[0..1]              |
     | timedReliability =3D 0         |
     | numberOfStreams {opt.}       |
     | numberOfStreams {opt.}       |
     +------------------------------+

If the Transport Session exists, the TransportSession object exists and=20
includes the complete state of the current Transport Session (including=20
state parameters which result from configuration, such as=20
destinationIpAddress).

In the case of the Collecting Process, we can do the same, allowing=20
multiple Transport Sessions per Receiver:


       +--------------------------+
       | SctpReceiver             |
       +--------------------------+    0..1 +------------------------+
       | name                     |<>-------| TransportLayerSecurity |
       | localIpAddress[0..*]     |         +------------------------+
       | localPort =3D 4739|4740    |
       | maxAllowedStreams {opt.} |    0..* +------------------------+
       +--------------------------+<>-------+ TransportSession       |
                                            +------------------------+

However, historic Transport Session would not be included.

Opinions?

Regards,
Gerhard

> In this case and Exporting Process wouldn't have a history of closed=20
> Transport Sessions, but would have information about the most recent on=
e=20
> per Destination.  The above example would become:
>=20
>    Primary Transport Session
>      Destination IP address: 10.2.100.2
>      Destination port:       4739
>      Status:                 Open
>      Start Time:             10:01.19
>      End Time:               --:--.--
>      Records sent:           5345
>      ...
>=20
>    Secondary Transport Session
>      Destination IP address: 10.3.0.15
>      Destination port:       4739
>      Status:                 Closed
>      Start Time:             09:45.16
>      End Time:               10:01.20
>      Records sent:           14022
>=20
>=20
>> Probably, it makes sense to have a container=20
>> <transportSessionStatistics> which groups the state parameters of a=20
>> Transport Session. This container may appear multiple times in the cas=
e=20
>> of the Receiver class, and only once in all other classes.
>>
>> Would you agree with these changes?
>=20
> I think we need a little more discussion :)
>=20
> Cheers
>=20
> Andrew

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms020703090909090707020407
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms020703090909090707020407--

From andrjohn@cisco.com  Thu Mar 11 08:00:44 2010
Return-Path: <andrjohn@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id A75613A6972 for <ipfix@core3.amsl.com>; Thu, 11 Mar 2010 08:00:44 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -9.266
X-Spam-Level: 
X-Spam-Status: No, score=-9.266 tagged_above=-999 required=5 tests=[AWL=1.333,  BAYES_00=-2.599, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id qeWYMsqFo5mI for <ipfix@core3.amsl.com>; Thu, 11 Mar 2010 08:00:42 -0800 (PST)
Received: from ams-iport-1.cisco.com (ams-iport-1.cisco.com [144.254.224.140]) by core3.amsl.com (Postfix) with ESMTP id 7C7703A685E for <ipfix@ietf.org>; Thu, 11 Mar 2010 08:00:06 -0800 (PST)
Authentication-Results: ams-iport-1.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AvQAAIuimEuQ/uCWe2dsb2JhbACaYRUBAQsLJAYcpEOYboR7BA
X-IronPort-AV: E=Sophos;i="4.49,620,1262563200"; d="scan'208";a="57933487"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 11 Mar 2010 16:00:10 +0000
Received: from cisco.com (mrwint.cisco.com [64.103.71.48]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2BG0AdH018213; Thu, 11 Mar 2010 16:00:10 GMT
Received: from [10.55.163.37] (ams-andrjohn-8714.cisco.com [10.55.163.37]) by cisco.com (8.11.7p3+Sun/8.8.8) with ESMTP id o2BG09X06105; Thu, 11 Mar 2010 16:00:09 GMT
Message-ID: <4B991389.2090209@cisco.com>
Date: Thu, 11 Mar 2010 16:00:09 +0000
From: Andrew Johnson <andrjohn@cisco.com>
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B9257CE.6040409@cisco.com> <4B98E288.4040204@net.in.tum.de>
In-Reply-To: <4B98E288.4040204@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 11 Mar 2010 16:00:44 -0000

Hi Gerhard

Gerhard Muenz wrote:
> 
> Hi Andrew, all,
> 
>>>> ======================================================================== 
>>>>
>>>> Section 3:
>>>> ======================================================================== 
>>>>
>>>> If a Selection Process is part of a Metering Process and a single
>>>> Selection Process can feed into more than one Cache, then it is
>>>> implied that a single Metering Process can have more than once cache.
>>> Note: The text has slightly changed after Atsushi's review. Now, it 
>>> explicitly says that a Metering Process may contain multiple 
>>> Selection Processes and multiple Caches.
>>
>> If a Metering Process can have multiple caches and an IPFIX device can 
>> have multiple Metering Processes, how do we associate the caches with 
>> Metering Processes?  Can this configuration system only be used to 
>> configure a single Metering Process within a device?
> 
> The model does not group Selection Processes and Caches into Metering 
> Processes. This is done on purpose because there is not a single correct 
> view.
> 
> All Selection Process and all Caches of a Monitoring Device can be 
> regarded as one Metering Process.
> 
> Alternatively, Selection Processes and Caches which are not directly or 
> indirectly linked to each other can be seen as belonging to different 
> Metering Processes.

OK, at this point I think I'm uncomfortable with the idea that a 
Metering Process can have more than one cache.  What is the reasoning 
behind this decision?


>>> Furthermore, it says that only packets observed at a single 
>>> Observation Point may enter a Selection Process. IMO, this results 
>>> from the PSAMP definitions of Observed Packet Stream and Selection 
>>> Process. Please tell me if this conclusion is wrong.
>>
>> A Selection process should only have a single feed, but can have 
>> multiple observers.  This results in the number of possible chains 
>> (Selection Sequences) being equal to the number of Selectors that 
>> directly feed a cache.
> 
> Ok. Single feed means single Observation Point as specified in the draft.

No, by "feed" I mean input stream.  Either from an Observation point or 
the output of a different Selection Process.


>>>> ======================================================================== 
>>>>
>>>> From section 3.1:
>>>> ======================================================================== 
>>>>
>>>> Why have so many SCTP or UDP only options?  Can't you use UML
>>>> inheritance to do something like:
>>>>
>>>>     +------------------------------------------------+
>>>>     | Destination                                    |
>>>>     +------------------------------------------------+
>>>>     | name                                           |
>>>>     | exportMemberType = parallel                    |
>>>>     | ipfixVersion = 10                              |
>>>>     | transportProtocol                              |
>>>>     | destinationIpAddress                           |
>>>>     | destinationPort = 4739|4740                    |
>>>>     | sendBufferSize {opt.}                          |
>>>>     | rateLimit[0..1]                                |
>>>>     +------------------------------------------------+
>>>>           ^
>>>>           | 1
>>>>      +----+-------------------+
>>>>      |                        |
>>>>      |    +------------------------------------------------+
>>>>      |    | SCTP Destination                               |
>>>>      |    +------------------------------------------------+
>>>>      |    | localIpAddress[0..*] {SCTP only}               
>>>> |<>-----+                |    | timedReliability = 0 {SCTP 
>>>> only}               |       | 0..1         |    | numberOfStreams 
>>>> {opt.} {SCTP only}             |       |                |    | 
>>>> sourceIpAddress[0..1] {UDP only}               |   +------------+
>>>>      |    | templateRefreshTimeout = 600 {UDP only}        |   | 
>>>> Transport- |
>>>>      |    | optionsTemplateRefreshTimeout = 600 {UDP only} |   | 
>>>> Layer-     |
>>>>      |    | templateRefreshPacket[0..1] {UDP only}         |   | 
>>>> Security   |
>>>>      |    | optionsTemplateRefreshPacket[0..1] {UDP only}  |   
>>>> +------------+
>>>>      |    +------------------------------------------------+
>>>>      |
>>>>      +---------------------+
>>>>                            |
>>>>     +------------------------------------------------+
>>>>     | UDP Destination                                |
>>>>     +------------------------------------------------+
>>>>     | sourceIpAddress[0..1] {UDP only}               
>>>> |<>-----+               | templateRefreshTimeout = 600 {UDP 
>>>> only}        |       | 0..1        | optionsTemplateRefreshTimeout = 
>>>> 600 {UDP only} |       |               | templateRefreshPacket[0..1] 
>>>> {UDP only}         |   +------------+
>>>>     | optionsTemplateRefreshPacket[0..1] {UDP only}  |   | Transport- |
>>>>     +------------------------------------------------+   | Layer-     |
>>>>                                    | Security   |
>>>>                                                          +------------+
>>> This looks nice. Nevertheless, there are some drawbacks:
>>>
>>> Since YANG does not support inheritance, it cannot be mapped to YANG 
>>> directly.
>>
>> I'm not that familiar with YANG, perhaps it deals with the situation 
>> more gracefully than the UML diagram shows.  Can you point the 
>> Destination to one of a choice of protocol specific objects?
> 
> Agreed. It is easier to model and describe protocol-specific Destination 
> classes. To be consistent, I propose to have protocol-specific Receiver 
> classes in the CollectingProcess class as well.
> 
> 
> 4.4.  ExportingProcess Class
> 
>              +--------------------+
>              | ExportingProcess   |
>              +--------------------+   0..* +------------------+
>              | name               |<>------| SctpDestination  |
>              |                    |        +------------------+
>              |                    |
>              |                    |   0..* +------------------+
>              |                    |<>------| UdpDestination   |
>              |                    |        +------------------+
>              |                    |
>              |                    |   0..* +------------------+
>              |                    |<>------| TcpDestination   |
>              |                    |        +------------------+
>              |                    |
>              |                    |   0..* +------------------+
>              |                    |<>------| FileWriter       |
>              |                    |        +------------------+
>              |                    |
>              |                    |   0..* +------------------+
>              |                    |<>------| Options          |
>              |                    |        +------------------+
>              |                    |
>              |                    |   0..* +------------------+
>              |                    |<>------| TransportSession |
>              +--------------------+        +------------------+
> 
> 
> 4.5.  CollectingProcess Class
> 
>            +-------------------+
>            | CollectingProcess |
>            +-------------------+
>            | name              |       0..* +------------------+
>            |                   |<>----------| SctpReceiver     |
>            |                   |            +------------------+
>            |                   |
>            |                   |       0..* +------------------+
>            |                   |<>----------| UdpReceiver      |
>            |                   |            +------------------+
>            |                   |
>            |                   |       0..* +------------------+
>            |                   |<>----------| TcpReceiver      |
>            |                   |            +------------------+
>            |                   |
>            |                   |       0..* +------------------+
>            |                   |<>----------| FileReader       |
>            |                   |            +------------------+
>            |                   |
>            |                   | 0..*  0..* +------------------+
>            |                   |----------->| ExportingProcess |
>            |                   |            +------------------+
>            |                   |
>            |                   |       0..* +------------------+
>            |                   |<>----------| TransportSession |
>            +-------------------+            +------------------+

OK, this approach seems fine to me.


>>>> ======================================================================== 
>>>>
>>>> From section 4.1:
>>>> ======================================================================== 
>>>>
>>>> Doesn't the direction belong in the Observation Point class, not the
>>>> Interface or LC class.
>>> If the current position of the direction parameter bothers you, we 
>>> can move it.
>>>
>>> However, my reasoning is the following:
>>> - If it appears in the ObservationPoint class, it looks like the 
>>> Observation Point is doing some filtering.
>>> - If it appears in the Interface/Linecard class, it looks like the 
>>> filtering is already done at the interface or linecard.
>>>
>>> I thought that the interface/linecard is doing the filtering.
>>
>> I don't see it as filtering, just part of the definition of the 
>> Observation Point.  The Interface is most likely able to send and 
>> receive traffic, but the observation point could be on only the send 
>> or receive channel.  i.e. It's not a input only interface, but the 
>> Observation Point could only see input packets.
>>
>> Currently you have a 1:1 mapping of OP to interface.  If the direction 
>> is moved to the Observation Point then you could have two observation 
>> points on the same interface, one on the input side and one on the 
>> output.
> 
> With the current model, there is no problem to configure two OPs on the 
> same interface:
> 
>      <observationPoint>
>        <name>interface ingress</name>
>        <interface>
>          <ifIndex>3</ifIndex>
>          <direction>ingress</direction>
>        </interface>
>        ...
>      </observationPoint>
> 
>      <observationPoint>
>        <name>interface egress</name>
>        <interface>
>          <ifIndex>3</ifIndex>
>          <direction>egress</direction>
>        </interface>
>        ...
>      </observationPoint>

OK, but my point was always that the interface is not defined by its 
direction.

These two "interfaces" are actually two observation points on the same 
interface:
        <interface>
          <ifIndex>3</ifIndex>
          <direction>ingress</direction>
        </interface>

        <interface>
          <ifIndex>3</ifIndex>
          <direction>ingress</direction>
        </interface>


>>> There is another small advantage: The current position of the 
>>> parameter makes it easier to add vendor or implementation-specific 
>>> packet sources for which the direction parameter has no meaning, for 
>>> example:
>>>
>>> +------------------+
>>> | PcapFile         |
>>> +------------------+
>>> | filename         |
>>> +------------------+
>>>
>>>> ======================================================================== 
>>>>
>>>> From Section 4.2.1
>>>> ======================================================================== 
>>>>
>>>> Why packetsDropped and not packetsSelected, when PSAMP talks about
>>>> packetsSelected in the stats it needs to export?
>>>
>>> All state parameters are adopted from the IPFIX-MIB. There, we have:
>>>
>>> +--ipfixSelectionProcessStatsTable(4)
>>>    |
>>>    +--ipfixSelectionProcessStatsEntry(1) 
>>> [ipfixMeteringProcessCacheId,ipfixSelectionProcessIndex,ipfixSelectionProcessSelectorIndex] 
>>>
>>>       |
>>>       +-- r-n Counter64 ipfixSelectionProcessStatsPacketsObserved(1)
>>>       +-- r-n Counter64 ipfixSelectionProcessStatsPacketsDropped(2)
>>>       +-- r-n TimeStamp ipfixSelectionProcessStatsDiscontinuityTime(3)
>>>
>>>
>>> Probably, ipfixSelectionProcessStatsPacketsSelected would make more 
>>> sense in the MIB.
>>>
>>> I assume that the MIB standardization cannot be stopped any more, right?
>>>
>>> So, what do you suggest?
>>
>> I don't know.  Why follow the MIB exactly?  Any idea about the history 
>> of this choice or dropped rather than selected?
> 
> No idea.
> 
> IMO, it does not make sense to have three counters (packetsDropped, 
> packetsSelected, packetsObserved) because two of them are sufficient.
> 
> For the moment, I'll keep packetsObserved and packetsDropped like in the 
> MIB.
> 
> Anyway, as concluded in the other e-mail thread, these counters are not 
> necessarily identical to those exported in the PSAMP statistic reports.

OK, on the other thread I think I want to go as far as saying they are 
unrelated.  Will reply to that thread shortly.


>>>> ======================================================================== 
>>>>
>>>> From Section 4.3.1
>>>> ======================================================================== 
>>>>
>>>> Isn't is possible to generate a cache without any keys?
>>>> This would generate a cache with a single flow?  Do we allow this?
>>> A the moment, there is no restriction which says that a flow key 
>>> field must exist.
>>>
>>> Do you want to add such a restriction?
>>
>> A cache without keys will result in a single flow, which seems a bit 
>> daft.  I guess there is no pressing reason to add this restriction 
>> though.
>  >
>>>> Would it be useful to have two lists, one for keys and one for
>>>> non-keys?
>>> There is an argument for having a single list:
>>>
>>> A Cache implementation may generate records using the specified field 
>>> order. So, the fields occur in the same order as specified in the 
>>> configuration.
>>
>> If we're worried about implementation specifics can't they extend the 
>> YANG model?
> 
> Yes, they can.
> 
>>> With separate lists for flow key fields and non-key fields, we cannot 
>>> arbitrarily mix these two types of fields.
>>>
>>> Is there any advantage of having separate lists?
>>
>> I was just thinking about how you'd specify that at least one key was 
>> required.
> 
> Caches of type "immediate" do not have Flow Keys. So, the restrictions 
> would not apply to such Caches.

Really?  Why not say that Caches of an "immediate" type have no 
non-keys?  For example, the counter fields should not be key fields, 
they make no sense as keys.  They also make no sense with the immediate 
cache.  It seems that an immediate cache is made up of entirely key fields.


>>>> For a permanent cache, update counters should be used and for a
>>>> timeout cache the delta counters should be used.  Where is this
>>>> sort of thing enforced?
>>> This is not enforced. I would not strictly enforce it because strange 
>>> configurations might make sense in specific use cases.
>>
>> OK.
>>
>>> However, we could add a note that totalCounters are typically used 
>>> for "permanent" and deltaCounters for "timeout".
>>>
>>> Could you provide a sentence or two?
>>
>> Perhaps a more general note that not all fields will be appropriate 
>> for all configurations. Is the Exporting Device free to change the fields
>  > if it needs total counters for a permanent cache?
> 
> No.
> 
> I added a new paragraph at the end of 4.3.1. (CacheLayout Class):
> 
>    Note that the use of Information Elements can be restricted to
>    certain Cache Modes or to Flow Key or non-key fields.  Such
>    restrictions may result from Information Element definitions or from
>    device-specific constraints.  According to Section 5, the Monitoring
>    Device MUST notify the user if a Cache field cannot be configured
>    with the given Information Element.
> 
> Ok?

Yes, seems reasonable.


>>>> ======================================================================== 
>>>>
>>>> From Section 4.4.1
>>>> ======================================================================== 
>>>>
>>>>    exportMemberType:  Configures the export member type that 
>>>> corresponds
>>>>       to the ipfixTransportSessionGroupMemberType object in
>>>>       [I-D.ietf-ipfix-mib].  The following parameter values are
>>>>       specified by the configuration data model:
>>>>       *  primary: primary target of the Exporting Process
>>>>       *  secondary: secondary target of the Exporting Process used when
>>>>          the primary target is not reachable
>>>>       *  parallel: parallel exporting to all destinations and files of
>>>>          the Exporting Process
>>>>       *  loadBalancing: load-balancing between different destinations
>>>>          and files of the Exporting Process
>>>>       "parallel" is the default value if this parameter is not
>>>>       configured.  If one destination or file is configured as 
>>>> "primary"
>>>>
>>>> ------------------------------------------------------------------------ 
>>>>
>>>> Comments
>>>> ------------------------------------------------------------------------ 
>>>>
>>>> It's not clear to me that this parameter belongs to the Destination
>>>> class.  The number of restrictions on when you can use the various
>>>> types suggests that we could simplify things.
>>>>
>>>> A Cache can pass flow records to multiple Exporting Processes, so we
>>>> don't really need a parallel option.  This would mean that each
>>>> Exporting Process is responsible for sending one complete set of
>>>> records.  i.e. one record in, one record out.  It could do that, either
>>>> using a redundancy model or with a load-balancing one.
>>>>
>>>> In other words, the exportType would be an Exporting Process attribute
>>>> which would indicate whether the list of Destinations should be load-
>>>> balanced across or treated as Primary, Seondary, Tertiary, etc.
>>> At the moment, the parameter is inline with the IPFIX-MIB:
>>>
>>> ipfixExportMemberType OBJECT-TYPE
>>>     SYNTAX      INTEGER {
>>>                     unknown(0),
>>>                     primary(1),
>>>                     secondary(2),
>>>                     parallel(3),
>>>                     loadBalancing(4)
>>>                 }
>>>     MAX-ACCESS  read-only
>>>     STATUS      current
>>>     DESCRIPTION
>>>         "The type of a member Transport Session in a Transport
>>>         Session group (identified by the value of ipfixExportIndex,
>>>         ipfixObservationDomainId and ipfixMeteringProcessCacheId).
>>>         The following values are valid:
>>>
>>>         unknown(0)
>>>             This value MUST be used if the status of the group
>>>             membership cannot be detected by the equipment. This
>>>             value should be avoided as far as possible.
>>>
>>>         primary(1)
>>>             This value is used for a group member that is used as
>>>             the primary target of an Exporter. Other group members
>>>             (with the same ipfixExportIndex and
>>>             ipfixMeteringProcessCacheId) MUST NOT have the value
>>>             primary(1) but MUST have the value secondary(2).
>>>             This value MUST also be specified if the Exporter does
>>>             not support Transport Session grouping. In this case
>>>             the group contains only one Transport Session.
>>>
>>>         secondary(2)
>>>             This value is used for a group member that is used as a
>>>             secondary target of an Exporter. The Exporter will use
>>>             one of the targets specified as secondary(2) within the
>>>             same Transport Session group when the primary target is
>>>             not reachable.
>>>
>>>         parallel(3)
>>>             This value is used for a group member that is used for
>>>             duplicate exporting i.e., all group members identified> 

>>>             by the ipfixExportIndex are exporting the same Records
>>>             in parallel. This implies that all group members MUST
>>>             have the same membertype parallel(3).
>>>
>>>         loadBalancing(4)
>>>             This value is used for a group member that is used
>>>             as one target for load-balancing. This means that a> 

>>>             Record is sent to one of the group members in this
>>>             group identified by ipfixExportIndex.
>>>             This implies that all group members MUST have the same
>>>             membertype loadBalancing(4)."
>>>     ::= { ipfixExportEntry 2 }
>>>
>>>
>>> I'm open to better proposals.
>>>
>>> "parallel" and "loadBalancing" can be configured once with a single 
>>> ExportingProcess class parameter.
>>>
>>> How would you configure "primary" and "secondary with a single 
>>> Exporting Process parameter?
> 
> This issue is on my todo list.

OK, it seems from a configuration point of view you have three modes. 
Parallel, loadBalancing and fallback.  In fallback mode you try the 
destinations in order until you get a connection.  The mode is a 
property of the Exporting Process and defines how to treat the list of 
destinations.

i.e.

   <Exporter>
     <name>Test 1</name>
     <mode>fallback</mode>
     <destination> ... </destination>
     <destination> ... </destination>
     <destination> ... </destination>
   </Exporter>

So parallel mode sends to all three at once, load balancing takes it in 
turns and fallback treats the first as primary, the second and secondary 
and the third as tertiary.


>>>> ======================================================================== 
>>>>
>>>> From Section 4.7: TransportSession class
>>>> ======================================================================== 
>>>>
>>>> I think you want to add a timestamp for when the session came up
>>>> and when it went down (unless it's still active).
>>> Currently, there is no such parameter in ipfixTransportSessionTable.
>>>
>>> Hm, it seems that we should have asked you to review the IPFIX-MIB :)
>>>
>>> Shall we go beyond IPFIX-MIB and add such state parameters?
>>
>> I don't see why not.  Is it possible to add to MIB after the initial 
>> definition?  Not necessarily now, but perhaps after it's been in use a 
>> while and certain additions are in demand.
>>
>>
>>>> There's no association with the relevant destination.  It seems to me
>>>> that your really want this as part of each Destination or FileReader.
>>>> Specifically, I think you want to separate the Transport Session from
>>>> the File Reader stats.  Potentially there could be a one to many
>>>> relationship of Destination and Transport Sessions, each with their own
>>>> timestamp for coming up and going down.
>>> Currently, TransportSession is a mere copy of 
>>> ipfixTransportSessionTable which exists for both, Exporters and 
>>> Collectors.
>>>
>>> In the case of the _Exporting Process_ , we could move the state 
>>> parameters to the _Destination_ and _FileWriter_ class.
>>>
>>> I think that there can be at most one Transport Session per 
>>> Destination and per FileWriter object. At least at the same point in 
>>> time.
>>> However, it may happen that no Transport Session exists, for example 
>>> in the case of a secondary export destination.
>>
>> Well let's imagine and Exporting process with a Primary Destination 
>> and a Secondary Destination.  The Primary Destination starts a 
>> Transport Session, which works for a while and then is lost.  The 
>> Secondary Destination then opens a Transport Session, which works 
>> fine.  The Exporting Process then manages to reconnect with Primary 
>> Destination and opens a new Transport Session with it and closes the 
>> Secondary Destination's Transport Session.
>>
>> So, we have seen three Transport Sessions and if the Exporting Process 
>> kept a full history of the statistics it could show:
>>
>>    Primary Destination (10.2.100.2 port 4739)
>>      - Transport Session (Open)
>>          Start Time: 10:01.19
>>          End Time: --:--.--
>>          Records sent: 5345
>>          ...
>>     -  Transport Session (Closed)
>>          Start Time: 08:00.00
>>          End Time: 09:45.14
>>          Records sent: 123210
>>
>>    Secondary Destination (10.3.0.15 port 4739)
>>     - Transport Session (Closed)
>>         Start Time: 09:45.16
>>         End Time: 10:01.20
>>         Records sent: 14022
>>
>> I guess it would be up to the implementation to control how much 
>> history was kept.
> 
> But what happens if you remove the primary destination from the 
> configuration?
> 
> If the transport session history is linked to the destination, it will 
> be lost.
 >
> If you want to keep a history of Transport Sessions, it is probably 
> better to keep the information at a distinct place at it is now.

Well you've changed the config, so I think that would be acceptable.


>>> So, we could move a single set of Transport Session state parameters 
>>> to the Destination and FileWriter classes. The parameters would 
>>> appear or not, depending on whether the Transport Session exists.
>>
>> I think this makes sense, otherwise you have to match the Destination 
>> configuration to the Transport Session statistics by matching the 
>> various address and port field.
>>
>>> In the case of the _Collecting Process_ , we can also move the 
>>> Transport Session state parameters to the _Receiver_ and _FileReader_ 
>>> classes. As you say, there can be multiple Transport Sessions per 
>>> Receiver.
>>
>> Well it seems that every field in the Destination Class is present in 
>> the Transport Session Class.  In fact, the Transport Session seems to 
>> be a copy of the Destination Class plus some statistics.  This is not 
>> true of the Receiver class.
> 
> There is an important difference between configuration parameter and 
> state parameters:
> 
> - configuration parameters are not supposed to change dynamically, only
>   after the (re)configuration
> - state parameters may change at any time
> 
> For example, if the user does not configure a source IP address in the 
> Destination class, the source IP address in the IPFIX packets changes if 
> the outgoing interface changes.

OK, that seems fine then.


>> Perhaps the Destination Class could be replaced by a Transport Session 
>> class that has some readonly stats and the Receiver Class can then 
>> contain multiple Transport Sessions.
> 
> IMO, it is wrong to think that Transport Sessions are configurable.
> Transport Sessions represent states which may change dynamically.
> 
> If you cut a link, the Transport Session does not exist any more, but 
> the configuration of the Exporting Process is still there.
> 
> Similarly, if you configure a backup destination, the Transport Session 
> won't exist until the backup is needed.
> 
> What we can do is having one Transport Session per Destination:
> 
>     +------------------------------+
>     | SctpDestination              |
>     +------------------------------+    0..1 +------------------------+
>     | name                         |<>-------| TransportLayerSecurity |
>     | exportMemberType = parallel  |         +------------------------+
>     | ipfixVersion = 10            |
>     | sourceIpAddress[0..*]        |    0..1 +------------------------+
>     | destinationIpAddress[1..*]   |<>-------+ TransportSession       |
>     | destinationPort = 4739|4740  |         +------------------------+
>     | ifName/ifIndex[0..1]         |
>     | sendBufferSize {opt.}        |
>     | rateLimit[0..1]              |
>     | timedReliability = 0         |
>     | numberOfStreams {opt.}       |
>     | numberOfStreams {opt.}       |
>     +------------------------------+
> 
> If the Transport Session exists, the TransportSession object exists and 
> includes the complete state of the current Transport Session (including 
> state parameters which result from configuration, such as 
> destinationIpAddress).
> 
> In the case of the Collecting Process, we can do the same, allowing 
> multiple Transport Sessions per Receiver:
> 
> 
>       +--------------------------+
>       | SctpReceiver             |
>       +--------------------------+    0..1 +------------------------+
>       | name                     |<>-------| TransportLayerSecurity |
>       | localIpAddress[0..*]     |         +------------------------+
>       | localPort = 4739|4740    |
>       | maxAllowedStreams {opt.} |    0..* +------------------------+
>       +--------------------------+<>-------+ TransportSession       |
>                                            +------------------------+
> 
> However, historic Transport Session would not be included.
> 
> Opinions?

I'm happy with that.

Cheers

Andrew


> Regards,
> Gerhard
> 
>> In this case and Exporting Process wouldn't have a history of closed 
>> Transport Sessions, but would have information about the most recent 
>> one per Destination.  The above example would become:
>>
>>    Primary Transport Session
>>      Destination IP address: 10.2.100.2
>>      Destination port:       4739
>>      Status:                 Open
>>      Start Time:             10:01.19
>>      End Time:               --:--.--
>>      Records sent:           5345
>>      ...
>>
>>    Secondary Transport Session
>>      Destination IP address: 10.3.0.15
>>      Destination port:       4739
>>      Status:                 Closed
>>      Start Time:             09:45.16
>>      End Time:               10:01.20
>>      Records sent:           14022
>>
>>
>>> Probably, it makes sense to have a container 
>>> <transportSessionStatistics> which groups the state parameters of a 
>>> Transport Session. This container may appear multiple times in the 
>>> case of the Receiver class, and only once in all other classes.
>>>
>>> Would you agree with these changes?
>>
>> I think we need a little more discussion :)
>>
>> Cheers
>>
>> Andrew
> 

From andrjohn@cisco.com  Thu Mar 11 08:22:34 2010
Return-Path: <andrjohn@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 0B6223A6BA4 for <ipfix@core3.amsl.com>; Thu, 11 Mar 2010 08:22:34 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.599
X-Spam-Level: 
X-Spam-Status: No, score=-5.599 tagged_above=-999 required=5 tests=[AWL=-3.000, BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id aFffW9SmS0Um for <ipfix@core3.amsl.com>; Thu, 11 Mar 2010 08:22:31 -0800 (PST)
Received: from ams-iport-2.cisco.com (ams-iport-2.cisco.com [144.254.224.141]) by core3.amsl.com (Postfix) with ESMTP id 49FB63A6922 for <ipfix@ietf.org>; Thu, 11 Mar 2010 08:21:54 -0800 (PST)
Authentication-Results: ams-iport-2.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AvQAANammEuQ/uCWe2dsb2JhbACaYRUBAQsLJAYcpEmYcYR7BA
X-IronPort-AV: E=Sophos;i="4.49,621,1262563200";  d="scan'208";a="4209829"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-2.cisco.com with ESMTP; 11 Mar 2010 15:48:42 +0000
Received: from cisco.com (mrwint.cisco.com [64.103.71.48]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2BGLw1B024456; Thu, 11 Mar 2010 16:21:58 GMT
Received: from [10.55.163.37] (ams-andrjohn-8714.cisco.com [10.55.163.37]) by cisco.com (8.11.7p3+Sun/8.8.8) with ESMTP id o2BGLvX08748; Thu, 11 Mar 2010 16:21:57 GMT
Message-ID: <4B9918A6.6010407@cisco.com>
Date: Thu, 11 Mar 2010 16:21:58 +0000
From: Andrew Johnson <andrjohn@cisco.com>
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com> <4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com> <4B98C5FC.5030909@net.in.tum.de>
In-Reply-To: <4B98C5FC.5030909@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 11 Mar 2010 16:22:34 -0000

Hi Gerhard


Gerhard Muenz wrote:
[SNIP]
> So, your concerns are not related to the counters included in the 
> configuration data model, but about how PSAMP statistics would be 
> exported in Selection Sequence Statistics Reports.
> 
> We have discussed two possibilities to maintain the necessary statistics:
> - reset all statistics with every reconfiguration of the device
> - maintain serapate statistics for every cache
> We can conclude that the configuration data model does not collide with 
> RFC5476, right?

They don't directly collide, but I don't think they directly align 
either.  I am not happy with the idea that a Metering Process has more 
than one cache.  I think one Metering Process should collect one set of 
statistics from a set of observation points.

That doesn't mean that you can't provide Selector configurations that 
feed more than one cache, it just means that more than one Selection 
Process will be created in response to that Selector configuration and 
each Selector Process will only need one set of stats.


> Nevertheless, I reworked the corresponding parts of the text with the 
> following objectives:
> - no reference to parameter descriptions in the MIB (as proposed by
>   Brian)
> - make adaptations to the state parameters adopted from the MIB where it
>   is useful
> - recommend re-initialization of counters at reconfiguration
> 
> As an example, there exists YANG counter types which are set to zero at 
> initialization (in contrast to MIB counters which start with an 
> arbitrary value). I propose to use these counter types.
> 
> So, please comment the current text:
> 
> 
> 4.2.1.  Selector Class
> 
>     +---------------------------------------+
>     | Selector                              |
>     +---------------------------------------+      1 +-----------------+
>     | name                                  |<>------+ SelectAll/      |
>     | selectorId {readOnly}                 |        | SampCountBased/ |
>     | packetsObserved {readOnly}            |        | SampTimeBased/  |
>     | packetsDropped {readOnly}             |        | SampRandOutOfN/ |
>     | selectorInitializationTime {readOnly} |        | SampUniProb/    |
>     |                                       |        | FilterMatch/    |
>     |                                       |        | FilterHash/     |
>     +---------------------------------------+        +-----------------+
> 
>                          Figure 8: Selector class
> 
> [snip]
> 
>    As state parameters, the Selector class also contains the Selector
>    statistics packetsObserved and packetsDropped as well as
>    selectorInitializationTime:
> 
>    packetsObserved:  The number of packets observed at the input of the
>       Selector since the point in time indicated by
>       selectorDiscontinuityTime.  In contrast to
>       ipfixSelectionProcessStatsPacketsObserved in the IPFIX MIB module
>       [I-D.ietf-ipfix-mib], the counter is initialized with zero.
> 
>    packetsDropped:  The number of packets discarded by the Selector
>       since the point in time indicated by selectorDiscontinuityTime.
>       In contrast to ipfixSelectionProcessStatsPacketsDropped in the
>       IPFIX MIB module [I-D.ietf-ipfix-mib], the counter is initialized
>       with zero.
> 
>    selectorInitializationTime:  The value of sysUpTime at the most
>       recent occasion at which the Selector was re-initialized and at
>       which the counters packetsObserved and packektsDropped were reset
>       to zero.  In contrast to
>       ipfixSelectionProcessStatsDiscontinuityTime in the IPFIX MIB
>       module [I-D.ietf-ipfix-mib], the zero value does not have any
>       specific meaning.
> 
>    The counter values of packetsObserved and packetsDropped MUST both be
>    set to zero at the time indicated by selectorInitializationTime.  It
>    is RECOMMENDED to reset the counters after every reconfiguration of
>    the Selector.

OK.  Clearing the stats on reconfiguration seems like a good idea. 
Perhaps we should make it clear that the statistics captured on a per 
Selector class basis are not the same as those that get exported.  The 
exported ones are captured on a Selection Sequence basis and can be 
reset whenever the implementation feels the need.


> [snip]
> 
> 4.3.  Cache Class
> 
>      +-------------------------------------+
>      | Cache                               |
>      +-------------------------------------+         1 +-------------+
>      | name                                |<>---------| CacheLayout |
>      | cacheMode                           |           +-------------+
>      | maxRecords {opt.}                   |
>      | activeTimeout {opt.} {Cache Modes   | 0..*
>      |   "timeout" and "natural" only}     |---------------+
>      | inactiveTimeout {opt.} {Cache Modes |               | 0..*
>      |   "timeout" and "natural" only}     |               V
>      | exportInterval {opt.} {Cache Mode   |      +------------------+
>      |   "permanent" only}                 |      | ExportingProcess |
>      | activeFlows {readOnly}              |      +------------------+
>      | inactiveFlows {readOnly}            |
>      | cacheDataRecords {readOnly}         |
>      | cacheInitializationTime {readOnly}  |
>      +-------------------------------------+
> 
>                           Figure 11: Cache class
> 
> [snip]
> 
>    As state parameters, the Cache class contains the Metering Process
>    statistics activeFlows, inactiveFlows, and cacheDataRecords, as well
>    as cacheInitializationTime:
> 
>    activeFlows:  The number of Flows currently active in this cache.
>       This value corresponds to ipfixMeteringProcessCacheActiveFlows in
>       the IPFIX MIB module [I-D.ietf-ipfix-mib].
> 
>    inactiveFlows:  The number of Flows currently inactive in this cache.
>       This value corresponds to ipfixMeteringProcessCacheInactiveFlows
>       in the IPFIX MIB module [I-D.ietf-ipfix-mib].
> 
>    cacheDataRecords:  The number of Data Records generated by this Cache
>       since the point in time indicated by cacheIntitializationTime.
>       In contrast to ipfixMeteringProcessDataRecords in the IPFIX MIB
>       module [I-D.ietf-ipfix-mib], the counter is initialized with zero.
> 
>    cacheInitializationTime:  The value of sysUpTime at the most recent
>       occasion at which the Cache was re-initialized and at which the
>       counter cacheDataRecords was reset to zero.  In contrast to
>       ipfixMeteringProcessDiscontinuityTime in the IPFIX MIB module
>       [I-D.ietf-ipfix-mib], the zero value does not have any specific
>       meaning.
> 
>    The counter value of cacheDataRecords MUST be set to zero at the time
>    indicated by cacheInitializationTime.  It is RECOMMENDED to reset the
>    counter after every reconfiguration of the Cache.

Again resetting counters on reconfiguration seems reasonable to me, 
however does changing a timeout value really necessitate resetting the 
counters?


> Question:
> Is it clear what inactive Flows are?
> I can imagine what this number is supposed to by, but I'm not sure.

No it isn't.  So if I have a cache with 1000 entries, some might be 
active, some might be free, ready to be populated with a new flow, and I 
suppose some might be considered ready for export, sitting in Limbo 
until the Exporting Process has the resources to send them.  Are these 
the inactive flows?

Cheers

Andrew

From bclaise@cisco.com  Fri Mar 12 00:28:55 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 016433A6B33 for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 00:28:55 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[AWL=0.000,  BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id YBjB9+Jlhnnf for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 00:28:54 -0800 (PST)
Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id A190B3A6BCC for <ipfix@ietf.org>; Fri, 12 Mar 2010 00:27:06 -0800 (PST)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2C8RAHL021424 for <ipfix@ietf.org>; Fri, 12 Mar 2010 09:27:10 +0100 (CET)
Received: from [10.55.43.57] (ams-bclaise-8718.cisco.com [10.55.43.57]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2C8RAeg001663 for <ipfix@ietf.org>; Fri, 12 Mar 2010 09:27:10 +0100 (CET)
Message-ID: <4B99FADE.7030006@cisco.com>
Date: Fri, 12 Mar 2010 09:27:10 +0100
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: "ipfix@ietf.org" <ipfix@ietf.org>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Subject: [IPFIX] New draft version: draft-claise-ipfix-mediation-protocol-01.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 12 Mar 2010 08:28:55 -0000

Dear all,

As we know, after the mediation problem statement and mediation 
framework, we will have to tackle the IPFIX mediation protocol. We 
received much feedback from Paul Aitken on the 00 version, and a new 
version of the draft has been posted 
http://tools.ietf.org/html/draft-claise-ipfix-mediation-protocol-01

This new version:
- is in line in terms of terminology with the mediation problem 
statement and framework
- a refined solution for the observation domain management
- some new text in the transport session management
- proposal to add a few new information elements
- a completed security section
- some more corrections here and there.

While it's definitively not a complete draft, this is a step in the 
right direction, which will give us some more input for our discussions 
in Anaheim

Regards, Benoit



From bclaise@cisco.com  Fri Mar 12 00:36:10 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 660383A6A50 for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 00:36:10 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[AWL=-0.001, BAYES_00=-2.599, HTML_MESSAGE=0.001]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id DCyts4riZWkM for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 00:36:09 -0800 (PST)
Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id 940F63A691C for <ipfix@ietf.org>; Fri, 12 Mar 2010 00:36:04 -0800 (PST)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2C7vFJ6011116; Fri, 12 Mar 2010 08:57:15 +0100 (CET)
Received: from [10.55.43.57] (ams-bclaise-8718.cisco.com [10.55.43.57]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2C7vDx8001137; Fri, 12 Mar 2010 08:57:14 +0100 (CET)
Message-ID: <4B99F3D9.6070503@cisco.com>
Date: Fri, 12 Mar 2010 08:57:13 +0100
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <C7BC34B3.77A3F%Quittek@nw.neclab.eu> <4B977127.2000903@net.in.tum.de>
In-Reply-To: <4B977127.2000903@net.in.tum.de>
Content-Type: multipart/alternative; boundary="------------080007040705010203030503"
Cc: IETF IPFIX Working Group <ipfix@ietf.org>, Juergen Quittek <Quittek@nw.neclab.eu>
Subject: Re: [IPFIX] draft-ietf-ipfix-configuration-model-05
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 12 Mar 2010 08:36:10 -0000

This is a multi-part message in MIME format.
--------------080007040705010203030503
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Sure.

Regards, Benoit.
>
> I assume that Benoit can give a short presentation as usual.
>
> Regards,
> Gerhard
>
> Juergen Quittek wrote:
>> Hi Gerhard,
>>
>> I appreciate the progress.
>>
>> Are you still planning a presentation in Anaheim?
>> Or can we take the draft off the agenda?
>>
>> Thanks,
>>
>>     Juergen
>>
>> On 08.03.10 11:49  "Gerhard Muenz" <muenz@net.in.tum.de> wrote:
>>
>>> Dear all,
>>>
>>> I've posted draft-ietf-ipfix-configuration-model-05 which shows the
>>> current status of the draft.
>>>
>>> Thanks a lot to the reviewers!
>>>
>>> WGLC comments of Atsushi, Lothar, and Brian have been addressed.
>>>
>>> Some issues raised by Andrew still need to be solved.
>>>
>>> Since I will not go to Anaheim, the mailing list is the right place to
>>> continue the discussion. I hope that the remaining issues can be solved
>>> until the IPFIX session. If so, a -06 version will follow very soon.
>>>
>>> Regards,
>>> Gerhard
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> IPFIX mailing list
>>> IPFIX@ietf.org
>>> https://www.ietf.org/mailman/listinfo/ipfix
>
>
> _______________________________________________
> IPFIX mailing list
> IPFIX@ietf.org
> https://www.ietf.org/mailman/listinfo/ipfix
>    


--------------080007040705010203030503
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <meta content="text/html; charset=ISO-8859-1"
 http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Sure.<br>
<br>
Regards, Benoit.<br>
<blockquote cite="mid:4B977127.2000903@net.in.tum.de" type="cite"><br>
I assume that Benoit can give a short presentation as usual.
  <br>
  <br>
Regards,
  <br>
Gerhard
  <br>
  <br>
Juergen Quittek wrote:
  <br>
  <blockquote type="cite">Hi Gerhard,
    <br>
    <br>
I appreciate the progress.
    <br>
    <br>
Are you still planning a presentation in Anaheim?
    <br>
Or can we take the draft off the agenda?
    <br>
    <br>
Thanks,
    <br>
    <br>
&nbsp;&nbsp;&nbsp; Juergen
    <br>
    <br>
On 08.03.10 11:49&nbsp; "Gerhard Muenz" <a class="moz-txt-link-rfc2396E" href="mailto:muenz@net.in.tum.de">&lt;muenz@net.in.tum.de&gt;</a> wrote:
    <br>
    <br>
    <blockquote type="cite">Dear all,
      <br>
      <br>
I've posted draft-ietf-ipfix-configuration-model-05 which shows the
      <br>
current status of the draft.
      <br>
      <br>
Thanks a lot to the reviewers!
      <br>
      <br>
WGLC comments of Atsushi, Lothar, and Brian have been addressed.
      <br>
      <br>
Some issues raised by Andrew still need to be solved.
      <br>
      <br>
Since I will not go to Anaheim, the mailing list is the right place to
      <br>
continue the discussion. I hope that the remaining issues can be solved
      <br>
until the IPFIX session. If so, a -06 version will follow very soon.
      <br>
      <br>
Regards,
      <br>
Gerhard
      <br>
      <br>
      <br>
      <br>
      <br>
      <br>
_______________________________________________
      <br>
IPFIX mailing list
      <br>
<a class="moz-txt-link-abbreviated" href="mailto:IPFIX@ietf.org">IPFIX@ietf.org</a>
      <br>
<a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/ipfix">https://www.ietf.org/mailman/listinfo/ipfix</a>
      <br>
    </blockquote>
  </blockquote>
  <br>
  <pre wrap="">
<fieldset class="mimeAttachmentHeader"></fieldset>
_______________________________________________
IPFIX mailing list
<a class="moz-txt-link-abbreviated" href="mailto:IPFIX@ietf.org">IPFIX@ietf.org</a>
<a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/ipfix">https://www.ietf.org/mailman/listinfo/ipfix</a>
  </pre>
</blockquote>
<br>
</body>
</html>

--------------080007040705010203030503--

From bclaise@cisco.com  Fri Mar 12 00:47:23 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 755063A69ED for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 00:47:23 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[AWL=0.000,  BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Xd9Y1LdmL-J8 for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 00:47:22 -0800 (PST)
Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id A704E3A6899 for <ipfix@ietf.org>; Fri, 12 Mar 2010 00:47:19 -0800 (PST)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2C8ZcF9024414; Fri, 12 Mar 2010 09:35:38 +0100 (CET)
Received: from [10.55.43.57] (ams-bclaise-8718.cisco.com [10.55.43.57]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2C8ZbF8011496; Fri, 12 Mar 2010 09:35:37 +0100 (CET)
Message-ID: <4B99FCD9.9040105@cisco.com>
Date: Fri, 12 Mar 2010 09:35:37 +0100
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: Lothar Braun <braun@net.in.tum.de>
References: <20100301094119.0F8423A8B2F@core3.amsl.com>	<930BF37B-7E33-440F-88CC-4771705A5C26@net.in.tum.de> <E79680ED-37CD-401B-BB2D-26BEF3F2852A@net.in.tum.de>
In-Reply-To: <E79680ED-37CD-401B-BB2D-26BEF3F2852A@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 8bit
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Fwd: New Version Notification for	draft-braun-core-compressed-ipfix-00
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 12 Mar 2010 08:47:23 -0000

Dear all,

In a nutshell, the draft name says it: compressed IPFIX, taking into 
accounts the hardware, energy, packet size, and transport protocol 
constraints of constrained networks (6lowapp).
The draft contains a series of examples for which a push model just 
makes sense. If we agree on the use case for a push model, then using 
some sort of IPFIX makes sense. Now, we don't want to say that 
compressed IPFIX would be the only protocol in constrained networks. For 
example, it's not suitable for applications where the sensor data 
triggers configuration changes or policy decisions

Regards, Benoit
> Hi all,
>
> we submitted an update on the draft "Compressed IPFIX for smart meters in constrained networks":
>
> http://tools.ietf.org/html/draft-braun-core-compressed-ipfix-01
>
> Best regards,
>    Lothar
>
> On Mar 3, 2010, at 10:33 AM, Lothar Braun wrote:
>
>    
>> Hi all,
>>
>> as you might have seen on the agenda, there is a draft Compressed IPFIX for smart meters in constrained networks:
>>
>> http://tools.ietf.org/id/draft-braun-core-compressed-ipfix-00.txt
>>
>> This draft replaces draft-schmitt-6lowapp-ipfix-ws and defines a modified version of IPFIX that can be used to export sensor data from constrained smart meters in 6lowpan networks. We are working on a 01-version for the meeting (there should only be rather small fixes) and we would like to hear your comments on the document.
>>
>> Best regards,
>> Lothar
>>
>> Begin forwarded message:
>>
>>      
>>> From: IETF I-D Submission Tool<idsubmission@ietf.org>
>>> Date: March 1, 2010 10:41:19 AM GMT+01:00
>>> To: braun@net.in.tum.de
>>> Cc: schmitt@net.in.tum.de,bclaise@cisco.com,carle@net.in.tum.de
>>> Subject: New Version Notification for draft-braun-core-compressed-ipfix-00
>>>
>>>
>>> A new version of I-D, draft-braun-core-compressed-ipfix-00.txt has been successfuly submitted by Lothar Braun and posted to the IETF repository.
>>>
>>> Filename:	 draft-braun-core-compressed-ipfix
>>> Revision:	 00
>>> Title:		 Compressed IPFIX for smart meters in constrained networks
>>> Creation_date:	 2010-03-01
>>> WG ID:		 Independent Submission
>>> Number_of_pages: 25
>>>
>>> Abstract:
>>> This document specifies the Compressed IPFIX protocol that serves for
>>> transmitting measurement data in 6LoWPAN networks [RFC4944].
>>> Compressed IPFIX is derived from IPFIX [RFC5101] and adopted to the
>>> needs of constrained networks.  This documents defines how the
>>> Compressed IPFIX Data and Template Records are transmitted in 6LoWPAN
>>> networks and how Compressed IPFIX data can be converted into
>>> uncompressed IPFIX data.
>>>
>>>
>>>
>>> The IETF Secretariat.
>>>
>>>
>>>        
>> --
>> Lothar Braun
>> Chair for Network Architectures and Services (I8)
>> Department of Informatics
>> Technische Universität München
>> Boltzmannstr. 3, 85748 Garching bei München, Germany
>> Phone:  +49 89 289-18010       Fax: +49 89 289-18033
>> E-mail: braun@net.in.tum.de
>>
>>
>>
>>
>>
>>
>> _______________________________________________
>> IPFIX mailing list
>> IPFIX@ietf.org
>> https://www.ietf.org/mailman/listinfo/ipfix
>>      
> --
> Lothar Braun
> Chair for Network Architectures and Services (I8)
> Department of Informatics
> Technische Universität München
> Boltzmannstr. 3, 85748 Garching bei München, Germany
> Phone:  +49 89 289-18010       Fax: +49 89 289-18033
> E-mail: braun@net.in.tum.de
>
>
>
>
>
>
> _______________________________________________
> IPFIX mailing list
> IPFIX@ietf.org
> https://www.ietf.org/mailman/listinfo/ipfix
>    


From muenz@net.in.tum.de  Fri Mar 12 06:27:53 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 46D673A6CA4 for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 06:27:53 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.182
X-Spam-Level: 
X-Spam-Status: No, score=-2.182 tagged_above=-999 required=5 tests=[AWL=0.067,  BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id f5CB5FN7GxRM for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 06:27:51 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 5A26B3A6CAA for <ipfix@ietf.org>; Fri, 12 Mar 2010 06:09:01 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 7B596208B0EB; Fri, 12 Mar 2010 15:09:05 +0100 (CET)
Message-ID: <4B9A4B15.3010402@net.in.tum.de>
Date: Fri, 12 Mar 2010 15:09:25 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Andrew Johnson <andrjohn@cisco.com>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B9257CE.6040409@cisco.com> <4B98E288.4040204@net.in.tum.de> <4B991389.2090209@cisco.com>
In-Reply-To: <4B991389.2090209@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms090105030803020704030705"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 12 Mar 2010 14:27:53 -0000

This is a cryptographically signed message in MIME format.

--------------ms090105030803020704030705
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Andrew, all,

Andrew Johnson wrote:
> Gerhard Muenz wrote:
>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D =

>>>>>
>>>>> Section 3:
>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D =

>>>>>
>>>>> If a Selection Process is part of a Metering Process and a single
>>>>> Selection Process can feed into more than one Cache, then it is
>>>>> implied that a single Metering Process can have more than once cach=
e.
>>>> Note: The text has slightly changed after Atsushi's review. Now, it =

>>>> explicitly says that a Metering Process may contain multiple=20
>>>> Selection Processes and multiple Caches.
>>> If a Metering Process can have multiple caches and an IPFIX device ca=
n=20
>>> have multiple Metering Processes, how do we associate the caches with=
=20
>>> Metering Processes?  Can this configuration system only be used to=20
>>> configure a single Metering Process within a device?
>> The model does not group Selection Processes and Caches into Metering =

>> Processes. This is done on purpose because there is not a single corre=
ct=20
>> view.
>>
>> All Selection Process and all Caches of a Monitoring Device can be=20
>> regarded as one Metering Process.
>>
>> Alternatively, Selection Processes and Caches which are not directly o=
r=20
>> indirectly linked to each other can be seen as belonging to different =

>> Metering Processes.
>=20
> OK, at this point I think I'm uncomfortable with the idea that a=20
> Metering Process can have more than one cache.  What is the reasoning=20
> behind this decision?

It should be possible that the output of the same Selection Process=20
enters two different Caches.

Example:

           +-> Cache 1 (PSAMP)
OP -> SP -+
           +-> Cache 2 (IPFIX)

It is difficult to say that this corresponds to two MPs.

>>>> Furthermore, it says that only packets observed at a single=20
>>>> Observation Point may enter a Selection Process. IMO, this results=20
>>>> from the PSAMP definitions of Observed Packet Stream and Selection=20
>>>> Process. Please tell me if this conclusion is wrong.
>>> A Selection process should only have a single feed, but can have=20
>>> multiple observers.  This results in the number of possible chains=20
>>> (Selection Sequences) being equal to the number of Selectors that=20
>>> directly feed a cache.
>> Ok. Single feed means single Observation Point as specified in the dra=
ft.
>=20
> No, by "feed" I mean input stream.  Either from an Observation point or=
=20
> the output of a different Selection Process.

Yes, this is what I wanted to say. Therefore, we have "0..1" in -05 (it=20
was "0..*" in -04):


                         +------------------+
                         | ObservationPoint |
                         +------------------+
                              0..1 |
                              =3D=3D=3D=3D |
                                   |
                              0..* V
                         +------------------+
                         | SelectionProcess |
                         +------------------+<-+
                              0..* |  0..1 |   | 0..*
                                   |  =3D=3D=3D=3D +---+
                              0..* V
                         +------------------+
                         | Cache            |
                         +------------------+
                              0..* |
                                   |
                              0..* V
                         +------------------+
                         | ExportingProcess |
                         +------------------+

Sorry, if this was not clear from the beginning.

>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D =

>>>>>
>>>>> From section 4.1:
>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D =

>>>>>
>>>>> Doesn't the direction belong in the Observation Point class, not th=
e
>>>>> Interface or LC class.
>>>> If the current position of the direction parameter bothers you, we=20
>>>> can move it.
>>>>
>>>> However, my reasoning is the following:
>>>> - If it appears in the ObservationPoint class, it looks like the=20
>>>> Observation Point is doing some filtering.
>>>> - If it appears in the Interface/Linecard class, it looks like the=20
>>>> filtering is already done at the interface or linecard.
>>>>
>>>> I thought that the interface/linecard is doing the filtering.
>>> I don't see it as filtering, just part of the definition of the=20
>>> Observation Point.  The Interface is most likely able to send and=20
>>> receive traffic, but the observation point could be on only the send =

>>> or receive channel.  i.e. It's not a input only interface, but the=20
>>> Observation Point could only see input packets.
>>>
>>> Currently you have a 1:1 mapping of OP to interface.  If the directio=
n=20
>>> is moved to the Observation Point then you could have two observation=
=20
>>> points on the same interface, one on the input side and one on the=20
>>> output.
>> With the current model, there is no problem to configure two OPs on th=
e=20
>> same interface:
>>
>>      <observationPoint>
>>        <name>interface ingress</name>
>>        <interface>
>>          <ifIndex>3</ifIndex>
>>          <direction>ingress</direction>
>>        </interface>
>>        ...
>>      </observationPoint>
>>
>>      <observationPoint>
>>        <name>interface egress</name>
>>        <interface>
>>          <ifIndex>3</ifIndex>
>>          <direction>egress</direction>
>>        </interface>
>>        ...
>>      </observationPoint>
>=20
> OK, but my point was always that the interface is not defined by its=20
> direction.
>=20
> These two "interfaces" are actually two observation points on the same =

> interface:
>         <interface>
>           <ifIndex>3</ifIndex>
>           <direction>ingress</direction>
>         </interface>
>=20
>         <interface>
>           <ifIndex>3</ifIndex>
>           <direction>ingress</direction>
>         </interface>

Ok. So it seems to you that <direction> belongs to the definition of the =

interface.

If we moved <direction> to the Observation Point, we could remove the=20
Interface/Linecard containers because there would be only one parameter=20
left:


     +-------------------------------------------------+
     | ObservationPoint                                |
     +-------------------------------------------------+
     | name                                            |
     | observationPointId {readOnly}                   |
     | observationDomainId                             |
     | ifIndex/ifName/entPhysicalIndex/entPhysicalName | 0..*
     | direction                                       |----+
     +-------------------------------------------------+    | 0..*
                                                            V
                                           +--------------------+
                                           | SelectionProcess   |
                                           +--------------------+


Would you agree with this change?

>>>> There is another small advantage: The current position of the=20
>>>> parameter makes it easier to add vendor or implementation-specific=20
>>>> packet sources for which the direction parameter has no meaning, for=
=20
>>>> example:
>>>>
>>>> +------------------+
>>>> | PcapFile         |
>>>> +------------------+
>>>> | filename         |
>>>> +------------------+
>>>>


>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D =

>>>>>
>>>>> From Section 4.3.1
>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D =

>>>>>
>>>>> Isn't is possible to generate a cache without any keys?
>>>>> This would generate a cache with a single flow?  Do we allow this?
>>>> A the moment, there is no restriction which says that a flow key=20
>>>> field must exist.
>>>>
>>>> Do you want to add such a restriction?
>>> A cache without keys will result in a single flow, which seems a bit =

>>> daft.  I guess there is no pressing reason to add this restriction=20
>>> though.
>>  >
>>>>> Would it be useful to have two lists, one for keys and one for
>>>>> non-keys?
>>>> There is an argument for having a single list:
>>>>
>>>> A Cache implementation may generate records using the specified fiel=
d=20
>>>> order. So, the fields occur in the same order as specified in the=20
>>>> configuration.
>>> If we're worried about implementation specifics can't they extend the=
=20
>>> YANG model?
>> Yes, they can.
>>
>>>> With separate lists for flow key fields and non-key fields, we canno=
t=20
>>>> arbitrarily mix these two types of fields.
>>>>
>>>> Is there any advantage of having separate lists?
>>> I was just thinking about how you'd specify that at least one key was=
=20
>>> required.
>> Caches of type "immediate" do not have Flow Keys. So, the restrictions=
=20
>> would not apply to such Caches.
>=20
> Really?  Why not say that Caches of an "immediate" type have no=20
> non-keys?  For example, the counter fields should not be key fields,=20
> they make no sense as keys.  They also make no sense with the immediate=
=20
> cache.  It seems that an immediate cache is made up of entirely key fie=
lds.

IMO, a Packet Report does not contain any Flow information, so it does=20
not make sense to talk about Flow Keys.

In -05, I restricted the <isFlowKey> parameter to Caches generating Flow =

Records because I think that it does not make sense to distinguish Flow=20
Key fields and non-key fields if Packet Reports are generated.

>>>>> For a permanent cache, update counters should be used and for a
>>>>> timeout cache the delta counters should be used.  Where is this
>>>>> sort of thing enforced?
>>>> This is not enforced. I would not strictly enforce it because strang=
e=20
>>>> configurations might make sense in specific use cases.
>>> OK.
>>>
>>>> However, we could add a note that totalCounters are typically used=20
>>>> for "permanent" and deltaCounters for "timeout".
>>>>
>>>> Could you provide a sentence or two?
>>> Perhaps a more general note that not all fields will be appropriate=20
>>> for all configurations. Is the Exporting Device free to change the fi=
elds
>>  > if it needs total counters for a permanent cache?
>>
>> No.
>>
>> I added a new paragraph at the end of 4.3.1. (CacheLayout Class):
>>
>>    Note that the use of Information Elements can be restricted to
>>    certain Cache Modes or to Flow Key or non-key fields.  Such
>>    restrictions may result from Information Element definitions or fro=
m
>>    device-specific constraints.  According to Section 5, the Monitorin=
g
>>    Device MUST notify the user if a Cache field cannot be configured
>>    with the given Information Element.
>>
>> Ok?
>=20
> Yes, seems reasonable.
>=20
>=20
>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D =

>>>>>
>>>>> From Section 4.4.1
>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D =

>>>>>
>>>>>    exportMemberType:  Configures the export member type that=20
>>>>> corresponds
>>>>>       to the ipfixTransportSessionGroupMemberType object in
>>>>>       [I-D.ietf-ipfix-mib].  The following parameter values are
>>>>>       specified by the configuration data model:
>>>>>       *  primary: primary target of the Exporting Process
>>>>>       *  secondary: secondary target of the Exporting Process used =
when
>>>>>          the primary target is not reachable
>>>>>       *  parallel: parallel exporting to all destinations and files=
 of
>>>>>          the Exporting Process
>>>>>       *  loadBalancing: load-balancing between different destinatio=
ns
>>>>>          and files of the Exporting Process
>>>>>       "parallel" is the default value if this parameter is not
>>>>>       configured.  If one destination or file is configured as=20
>>>>> "primary"
>>>>>
>>>>> -------------------------------------------------------------------=
-----=20
>>>>>
>>>>> Comments
>>>>> -------------------------------------------------------------------=
-----=20
>>>>>
>>>>> It's not clear to me that this parameter belongs to the Destination=

>>>>> class.  The number of restrictions on when you can use the various
>>>>> types suggests that we could simplify things.
>>>>>
>>>>> A Cache can pass flow records to multiple Exporting Processes, so w=
e
>>>>> don't really need a parallel option.  This would mean that each
>>>>> Exporting Process is responsible for sending one complete set of
>>>>> records.  i.e. one record in, one record out.  It could do that, ei=
ther
>>>>> using a redundancy model or with a load-balancing one.
>>>>>
>>>>> In other words, the exportType would be an Exporting Process attrib=
ute
>>>>> which would indicate whether the list of Destinations should be loa=
d-
>>>>> balanced across or treated as Primary, Seondary, Tertiary, etc.
>>>> At the moment, the parameter is inline with the IPFIX-MIB:
>>>>
>>>> ipfixExportMemberType OBJECT-TYPE
>>>>     SYNTAX      INTEGER {
>>>>                     unknown(0),
>>>>                     primary(1),
>>>>                     secondary(2),
>>>>                     parallel(3),
>>>>                     loadBalancing(4)
>>>>                 }
>>>>     MAX-ACCESS  read-only
>>>>     STATUS      current
>>>>     DESCRIPTION
>>>>         "The type of a member Transport Session in a Transport
>>>>         Session group (identified by the value of ipfixExportIndex,
>>>>         ipfixObservationDomainId and ipfixMeteringProcessCacheId).
>>>>         The following values are valid:
>>>>
>>>>         unknown(0)
>>>>             This value MUST be used if the status of the group
>>>>             membership cannot be detected by the equipment. This
>>>>             value should be avoided as far as possible.
>>>>
>>>>         primary(1)
>>>>             This value is used for a group member that is used as
>>>>             the primary target of an Exporter. Other group members
>>>>             (with the same ipfixExportIndex and
>>>>             ipfixMeteringProcessCacheId) MUST NOT have the value
>>>>             primary(1) but MUST have the value secondary(2).
>>>>             This value MUST also be specified if the Exporter does
>>>>             not support Transport Session grouping. In this case
>>>>             the group contains only one Transport Session.
>>>>
>>>>         secondary(2)
>>>>             This value is used for a group member that is used as a
>>>>             secondary target of an Exporter. The Exporter will use
>>>>             one of the targets specified as secondary(2) within the
>>>>             same Transport Session group when the primary target is
>>>>             not reachable.
>>>>
>>>>         parallel(3)
>>>>             This value is used for a group member that is used for
>>>>             duplicate exporting i.e., all group members identified> =

>=20
>>>>             by the ipfixExportIndex are exporting the same Records
>>>>             in parallel. This implies that all group members MUST
>>>>             have the same membertype parallel(3).
>>>>
>>>>         loadBalancing(4)
>>>>             This value is used for a group member that is used
>>>>             as one target for load-balancing. This means that a>=20
>=20
>>>>             Record is sent to one of the group members in this
>>>>             group identified by ipfixExportIndex.
>>>>             This implies that all group members MUST have the same
>>>>             membertype loadBalancing(4)."
>>>>     ::=3D { ipfixExportEntry 2 }
>>>>
>>>>
>>>> I'm open to better proposals.
>>>>
>>>> "parallel" and "loadBalancing" can be configured once with a single =

>>>> ExportingProcess class parameter.
>>>>
>>>> How would you configure "primary" and "secondary with a single=20
>>>> Exporting Process parameter?
>> This issue is on my todo list.
>=20
> OK, it seems from a configuration point of view you have three modes.=20
> Parallel, loadBalancing and fallback.  In fallback mode you try the=20
> destinations in order until you get a connection.  The mode is a=20
> property of the Exporting Process and defines how to treat the list of =

> destinations.
>=20
> i.e.
>=20
>    <Exporter>
>      <name>Test 1</name>
>      <mode>fallback</mode>
>      <destination> ... </destination>
>      <destination> ... </destination>
>      <destination> ... </destination>
>    </Exporter>
>=20
> So parallel mode sends to all three at once, load balancing takes it in=
=20
> turns and fallback treats the first as primary, the second and secondar=
y=20
> and the third as tertiary.

Yes, this sounds good!

Cheers,
Gerhard

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms090105030803020704030705
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms090105030803020704030705--

From andrjohn@cisco.com  Fri Mar 12 08:05:26 2010
Return-Path: <andrjohn@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id E6ED63A6AFC for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 08:05:26 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -8.999
X-Spam-Level: 
X-Spam-Status: No, score=-8.999 tagged_above=-999 required=5 tests=[AWL=1.600,  BAYES_00=-2.599, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id RYfkP3Hp1qoG for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 08:05:25 -0800 (PST)
Received: from ams-iport-1.cisco.com (ams-iport-1.cisco.com [144.254.224.140]) by core3.amsl.com (Postfix) with ESMTP id DF0103A6BFE for <ipfix@ietf.org>; Fri, 12 Mar 2010 07:52:37 -0800 (PST)
Authentication-Results: ams-iport-1.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AtwAALLxmUuQ/uCWe2dsb2JhbACaahUBAQsLJAYconiYUYR7BA
X-IronPort-AV: E=Sophos;i="4.49,626,1262563200"; d="scan'208";a="57993854"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 12 Mar 2010 15:52:42 +0000
Received: from cisco.com (mrwint.cisco.com [64.103.71.48]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2CFqg7E015444; Fri, 12 Mar 2010 15:52:42 GMT
Received: from [10.55.163.37] (ams-andrjohn-8714.cisco.com [10.55.163.37]) by cisco.com (8.11.7p3+Sun/8.8.8) with ESMTP id o2CFqfX27282; Fri, 12 Mar 2010 15:52:41 GMT
Message-ID: <4B9A634C.4080503@cisco.com>
Date: Fri, 12 Mar 2010 15:52:44 +0000
From: Andrew Johnson <andrjohn@cisco.com>
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B9257CE.6040409@cisco.com> <4B98E288.4040204@net.in.tum.de> <4B991389.2090209@cisco.com> <4B9A4B15.3010402@net.in.tum.de>
In-Reply-To: <4B9A4B15.3010402@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 12 Mar 2010 16:05:27 -0000

Hi Gerhard

Gerhard Muenz wrote:
> 
> Hi Andrew, all,
> 
> Andrew Johnson wrote:
>> Gerhard Muenz wrote:
>>>>>> ======================================================================== 
>>>>>>
>>>>>> Section 3:
>>>>>> ======================================================================== 
>>>>>>
>>>>>> If a Selection Process is part of a Metering Process and a single
>>>>>> Selection Process can feed into more than one Cache, then it is
>>>>>> implied that a single Metering Process can have more than once cache.
>>>>> Note: The text has slightly changed after Atsushi's review. Now, it 
>>>>> explicitly says that a Metering Process may contain multiple 
>>>>> Selection Processes and multiple Caches.
>>>> If a Metering Process can have multiple caches and an IPFIX device 
>>>> can have multiple Metering Processes, how do we associate the caches 
>>>> with Metering Processes?  Can this configuration system only be used 
>>>> to configure a single Metering Process within a device?
>>> The model does not group Selection Processes and Caches into Metering 
>>> Processes. This is done on purpose because there is not a single 
>>> correct view.
>>>
>>> All Selection Process and all Caches of a Monitoring Device can be 
>>> regarded as one Metering Process.
>>>
>>> Alternatively, Selection Processes and Caches which are not directly 
>>> or indirectly linked to each other can be seen as belonging to 
>>> different Metering Processes.
>>
>> OK, at this point I think I'm uncomfortable with the idea that a 
>> Metering Process can have more than one cache.  What is the reasoning 
>> behind this decision?
> 
> It should be possible that the output of the same Selection Process 
> enters two different Caches.
> 
> Example:
> 
>           +-> Cache 1 (PSAMP)
> OP -> SP -+
>           +-> Cache 2 (IPFIX)
> 
> It is difficult to say that this corresponds to two MPs.

Is the only argument for these two caches being in the same MP is so 
they can share a selection process?

I say that the cache is the heart of a Metering Process and it should 
only have one!  I realise that it may not be explicitly stated in 
current IPFIX documents, but I think it's implied in the Architecture 
document and is the paradigm in which IPFIX and PSAMP have been developed.

It seems that what you really object to is the PSAMP declaration that a 
selection process belongs to a single Metering Process, but as we've 
seen, the statistics held by a Metering Process need to be kept on a 
per-cache basis.

I don't object to the configuration shorthand of using the same config 
across different Metering Processes, but with this model the stats are 
going to be quite different to those needed for PSAMP.  That's fine though!


>>>>> Furthermore, it says that only packets observed at a single 
>>>>> Observation Point may enter a Selection Process. IMO, this results 
>>>>> from the PSAMP definitions of Observed Packet Stream and Selection 
>>>>> Process. Please tell me if this conclusion is wrong.
>>>> A Selection process should only have a single feed, but can have 
>>>> multiple observers.  This results in the number of possible chains 
>>>> (Selection Sequences) being equal to the number of Selectors that 
>>>> directly feed a cache.
>>> Ok. Single feed means single Observation Point as specified in the 
>>> draft.
>>
>> No, by "feed" I mean input stream.  Either from an Observation point 
>> or the output of a different Selection Process.
> 
> Yes, this is what I wanted to say. Therefore, we have "0..1" in -05 (it 
> was "0..*" in -04):
> 
> 
>                         +------------------+
>                         | ObservationPoint |
>                         +------------------+
>                              0..1 |
>                              ==== |
>                                   |
>                              0..* V
>                         +------------------+
>                         | SelectionProcess |
>                         +------------------+<-+
>                              0..* |  0..1 |   | 0..*
>                                   |  ==== +---+
>                              0..* V
>                         +------------------+
>                         | Cache            |
>                         +------------------+
>                              0..* |
>                                   |
>                              0..* V
>                         +------------------+
>                         | ExportingProcess |
>                         +------------------+
> 
> Sorry, if this was not clear from the beginning.

This has always been clear... I'm not sure what the original points were 
now.


>>>>>> ======================================================================== 
>>>>>>
>>>>>> From section 4.1:
>>>>>> ======================================================================== 
>>>>>>
>>>>>> Doesn't the direction belong in the Observation Point class, not the
>>>>>> Interface or LC class.
>>>>> If the current position of the direction parameter bothers you, we 
>>>>> can move it.
>>>>>
>>>>> However, my reasoning is the following:
>>>>> - If it appears in the ObservationPoint class, it looks like the 
>>>>> Observation Point is doing some filtering.
>>>>> - If it appears in the Interface/Linecard class, it looks like the 
>>>>> filtering is already done at the interface or linecard.
>>>>>
>>>>> I thought that the interface/linecard is doing the filtering.
>>>> I don't see it as filtering, just part of the definition of the 
>>>> Observation Point.  The Interface is most likely able to send and 
>>>> receive traffic, but the observation point could be on only the send 
>>>> or receive channel.  i.e. It's not a input only interface, but the 
>>>> Observation Point could only see input packets.
>>>>
>>>> Currently you have a 1:1 mapping of OP to interface.  If the 
>>>> direction is moved to the Observation Point then you could have two 
>>>> observation points on the same interface, one on the input side and 
>>>> one on the output.
>>> With the current model, there is no problem to configure two OPs on 
>>> the same interface:
>>>
>>>      <observationPoint>
>>>        <name>interface ingress</name>
>>>        <interface>
>>>          <ifIndex>3</ifIndex>
>>>          <direction>ingress</direction>
>>>        </interface>
>>>        ...
>>>      </observationPoint>
>>>
>>>      <observationPoint>
>>>        <name>interface egress</name>
>>>        <interface>
>>>          <ifIndex>3</ifIndex>
>>>          <direction>egress</direction>
>>>        </interface>
>>>        ...
>>>      </observationPoint>
>>
>> OK, but my point was always that the interface is not defined by its 
>> direction.
>>
>> These two "interfaces" are actually two observation points on the same 
>> interface:
>>         <interface>
>>           <ifIndex>3</ifIndex>
>>           <direction>ingress</direction>
>>         </interface>
>>
>>         <interface>
>>           <ifIndex>3</ifIndex>
>>           <direction>ingress</direction>
>>         </interface>
> 
> Ok. So it seems to you that <direction> belongs to the definition of the 
> interface.
> 
> If we moved <direction> to the Observation Point, we could remove the 
> Interface/Linecard containers because there would be only one parameter 
> left:
> 
> 
>     +-------------------------------------------------+
>     | ObservationPoint                                |
>     +-------------------------------------------------+
>     | name                                            |
>     | observationPointId {readOnly}                   |
>     | observationDomainId                             |
>     | ifIndex/ifName/entPhysicalIndex/entPhysicalName | 0..*
>     | direction                                       |----+
>     +-------------------------------------------------+    | 0..*
>                                                            V
>                                           +--------------------+
>                                           | SelectionProcess   |
>                                           +--------------------+
> 
> 
> Would you agree with this change?

Yes, this seems fine.


>>>>> There is another small advantage: The current position of the 
>>>>> parameter makes it easier to add vendor or implementation-specific 
>>>>> packet sources for which the direction parameter has no meaning, 
>>>>> for example:
>>>>>
>>>>> +------------------+
>>>>> | PcapFile         |
>>>>> +------------------+
>>>>> | filename         |
>>>>> +------------------+
>>>>>
> 
> 
>>>>>> ======================================================================== 
>>>>>>
>>>>>> From Section 4.3.1
>>>>>> ======================================================================== 
>>>>>>
>>>>>> Isn't is possible to generate a cache without any keys?
>>>>>> This would generate a cache with a single flow?  Do we allow this?
>>>>> A the moment, there is no restriction which says that a flow key 
>>>>> field must exist.
>>>>>
>>>>> Do you want to add such a restriction?
>>>> A cache without keys will result in a single flow, which seems a bit 
>>>> daft.  I guess there is no pressing reason to add this restriction 
>>>> though.
>>>  >
>>>>>> Would it be useful to have two lists, one for keys and one for
>>>>>> non-keys?
>>>>> There is an argument for having a single list:
>>>>>
>>>>> A Cache implementation may generate records using the specified 
>>>>> field order. So, the fields occur in the same order as specified in 
>>>>> the configuration.
>>>> If we're worried about implementation specifics can't they extend 
>>>> the YANG model?
>>> Yes, they can.
>>>
>>>>> With separate lists for flow key fields and non-key fields, we 
>>>>> cannot arbitrarily mix these two types of fields.
>>>>>
>>>>> Is there any advantage of having separate lists?
>>>> I was just thinking about how you'd specify that at least one key 
>>>> was required.
>>> Caches of type "immediate" do not have Flow Keys. So, the 
>>> restrictions would not apply to such Caches.
>>
>> Really?  Why not say that Caches of an "immediate" type have no 
>> non-keys?  For example, the counter fields should not be key fields, 
>> they make no sense as keys.  They also make no sense with the 
>> immediate cache.  It seems that an immediate cache is made up of 
>> entirely key fields.
> 
> IMO, a Packet Report does not contain any Flow information, so it does 
> not make sense to talk about Flow Keys.
> 
> In -05, I restricted the <isFlowKey> parameter to Caches generating Flow 
> Records because I think that it does not make sense to distinguish Flow 
> Key fields and non-key fields if Packet Reports are generated.

Ah I see.  I guess it makes sense to simply not say anything about 
whether it is a key or not.  So these fields are not keys, but neither 
are they non-keys.  The are simply collected fields.  Certain fields 
only make sense for flows and not for packet reports, which might be 
worth mentioning.


>>>>>> For a permanent cache, update counters should be used and for a
>>>>>> timeout cache the delta counters should be used.  Where is this
>>>>>> sort of thing enforced?
>>>>> This is not enforced. I would not strictly enforce it because 
>>>>> strange configurations might make sense in specific use cases.
>>>> OK.
>>>>
>>>>> However, we could add a note that totalCounters are typically used 
>>>>> for "permanent" and deltaCounters for "timeout".
>>>>>
>>>>> Could you provide a sentence or two?
>>>> Perhaps a more general note that not all fields will be appropriate 
>>>> for all configurations. Is the Exporting Device free to change the 
>>>> fields
>>>  > if it needs total counters for a permanent cache?
>>>
>>> No.
>>>
>>> I added a new paragraph at the end of 4.3.1. (CacheLayout Class):
>>>
>>>    Note that the use of Information Elements can be restricted to
>>>    certain Cache Modes or to Flow Key or non-key fields.  Such
>>>    restrictions may result from Information Element definitions or from
>>>    device-specific constraints.  According to Section 5, the Monitoring
>>>    Device MUST notify the user if a Cache field cannot be configured
>>>    with the given Information Element.
>>>
>>> Ok?
>>
>> Yes, seems reasonable.
>>
>>
>>>>>> ======================================================================== 
>>>>>>
>>>>>> From Section 4.4.1
>>>>>> ======================================================================== 
>>>>>>
>>>>>>    exportMemberType:  Configures the export member type that 
>>>>>> corresponds
>>>>>>       to the ipfixTransportSessionGroupMemberType object in
>>>>>>       [I-D.ietf-ipfix-mib].  The following parameter values are
>>>>>>       specified by the configuration data model:
>>>>>>       *  primary: primary target of the Exporting Process
>>>>>>       *  secondary: secondary target of the Exporting Process used 
>>>>>> when
>>>>>>          the primary target is not reachable
>>>>>>       *  parallel: parallel exporting to all destinations and 
>>>>>> files of
>>>>>>          the Exporting Process
>>>>>>       *  loadBalancing: load-balancing between different destinations
>>>>>>          and files of the Exporting Process
>>>>>>       "parallel" is the default value if this parameter is not
>>>>>>       configured.  If one destination or file is configured as 
>>>>>> "primary"
>>>>>>
>>>>>> ------------------------------------------------------------------------ 
>>>>>>
>>>>>> Comments
>>>>>> ------------------------------------------------------------------------ 
>>>>>>
>>>>>> It's not clear to me that this parameter belongs to the Destination
>>>>>> class.  The number of restrictions on when you can use the various
>>>>>> types suggests that we could simplify things.
>>>>>>
>>>>>> A Cache can pass flow records to multiple Exporting Processes, so we
>>>>>> don't really need a parallel option.  This would mean that each
>>>>>> Exporting Process is responsible for sending one complete set of
>>>>>> records.  i.e. one record in, one record out.  It could do that, 
>>>>>> either
>>>>>> using a redundancy model or with a load-balancing one.
>>>>>>
>>>>>> In other words, the exportType would be an Exporting Process 
>>>>>> attribute
>>>>>> which would indicate whether the list of Destinations should be load-
>>>>>> balanced across or treated as Primary, Seondary, Tertiary, etc.
>>>>> At the moment, the parameter is inline with the IPFIX-MIB:
>>>>>
>>>>> ipfixExportMemberType OBJECT-TYPE
>>>>>     SYNTAX      INTEGER {
>>>>>                     unknown(0),
>>>>>                     primary(1),
>>>>>                     secondary(2),
>>>>>                     parallel(3),
>>>>>                     loadBalancing(4)
>>>>>                 }
>>>>>     MAX-ACCESS  read-only
>>>>>     STATUS      current
>>>>>     DESCRIPTION
>>>>>         "The type of a member Transport Session in a Transport
>>>>>         Session group (identified by the value of ipfixExportIndex,
>>>>>         ipfixObservationDomainId and ipfixMeteringProcessCacheId).
>>>>>         The following values are valid:
>>>>>
>>>>>         unknown(0)
>>>>>             This value MUST be used if the status of the group
>>>>>             membership cannot be detected by the equipment. This
>>>>>             value should be avoided as far as possible.
>>>>>
>>>>>         primary(1)
>>>>>             This value is used for a group member that is used as
>>>>>             the primary target of an Exporter. Other group members
>>>>>             (with the same ipfixExportIndex and
>>>>>             ipfixMeteringProcessCacheId) MUST NOT have the value
>>>>>             primary(1) but MUST have the value secondary(2).
>>>>>             This value MUST also be specified if the Exporter does
>>>>>             not support Transport Session grouping. In this case
>>>>>             the group contains only one Transport Session.
>>>>>
>>>>>         secondary(2)
>>>>>             This value is used for a group member that is used as a
>>>>>             secondary target of an Exporter. The Exporter will use
>>>>>             one of the targets specified as secondary(2) within the
>>>>>             same Transport Session group when the primary target is
>>>>>             not reachable.
>>>>>
>>>>>         parallel(3)
>>>>>             This value is used for a group member that is used for
>>>>>             duplicate exporting i.e., all group members identified> 
>>
>>>>>             by the ipfixExportIndex are exporting the same Records
>>>>>             in parallel. This implies that all group members MUST
>>>>>             have the same membertype parallel(3).
>>>>>
>>>>>         loadBalancing(4)
>>>>>             This value is used for a group member that is used
>>>>>             as one target for load-balancing. This means that a> 
>>
>>>>>             Record is sent to one of the group members in this
>>>>>             group identified by ipfixExportIndex.
>>>>>             This implies that all group members MUST have the same
>>>>>             membertype loadBalancing(4)."
>>>>>     ::= { ipfixExportEntry 2 }
>>>>>
>>>>>
>>>>> I'm open to better proposals.
>>>>>
>>>>> "parallel" and "loadBalancing" can be configured once with a single 
>>>>> ExportingProcess class parameter.
>>>>>
>>>>> How would you configure "primary" and "secondary with a single 
>>>>> Exporting Process parameter?
>>> This issue is on my todo list.
>>
>> OK, it seems from a configuration point of view you have three modes. 
>> Parallel, loadBalancing and fallback.  In fallback mode you try the 
>> destinations in order until you get a connection.  The mode is a 
>> property of the Exporting Process and defines how to treat the list of 
>> destinations.
>>
>> i.e.
>>
>>    <Exporter>
>>      <name>Test 1</name>
>>      <mode>fallback</mode>
>>      <destination> ... </destination>
>>      <destination> ... </destination>
>>      <destination> ... </destination>
>>    </Exporter>
>>
>> So parallel mode sends to all three at once, load balancing takes it 
>> in turns and fallback treats the first as primary, the second and 
>> secondary and the third as tertiary.
> 
> Yes, this sounds good!

Great.

Thanks

Andrew

From trammell@tik.ee.ethz.ch  Fri Mar 12 08:45:54 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 115753A68B0 for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 08:45:54 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.599
X-Spam-Level: 
X-Spam-Status: No, score=-6.599 tagged_above=-999 required=5 tests=[AWL=0.000,  BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id RPCunjSLswvZ for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 08:45:52 -0800 (PST)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id 922A73A68CC for <ipfix@ietf.org>; Fri, 12 Mar 2010 08:35:24 -0800 (PST)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id 02200D93F8; Fri, 12 Mar 2010 17:35:29 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 5+e741muZGTT; Fri, 12 Mar 2010 17:35:28 +0100 (MET)
Received: from pb-10072.ethz.ch (pb-10072.ethz.ch [82.130.103.195]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id BA8CCD93A9; Fri, 12 Mar 2010 17:35:28 +0100 (MET)
Mime-Version: 1.0 (Apple Message framework v1077)
Content-Type: text/plain; charset=us-ascii
From: Brian Trammell <trammell@tik.ee.ethz.ch>
In-Reply-To: <4B9A634C.4080503@cisco.com>
Date: Fri, 12 Mar 2010 17:35:28 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <3BC9C9EF-4A1C-4A48-9AB6-E7CE5EF0C0BC@tik.ee.ethz.ch>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B9257CE.6040409@cisco.com> <4B98E288.4040204@net.in.tum.de> <4B991389.2090209@cisco.com> <4B9A4B15.3010402@net.in.tum.de> <4B9A634C.4080503@cisco.com>
To: Andrew Johnson <andrjohn@cisco.com>
X-Mailer: Apple Mail (2.1077)
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 12 Mar 2010 16:45:54 -0000

hi Andrew, Gerhard,

Let me step in here quickly, as I can easily see a place for a =
multicache MP.. the following is something I was asked to do once upon a =
time:

Suppose we have an MP at an observation point in a security monitoring =
context. This MP is driving two separate processes: one which filters =
traffic according to some current set of criteria defining "bad" =
traffic, then displays in "real time" (~5s delay) this "bad" traffic =
volume (the fast stream), and one which prefilters and stores full flow =
data for retrospective analysis (the full stream). The fast stream gets =
used to drive a display and maybe turn on a blinkenlight somewhere, then =
it gets thrown away. The full stream has richer data about each flow and =
gets stored and used by the analysis tools for doing actual work.

It seemed to me at the time that the right way to handle this situation =
is two caches in the same MP: one with very fast active timeouts (we =
don't care about termination conditions, we just want to see flow-level =
volumes), and one with reasonable timeouts (because the analyses are =
simpler when you don't have to stitch together all the flows from active =
timeouts.) Everything else -> observation point, selection process, even =
the EP and CP, are the same.

Or is this two MPs in your reading of the architecture?

Cheers,

Brian

On Mar 12, 2010, at 4:52 PM, Andrew Johnson wrote:

> Hi Gerhard
>=20
> Gerhard Muenz wrote:
>> Hi Andrew, all,
>> Andrew Johnson wrote:
>>> Gerhard Muenz wrote:
>>>>>>> =
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=20
>>>>>>> Section 3:
>>>>>>> =
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=20
>>>>>>> If a Selection Process is part of a Metering Process and a =
single
>>>>>>> Selection Process can feed into more than one Cache, then it is
>>>>>>> implied that a single Metering Process can have more than once =
cache.
>>>>>> Note: The text has slightly changed after Atsushi's review. Now, =
it explicitly says that a Metering Process may contain multiple =
Selection Processes and multiple Caches.
>>>>> If a Metering Process can have multiple caches and an IPFIX device =
can have multiple Metering Processes, how do we associate the caches =
with Metering Processes?  Can this configuration system only be used to =
configure a single Metering Process within a device?
>>>> The model does not group Selection Processes and Caches into =
Metering Processes. This is done on purpose because there is not a =
single correct view.
>>>>=20
>>>> All Selection Process and all Caches of a Monitoring Device can be =
regarded as one Metering Process.
>>>>=20
>>>> Alternatively, Selection Processes and Caches which are not =
directly or indirectly linked to each other can be seen as belonging to =
different Metering Processes.
>>>=20
>>> OK, at this point I think I'm uncomfortable with the idea that a =
Metering Process can have more than one cache.  What is the reasoning =
behind this decision?
>> It should be possible that the output of the same Selection Process =
enters two different Caches.
>> Example:
>>          +-> Cache 1 (PSAMP)
>> OP -> SP -+
>>          +-> Cache 2 (IPFIX)
>> It is difficult to say that this corresponds to two MPs.
>=20
> Is the only argument for these two caches being in the same MP is so =
they can share a selection process?
>=20
> I say that the cache is the heart of a Metering Process and it should =
only have one!  I realise that it may not be explicitly stated in =
current IPFIX documents, but I think it's implied in the Architecture =
document and is the paradigm in which IPFIX and PSAMP have been =
developed.
>=20
> It seems that what you really object to is the PSAMP declaration that =
a selection process belongs to a single Metering Process, but as we've =
seen, the statistics held by a Metering Process need to be kept on a =
per-cache basis.
>=20
> I don't object to the configuration shorthand of using the same config =
across different Metering Processes, but with this model the stats are =
going to be quite different to those needed for PSAMP.  That's fine =
though!
>=20
>=20
>>>>>> Furthermore, it says that only packets observed at a single =
Observation Point may enter a Selection Process. IMO, this results from =
the PSAMP definitions of Observed Packet Stream and Selection Process. =
Please tell me if this conclusion is wrong.
>>>>> A Selection process should only have a single feed, but can have =
multiple observers.  This results in the number of possible chains =
(Selection Sequences) being equal to the number of Selectors that =
directly feed a cache.
>>>> Ok. Single feed means single Observation Point as specified in the =
draft.
>>>=20
>>> No, by "feed" I mean input stream.  Either from an Observation point =
or the output of a different Selection Process.
>> Yes, this is what I wanted to say. Therefore, we have "0..1" in -05 =
(it was "0..*" in -04):
>>                        +------------------+
>>                        | ObservationPoint |
>>                        +------------------+
>>                             0..1 |
>>                             =3D=3D=3D=3D |
>>                                  |
>>                             0..* V
>>                        +------------------+
>>                        | SelectionProcess |
>>                        +------------------+<-+
>>                             0..* |  0..1 |   | 0..*
>>                                  |  =3D=3D=3D=3D +---+
>>                             0..* V
>>                        +------------------+
>>                        | Cache            |
>>                        +------------------+
>>                             0..* |
>>                                  |
>>                             0..* V
>>                        +------------------+
>>                        | ExportingProcess |
>>                        +------------------+
>> Sorry, if this was not clear from the beginning.
>=20
> This has always been clear... I'm not sure what the original points =
were now.
>=20
>=20
>>>>>>> =
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=20
>>>>>>> =46rom section 4.1:
>>>>>>> =
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=20
>>>>>>> Doesn't the direction belong in the Observation Point class, not =
the
>>>>>>> Interface or LC class.
>>>>>> If the current position of the direction parameter bothers you, =
we can move it.
>>>>>>=20
>>>>>> However, my reasoning is the following:
>>>>>> - If it appears in the ObservationPoint class, it looks like the =
Observation Point is doing some filtering.
>>>>>> - If it appears in the Interface/Linecard class, it looks like =
the filtering is already done at the interface or linecard.
>>>>>>=20
>>>>>> I thought that the interface/linecard is doing the filtering.
>>>>> I don't see it as filtering, just part of the definition of the =
Observation Point.  The Interface is most likely able to send and =
receive traffic, but the observation point could be on only the send or =
receive channel.  i.e. It's not a input only interface, but the =
Observation Point could only see input packets.
>>>>>=20
>>>>> Currently you have a 1:1 mapping of OP to interface.  If the =
direction is moved to the Observation Point then you could have two =
observation points on the same interface, one on the input side and one =
on the output.
>>>> With the current model, there is no problem to configure two OPs on =
the same interface:
>>>>=20
>>>>     <observationPoint>
>>>>       <name>interface ingress</name>
>>>>       <interface>
>>>>         <ifIndex>3</ifIndex>
>>>>         <direction>ingress</direction>
>>>>       </interface>
>>>>       ...
>>>>     </observationPoint>
>>>>=20
>>>>     <observationPoint>
>>>>       <name>interface egress</name>
>>>>       <interface>
>>>>         <ifIndex>3</ifIndex>
>>>>         <direction>egress</direction>
>>>>       </interface>
>>>>       ...
>>>>     </observationPoint>
>>>=20
>>> OK, but my point was always that the interface is not defined by its =
direction.
>>>=20
>>> These two "interfaces" are actually two observation points on the =
same interface:
>>>        <interface>
>>>          <ifIndex>3</ifIndex>
>>>          <direction>ingress</direction>
>>>        </interface>
>>>=20
>>>        <interface>
>>>          <ifIndex>3</ifIndex>
>>>          <direction>ingress</direction>
>>>        </interface>
>> Ok. So it seems to you that <direction> belongs to the definition of =
the interface.
>> If we moved <direction> to the Observation Point, we could remove the =
Interface/Linecard containers because there would be only one parameter =
left:
>>    +-------------------------------------------------+
>>    | ObservationPoint                                |
>>    +-------------------------------------------------+
>>    | name                                            |
>>    | observationPointId {readOnly}                   |
>>    | observationDomainId                             |
>>    | ifIndex/ifName/entPhysicalIndex/entPhysicalName | 0..*
>>    | direction                                       |----+
>>    +-------------------------------------------------+    | 0..*
>>                                                           V
>>                                          +--------------------+
>>                                          | SelectionProcess   |
>>                                          +--------------------+
>> Would you agree with this change?
>=20
> Yes, this seems fine.
>=20
>=20
>>>>>> There is another small advantage: The current position of the =
parameter makes it easier to add vendor or implementation-specific =
packet sources for which the direction parameter has no meaning, for =
example:
>>>>>>=20
>>>>>> +------------------+
>>>>>> | PcapFile         |
>>>>>> +------------------+
>>>>>> | filename         |
>>>>>> +------------------+
>>>>>>=20
>>>>>>> =
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=20
>>>>>>> =46rom Section 4.3.1
>>>>>>> =
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=20
>>>>>>> Isn't is possible to generate a cache without any keys?
>>>>>>> This would generate a cache with a single flow?  Do we allow =
this?
>>>>>> A the moment, there is no restriction which says that a flow key =
field must exist.
>>>>>>=20
>>>>>> Do you want to add such a restriction?
>>>>> A cache without keys will result in a single flow, which seems a =
bit daft.  I guess there is no pressing reason to add this restriction =
though.
>>>> >
>>>>>>> Would it be useful to have two lists, one for keys and one for
>>>>>>> non-keys?
>>>>>> There is an argument for having a single list:
>>>>>>=20
>>>>>> A Cache implementation may generate records using the specified =
field order. So, the fields occur in the same order as specified in the =
configuration.
>>>>> If we're worried about implementation specifics can't they extend =
the YANG model?
>>>> Yes, they can.
>>>>=20
>>>>>> With separate lists for flow key fields and non-key fields, we =
cannot arbitrarily mix these two types of fields.
>>>>>>=20
>>>>>> Is there any advantage of having separate lists?
>>>>> I was just thinking about how you'd specify that at least one key =
was required.
>>>> Caches of type "immediate" do not have Flow Keys. So, the =
restrictions would not apply to such Caches.
>>>=20
>>> Really?  Why not say that Caches of an "immediate" type have no =
non-keys?  For example, the counter fields should not be key fields, =
they make no sense as keys.  They also make no sense with the immediate =
cache.  It seems that an immediate cache is made up of entirely key =
fields.
>> IMO, a Packet Report does not contain any Flow information, so it =
does not make sense to talk about Flow Keys.
>> In -05, I restricted the <isFlowKey> parameter to Caches generating =
Flow Records because I think that it does not make sense to distinguish =
Flow Key fields and non-key fields if Packet Reports are generated.
>=20
> Ah I see.  I guess it makes sense to simply not say anything about =
whether it is a key or not.  So these fields are not keys, but neither =
are they non-keys.  The are simply collected fields.  Certain fields =
only make sense for flows and not for packet reports, which might be =
worth mentioning.
>=20
>=20
>>>>>>> For a permanent cache, update counters should be used and for a
>>>>>>> timeout cache the delta counters should be used.  Where is this
>>>>>>> sort of thing enforced?
>>>>>> This is not enforced. I would not strictly enforce it because =
strange configurations might make sense in specific use cases.
>>>>> OK.
>>>>>=20
>>>>>> However, we could add a note that totalCounters are typically =
used for "permanent" and deltaCounters for "timeout".
>>>>>>=20
>>>>>> Could you provide a sentence or two?
>>>>> Perhaps a more general note that not all fields will be =
appropriate for all configurations. Is the Exporting Device free to =
change the fields
>>>> > if it needs total counters for a permanent cache?
>>>>=20
>>>> No.
>>>>=20
>>>> I added a new paragraph at the end of 4.3.1. (CacheLayout Class):
>>>>=20
>>>>   Note that the use of Information Elements can be restricted to
>>>>   certain Cache Modes or to Flow Key or non-key fields.  Such
>>>>   restrictions may result from Information Element definitions or =
from
>>>>   device-specific constraints.  According to Section 5, the =
Monitoring
>>>>   Device MUST notify the user if a Cache field cannot be configured
>>>>   with the given Information Element.
>>>>=20
>>>> Ok?
>>>=20
>>> Yes, seems reasonable.
>>>=20
>>>=20
>>>>>>> =
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=20
>>>>>>> =46rom Section 4.4.1
>>>>>>> =
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=20
>>>>>>>   exportMemberType:  Configures the export member type that =
corresponds
>>>>>>>      to the ipfixTransportSessionGroupMemberType object in
>>>>>>>      [I-D.ietf-ipfix-mib].  The following parameter values are
>>>>>>>      specified by the configuration data model:
>>>>>>>      *  primary: primary target of the Exporting Process
>>>>>>>      *  secondary: secondary target of the Exporting Process =
used when
>>>>>>>         the primary target is not reachable
>>>>>>>      *  parallel: parallel exporting to all destinations and =
files of
>>>>>>>         the Exporting Process
>>>>>>>      *  loadBalancing: load-balancing between different =
destinations
>>>>>>>         and files of the Exporting Process
>>>>>>>      "parallel" is the default value if this parameter is not
>>>>>>>      configured.  If one destination or file is configured as =
"primary"
>>>>>>>=20
>>>>>>> =
------------------------------------------------------------------------=20=

>>>>>>> Comments
>>>>>>> =
------------------------------------------------------------------------=20=

>>>>>>> It's not clear to me that this parameter belongs to the =
Destination
>>>>>>> class.  The number of restrictions on when you can use the =
various
>>>>>>> types suggests that we could simplify things.
>>>>>>>=20
>>>>>>> A Cache can pass flow records to multiple Exporting Processes, =
so we
>>>>>>> don't really need a parallel option.  This would mean that each
>>>>>>> Exporting Process is responsible for sending one complete set of
>>>>>>> records.  i.e. one record in, one record out.  It could do that, =
either
>>>>>>> using a redundancy model or with a load-balancing one.
>>>>>>>=20
>>>>>>> In other words, the exportType would be an Exporting Process =
attribute
>>>>>>> which would indicate whether the list of Destinations should be =
load-
>>>>>>> balanced across or treated as Primary, Seondary, Tertiary, etc.
>>>>>> At the moment, the parameter is inline with the IPFIX-MIB:
>>>>>>=20
>>>>>> ipfixExportMemberType OBJECT-TYPE
>>>>>>    SYNTAX      INTEGER {
>>>>>>                    unknown(0),
>>>>>>                    primary(1),
>>>>>>                    secondary(2),
>>>>>>                    parallel(3),
>>>>>>                    loadBalancing(4)
>>>>>>                }
>>>>>>    MAX-ACCESS  read-only
>>>>>>    STATUS      current
>>>>>>    DESCRIPTION
>>>>>>        "The type of a member Transport Session in a Transport
>>>>>>        Session group (identified by the value of =
ipfixExportIndex,
>>>>>>        ipfixObservationDomainId and ipfixMeteringProcessCacheId).
>>>>>>        The following values are valid:
>>>>>>=20
>>>>>>        unknown(0)
>>>>>>            This value MUST be used if the status of the group
>>>>>>            membership cannot be detected by the equipment. This
>>>>>>            value should be avoided as far as possible.
>>>>>>=20
>>>>>>        primary(1)
>>>>>>            This value is used for a group member that is used as
>>>>>>            the primary target of an Exporter. Other group members
>>>>>>            (with the same ipfixExportIndex and
>>>>>>            ipfixMeteringProcessCacheId) MUST NOT have the value
>>>>>>            primary(1) but MUST have the value secondary(2).
>>>>>>            This value MUST also be specified if the Exporter does
>>>>>>            not support Transport Session grouping. In this case
>>>>>>            the group contains only one Transport Session.
>>>>>>=20
>>>>>>        secondary(2)
>>>>>>            This value is used for a group member that is used as =
a
>>>>>>            secondary target of an Exporter. The Exporter will use
>>>>>>            one of the targets specified as secondary(2) within =
the
>>>>>>            same Transport Session group when the primary target =
is
>>>>>>            not reachable.
>>>>>>=20
>>>>>>        parallel(3)
>>>>>>            This value is used for a group member that is used for
>>>>>>            duplicate exporting i.e., all group members =
identified>=20
>>>=20
>>>>>>            by the ipfixExportIndex are exporting the same Records
>>>>>>            in parallel. This implies that all group members MUST
>>>>>>            have the same membertype parallel(3).
>>>>>>=20
>>>>>>        loadBalancing(4)
>>>>>>            This value is used for a group member that is used
>>>>>>            as one target for load-balancing. This means that a>=20=

>>>=20
>>>>>>            Record is sent to one of the group members in this
>>>>>>            group identified by ipfixExportIndex.
>>>>>>            This implies that all group members MUST have the same
>>>>>>            membertype loadBalancing(4)."
>>>>>>    ::=3D { ipfixExportEntry 2 }
>>>>>>=20
>>>>>>=20
>>>>>> I'm open to better proposals.
>>>>>>=20
>>>>>> "parallel" and "loadBalancing" can be configured once with a =
single ExportingProcess class parameter.
>>>>>>=20
>>>>>> How would you configure "primary" and "secondary with a single =
Exporting Process parameter?
>>>> This issue is on my todo list.
>>>=20
>>> OK, it seems from a configuration point of view you have three =
modes. Parallel, loadBalancing and fallback.  In fallback mode you try =
the destinations in order until you get a connection.  The mode is a =
property of the Exporting Process and defines how to treat the list of =
destinations.
>>>=20
>>> i.e.
>>>=20
>>>   <Exporter>
>>>     <name>Test 1</name>
>>>     <mode>fallback</mode>
>>>     <destination> ... </destination>
>>>     <destination> ... </destination>
>>>     <destination> ... </destination>
>>>   </Exporter>
>>>=20
>>> So parallel mode sends to all three at once, load balancing takes it =
in turns and fallback treats the first as primary, the second and =
secondary and the third as tertiary.
>> Yes, this sounds good!
>=20
> Great.
>=20
> Thanks
>=20
> Andrew
> _______________________________________________
> IPFIX mailing list
> IPFIX@ietf.org
> https://www.ietf.org/mailman/listinfo/ipfix


From muenz@net.in.tum.de  Fri Mar 12 09:11:31 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 21D873A6992 for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 09:11:28 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.185
X-Spam-Level: 
X-Spam-Status: No, score=-2.185 tagged_above=-999 required=5 tests=[AWL=0.064,  BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 1PUkNNwO4Lft for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 09:11:24 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 6CC483A6820 for <ipfix@ietf.org>; Fri, 12 Mar 2010 09:09:30 -0800 (PST)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 62B21208CD3D; Fri, 12 Mar 2010 18:09:35 +0100 (CET)
Message-ID: <4B9A7563.3010200@net.in.tum.de>
Date: Fri, 12 Mar 2010 18:09:55 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Brian Trammell <trammell@tik.ee.ethz.ch>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B9257CE.6040409@cisco.com> <4B98E288.4040204@net.in.tum.de> <4B991389.2090209@cisco.com> <4B9A4B15.3010402@net.in.tum.de> <4B9A634C.4080503@cisco.com> <3BC9C9EF-4A1C-4A48-9AB6-E7CE5EF0C0BC@tik.ee.ethz.ch>
In-Reply-To: <3BC9C9EF-4A1C-4A48-9AB6-E7CE5EF0C0BC@tik.ee.ethz.ch>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms010202090504090502000005"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 12 Mar 2010 17:11:34 -0000

This is a cryptographically signed message in MIME format.

--------------ms010202090504090502000005
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Brian,

If it does not matter whether the Selected Packet Stream at the input of =

the two Caches is identical, you could use two distinct Selection=20
Processes (with identical configuration). If so, you can see it as two=20
different Metering Process, each having a distinct SP and distinct Cache.=


If the Selected Packet Stream at the input of the two Caches needs to be =

identical (e.g., because you want to generate Packet Reports and Flow=20
Records from the identical subset of packets), then the same SP is=20
shared by two Caches.

There is no IPFIX Standard Track document which disallows such a=20
configuration. Therefore, the configuration data model allows both=20
configurations.

I think that such a configuration can be useful. And I consider it as a=20
single MP with two Caches.

Andrew feels uncomfortable with such a configuration because it=20
contradicts "the paradigm in which IPFIX and PSAMP have been developed" ;=
)

Regards,
Gerhard


Brian Trammell wrote:
> hi Andrew, Gerhard,
>=20
> Let me step in here quickly, as I can easily see a place for a multicac=
he MP.. the following is something I was asked to do once upon a time:
>=20
> Suppose we have an MP at an observation point in a security monitoring =
context. This MP is driving two separate processes: one which filters tra=
ffic according to some current set of criteria defining "bad" traffic, th=
en displays in "real time" (~5s delay) this "bad" traffic volume (the fas=
t stream), and one which prefilters and stores full flow data for retrosp=
ective analysis (the full stream). The fast stream gets used to drive a d=
isplay and maybe turn on a blinkenlight somewhere, then it gets thrown aw=
ay. The full stream has richer data about each flow and gets stored and u=
sed by the analysis tools for doing actual work.
>=20
> It seemed to me at the time that the right way to handle this situation=
 is two caches in the same MP: one with very fast active timeouts (we don=
't care about termination conditions, we just want to see flow-level volu=
mes), and one with reasonable timeouts (because the analyses are simpler =
when you don't have to stitch together all the flows from active timeouts=
=2E) Everything else -> observation point, selection process, even the EP=
 and CP, are the same.
>=20
> Or is this two MPs in your reading of the architecture?
>=20
> Cheers,
>=20
> Brian
>=20
> On Mar 12, 2010, at 4:52 PM, Andrew Johnson wrote:
>=20
>> Hi Gerhard
>>
>> Gerhard Muenz wrote:
>>> Hi Andrew, all,
>>> Andrew Johnson wrote:
>>>> Gerhard Muenz wrote:
>>>>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=20
>>>>>>>> Section 3:
>>>>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=20
>>>>>>>> If a Selection Process is part of a Metering Process and a singl=
e
>>>>>>>> Selection Process can feed into more than one Cache, then it is
>>>>>>>> implied that a single Metering Process can have more than once c=
ache.
>>>>>>> Note: The text has slightly changed after Atsushi's review. Now, =
it explicitly says that a Metering Process may contain multiple Selection=
 Processes and multiple Caches.
>>>>>> If a Metering Process can have multiple caches and an IPFIX device=
 can have multiple Metering Processes, how do we associate the caches wit=
h Metering Processes?  Can this configuration system only be used to conf=
igure a single Metering Process within a device?
>>>>> The model does not group Selection Processes and Caches into Meteri=
ng Processes. This is done on purpose because there is not a single corre=
ct view.
>>>>>
>>>>> All Selection Process and all Caches of a Monitoring Device can be =
regarded as one Metering Process.
>>>>>
>>>>> Alternatively, Selection Processes and Caches which are not directl=
y or indirectly linked to each other can be seen as belonging to differen=
t Metering Processes.
>>>> OK, at this point I think I'm uncomfortable with the idea that a Met=
ering Process can have more than one cache.  What is the reasoning behind=
 this decision?
>>> It should be possible that the output of the same Selection Process e=
nters two different Caches.
>>> Example:
>>>          +-> Cache 1 (PSAMP)
>>> OP -> SP -+
>>>          +-> Cache 2 (IPFIX)
>>> It is difficult to say that this corresponds to two MPs.
>> Is the only argument for these two caches being in the same MP is so t=
hey can share a selection process?
>>
>> I say that the cache is the heart of a Metering Process and it should =
only have one!  I realise that it may not be explicitly stated in current=
 IPFIX documents, but I think it's implied in the Architecture document a=
nd is the paradigm in which IPFIX and PSAMP have been developed.
>>
>> It seems that what you really object to is the PSAMP declaration that =
a selection process belongs to a single Metering Process, but as we've se=
en, the statistics held by a Metering Process need to be kept on a per-ca=
che basis.
>>
>> I don't object to the configuration shorthand of using the same config=
 across different Metering Processes, but with this model the stats are g=
oing to be quite different to those needed for PSAMP.  That's fine though=
!
>>
>>
>>>>>>> Furthermore, it says that only packets observed at a single Obser=
vation Point may enter a Selection Process. IMO, this results from the PS=
AMP definitions of Observed Packet Stream and Selection Process. Please t=
ell me if this conclusion is wrong.
>>>>>> A Selection process should only have a single feed, but can have m=
ultiple observers.  This results in the number of possible chains (Select=
ion Sequences) being equal to the number of Selectors that directly feed =
a cache.
>>>>> Ok. Single feed means single Observation Point as specified in the =
draft.
>>>> No, by "feed" I mean input stream.  Either from an Observation point=
 or the output of a different Selection Process.
>>> Yes, this is what I wanted to say. Therefore, we have "0..1" in -05 (=
it was "0..*" in -04):
>>>                        +------------------+
>>>                        | ObservationPoint |
>>>                        +------------------+
>>>                             0..1 |
>>>                             =3D=3D=3D=3D |
>>>                                  |
>>>                             0..* V
>>>                        +------------------+
>>>                        | SelectionProcess |
>>>                        +------------------+<-+
>>>                             0..* |  0..1 |   | 0..*
>>>                                  |  =3D=3D=3D=3D +---+
>>>                             0..* V
>>>                        +------------------+
>>>                        | Cache            |
>>>                        +------------------+
>>>                             0..* |
>>>                                  |
>>>                             0..* V
>>>                        +------------------+
>>>                        | ExportingProcess |
>>>                        +------------------+
>>> Sorry, if this was not clear from the beginning.
>> This has always been clear... I'm not sure what the original points we=
re now.
>>
>>
>>>>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=20
>>>>>>>> From section 4.1:
>>>>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=20
>>>>>>>> Doesn't the direction belong in the Observation Point class, not=
 the
>>>>>>>> Interface or LC class.
>>>>>>> If the current position of the direction parameter bothers you, w=
e can move it.
>>>>>>>
>>>>>>> However, my reasoning is the following:
>>>>>>> - If it appears in the ObservationPoint class, it looks like the =
Observation Point is doing some filtering.
>>>>>>> - If it appears in the Interface/Linecard class, it looks like th=
e filtering is already done at the interface or linecard.
>>>>>>>
>>>>>>> I thought that the interface/linecard is doing the filtering.
>>>>>> I don't see it as filtering, just part of the definition of the Ob=
servation Point.  The Interface is most likely able to send and receive t=
raffic, but the observation point could be on only the send or receive ch=
annel.  i.e. It's not a input only interface, but the Observation Point c=
ould only see input packets.
>>>>>>
>>>>>> Currently you have a 1:1 mapping of OP to interface.  If the direc=
tion is moved to the Observation Point then you could have two observatio=
n points on the same interface, one on the input side and one on the outp=
ut.
>>>>> With the current model, there is no problem to configure two OPs on=
 the same interface:
>>>>>
>>>>>     <observationPoint>
>>>>>       <name>interface ingress</name>
>>>>>       <interface>
>>>>>         <ifIndex>3</ifIndex>
>>>>>         <direction>ingress</direction>
>>>>>       </interface>
>>>>>       ...
>>>>>     </observationPoint>
>>>>>
>>>>>     <observationPoint>
>>>>>       <name>interface egress</name>
>>>>>       <interface>
>>>>>         <ifIndex>3</ifIndex>
>>>>>         <direction>egress</direction>
>>>>>       </interface>
>>>>>       ...
>>>>>     </observationPoint>
>>>> OK, but my point was always that the interface is not defined by its=
 direction.
>>>>
>>>> These two "interfaces" are actually two observation points on the sa=
me interface:
>>>>        <interface>
>>>>          <ifIndex>3</ifIndex>
>>>>          <direction>ingress</direction>
>>>>        </interface>
>>>>
>>>>        <interface>
>>>>          <ifIndex>3</ifIndex>
>>>>          <direction>ingress</direction>
>>>>        </interface>
>>> Ok. So it seems to you that <direction> belongs to the definition of =
the interface.
>>> If we moved <direction> to the Observation Point, we could remove the=
 Interface/Linecard containers because there would be only one parameter =
left:
>>>    +-------------------------------------------------+
>>>    | ObservationPoint                                |
>>>    +-------------------------------------------------+
>>>    | name                                            |
>>>    | observationPointId {readOnly}                   |
>>>    | observationDomainId                             |
>>>    | ifIndex/ifName/entPhysicalIndex/entPhysicalName | 0..*
>>>    | direction                                       |----+
>>>    +-------------------------------------------------+    | 0..*
>>>                                                           V
>>>                                          +--------------------+
>>>                                          | SelectionProcess   |
>>>                                          +--------------------+
>>> Would you agree with this change?
>> Yes, this seems fine.
>>
>>
>>>>>>> There is another small advantage: The current position of the par=
ameter makes it easier to add vendor or implementation-specific packet so=
urces for which the direction parameter has no meaning, for example:
>>>>>>>
>>>>>>> +------------------+
>>>>>>> | PcapFile         |
>>>>>>> +------------------+
>>>>>>> | filename         |
>>>>>>> +------------------+
>>>>>>>
>>>>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=20
>>>>>>>> From Section 4.3.1
>>>>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=20
>>>>>>>> Isn't is possible to generate a cache without any keys?
>>>>>>>> This would generate a cache with a single flow?  Do we allow thi=
s?
>>>>>>> A the moment, there is no restriction which says that a flow key =
field must exist.
>>>>>>>
>>>>>>> Do you want to add such a restriction?
>>>>>> A cache without keys will result in a single flow, which seems a b=
it daft.  I guess there is no pressing reason to add this restriction tho=
ugh.
>>>>>>
>>>>>>>> Would it be useful to have two lists, one for keys and one for
>>>>>>>> non-keys?
>>>>>>> There is an argument for having a single list:
>>>>>>>
>>>>>>> A Cache implementation may generate records using the specified f=
ield order. So, the fields occur in the same order as specified in the co=
nfiguration.
>>>>>> If we're worried about implementation specifics can't they extend =
the YANG model?
>>>>> Yes, they can.
>>>>>
>>>>>>> With separate lists for flow key fields and non-key fields, we ca=
nnot arbitrarily mix these two types of fields.
>>>>>>>
>>>>>>> Is there any advantage of having separate lists?
>>>>>> I was just thinking about how you'd specify that at least one key =
was required.
>>>>> Caches of type "immediate" do not have Flow Keys. So, the restricti=
ons would not apply to such Caches.
>>>> Really?  Why not say that Caches of an "immediate" type have no non-=
keys?  For example, the counter fields should not be key fields, they mak=
e no sense as keys.  They also make no sense with the immediate cache.  I=
t seems that an immediate cache is made up of entirely key fields.
>>> IMO, a Packet Report does not contain any Flow information, so it doe=
s not make sense to talk about Flow Keys.
>>> In -05, I restricted the <isFlowKey> parameter to Caches generating F=
low Records because I think that it does not make sense to distinguish Fl=
ow Key fields and non-key fields if Packet Reports are generated.
>> Ah I see.  I guess it makes sense to simply not say anything about whe=
ther it is a key or not.  So these fields are not keys, but neither are t=
hey non-keys.  The are simply collected fields.  Certain fields only make=
 sense for flows and not for packet reports, which might be worth mention=
ing.
>>
>>
>>>>>>>> For a permanent cache, update counters should be used and for a
>>>>>>>> timeout cache the delta counters should be used.  Where is this
>>>>>>>> sort of thing enforced?
>>>>>>> This is not enforced. I would not strictly enforce it because str=
ange configurations might make sense in specific use cases.
>>>>>> OK.
>>>>>>
>>>>>>> However, we could add a note that totalCounters are typically use=
d for "permanent" and deltaCounters for "timeout".
>>>>>>>
>>>>>>> Could you provide a sentence or two?
>>>>>> Perhaps a more general note that not all fields will be appropriat=
e for all configurations. Is the Exporting Device free to change the fiel=
ds
>>>>>> if it needs total counters for a permanent cache?
>>>>> No.
>>>>>
>>>>> I added a new paragraph at the end of 4.3.1. (CacheLayout Class):
>>>>>
>>>>>   Note that the use of Information Elements can be restricted to
>>>>>   certain Cache Modes or to Flow Key or non-key fields.  Such
>>>>>   restrictions may result from Information Element definitions or f=
rom
>>>>>   device-specific constraints.  According to Section 5, the Monitor=
ing
>>>>>   Device MUST notify the user if a Cache field cannot be configured=

>>>>>   with the given Information Element.
>>>>>
>>>>> Ok?
>>>> Yes, seems reasonable.
>>>>
>>>>
>>>>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=20
>>>>>>>> From Section 4.4.1
>>>>>>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=20
>>>>>>>>   exportMemberType:  Configures the export member type that corr=
esponds
>>>>>>>>      to the ipfixTransportSessionGroupMemberType object in
>>>>>>>>      [I-D.ietf-ipfix-mib].  The following parameter values are
>>>>>>>>      specified by the configuration data model:
>>>>>>>>      *  primary: primary target of the Exporting Process
>>>>>>>>      *  secondary: secondary target of the Exporting Process use=
d when
>>>>>>>>         the primary target is not reachable
>>>>>>>>      *  parallel: parallel exporting to all destinations and fil=
es of
>>>>>>>>         the Exporting Process
>>>>>>>>      *  loadBalancing: load-balancing between different destinat=
ions
>>>>>>>>         and files of the Exporting Process
>>>>>>>>      "parallel" is the default value if this parameter is not
>>>>>>>>      configured.  If one destination or file is configured as "p=
rimary"
>>>>>>>>
>>>>>>>> ----------------------------------------------------------------=
--------=20
>>>>>>>> Comments
>>>>>>>> ----------------------------------------------------------------=
--------=20
>>>>>>>> It's not clear to me that this parameter belongs to the Destinat=
ion
>>>>>>>> class.  The number of restrictions on when you can use the vario=
us
>>>>>>>> types suggests that we could simplify things.
>>>>>>>>
>>>>>>>> A Cache can pass flow records to multiple Exporting Processes, s=
o we
>>>>>>>> don't really need a parallel option.  This would mean that each
>>>>>>>> Exporting Process is responsible for sending one complete set of=

>>>>>>>> records.  i.e. one record in, one record out.  It could do that,=
 either
>>>>>>>> using a redundancy model or with a load-balancing one.
>>>>>>>>
>>>>>>>> In other words, the exportType would be an Exporting Process att=
ribute
>>>>>>>> which would indicate whether the list of Destinations should be =
load-
>>>>>>>> balanced across or treated as Primary, Seondary, Tertiary, etc.
>>>>>>> At the moment, the parameter is inline with the IPFIX-MIB:
>>>>>>>
>>>>>>> ipfixExportMemberType OBJECT-TYPE
>>>>>>>    SYNTAX      INTEGER {
>>>>>>>                    unknown(0),
>>>>>>>                    primary(1),
>>>>>>>                    secondary(2),
>>>>>>>                    parallel(3),
>>>>>>>                    loadBalancing(4)
>>>>>>>                }
>>>>>>>    MAX-ACCESS  read-only
>>>>>>>    STATUS      current
>>>>>>>    DESCRIPTION
>>>>>>>        "The type of a member Transport Session in a Transport
>>>>>>>        Session group (identified by the value of ipfixExportIndex=
,
>>>>>>>        ipfixObservationDomainId and ipfixMeteringProcessCacheId).=

>>>>>>>        The following values are valid:
>>>>>>>
>>>>>>>        unknown(0)
>>>>>>>            This value MUST be used if the status of the group
>>>>>>>            membership cannot be detected by the equipment. This
>>>>>>>            value should be avoided as far as possible.
>>>>>>>
>>>>>>>        primary(1)
>>>>>>>            This value is used for a group member that is used as
>>>>>>>            the primary target of an Exporter. Other group members=

>>>>>>>            (with the same ipfixExportIndex and
>>>>>>>            ipfixMeteringProcessCacheId) MUST NOT have the value
>>>>>>>            primary(1) but MUST have the value secondary(2).
>>>>>>>            This value MUST also be specified if the Exporter does=

>>>>>>>            not support Transport Session grouping. In this case
>>>>>>>            the group contains only one Transport Session.
>>>>>>>
>>>>>>>        secondary(2)
>>>>>>>            This value is used for a group member that is used as =
a
>>>>>>>            secondary target of an Exporter. The Exporter will use=

>>>>>>>            one of the targets specified as secondary(2) within th=
e
>>>>>>>            same Transport Session group when the primary target i=
s
>>>>>>>            not reachable.
>>>>>>>
>>>>>>>        parallel(3)
>>>>>>>            This value is used for a group member that is used for=

>>>>>>>            duplicate exporting i.e., all group members identified=
>=20
>>>>>>>            by the ipfixExportIndex are exporting the same Records=

>>>>>>>            in parallel. This implies that all group members MUST
>>>>>>>            have the same membertype parallel(3).
>>>>>>>
>>>>>>>        loadBalancing(4)
>>>>>>>            This value is used for a group member that is used
>>>>>>>            as one target for load-balancing. This means that a>=20
>>>>>>>            Record is sent to one of the group members in this
>>>>>>>            group identified by ipfixExportIndex.
>>>>>>>            This implies that all group members MUST have the same=

>>>>>>>            membertype loadBalancing(4)."
>>>>>>>    ::=3D { ipfixExportEntry 2 }
>>>>>>>
>>>>>>>
>>>>>>> I'm open to better proposals.
>>>>>>>
>>>>>>> "parallel" and "loadBalancing" can be configured once with a sing=
le ExportingProcess class parameter.
>>>>>>>
>>>>>>> How would you configure "primary" and "secondary with a single Ex=
porting Process parameter?
>>>>> This issue is on my todo list.
>>>> OK, it seems from a configuration point of view you have three modes=
=2E Parallel, loadBalancing and fallback.  In fallback mode you try the d=
estinations in order until you get a connection.  The mode is a property =
of the Exporting Process and defines how to treat the list of destination=
s.
>>>>
>>>> i.e.
>>>>
>>>>   <Exporter>
>>>>     <name>Test 1</name>
>>>>     <mode>fallback</mode>
>>>>     <destination> ... </destination>
>>>>     <destination> ... </destination>
>>>>     <destination> ... </destination>
>>>>   </Exporter>
>>>>
>>>> So parallel mode sends to all three at once, load balancing takes it=
 in turns and fallback treats the first as primary, the second and second=
ary and the third as tertiary.
>>> Yes, this sounds good!
>> Great.
>>
>> Thanks
>>
>> Andrew
>> _______________________________________________
>> IPFIX mailing list
>> IPFIX@ietf.org
>> https://www.ietf.org/mailman/listinfo/ipfix
>=20

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms010202090504090502000005
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms010202090504090502000005--

From paitken@cisco.com  Fri Mar 12 14:40:53 2010
Return-Path: <paitken@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 8A75C3A682F for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 14:40:53 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -7.524
X-Spam-Level: 
X-Spam-Status: No, score=-7.524 tagged_above=-999 required=5 tests=[AWL=3.075,  BAYES_00=-2.599, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id hGTV9k9Sj9xT for <ipfix@core3.amsl.com>; Fri, 12 Mar 2010 14:40:52 -0800 (PST)
Received: from ams-iport-1.cisco.com (ams-iport-1.cisco.com [144.254.224.140]) by core3.amsl.com (Postfix) with ESMTP id 619853A681D for <ipfix@ietf.org>; Fri, 12 Mar 2010 14:40:52 -0800 (PST)
Authentication-Results: ams-iport-1.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AtQAAMpRmkuQ/uCWe2dsb2JhbACaahUBAQsLJAYciB+bB5hahHsE
X-IronPort-AV: E=Sophos;i="4.49,628,1262563200"; d="scan'208";a="58003509"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 12 Mar 2010 22:40:57 +0000
Received: from [10.61.95.53] (ams3-vpn-dhcp7990.cisco.com [10.61.95.53]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2CMevb9015581; Fri, 12 Mar 2010 22:40:57 GMT
Message-ID: <4B9AC2F8.1090504@cisco.com>
Date: Fri, 12 Mar 2010 22:40:56 +0000
From: Paul Aitken <paitken@cisco.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-GB; rv:1.8.1.22) Gecko/20090605 SeaMonkey/1.1.17 (Ubuntu-1.1.17+nobinonly-0ubuntu0.9.04.1)
MIME-Version: 1.0
To: "ipfix@ietf.org" <ipfix@ietf.org>, "Romascanu, Dan (Dan)" <dromasca@avaya.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Subject: [IPFIX] IPFIX errata 2052 to 2059
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 12 Mar 2010 22:40:53 -0000

Dear all,

Discussion of these errata ended ~ 2 weeks ago. Presumably nobody has 
any objections?

http://www.rfc-editor.org/errata_search.php?rfc=5477
http://www.rfc-editor.org/errata_search.php?rfc=5476

http://www.ietf.org/mail-archive/web/ipfix/current/msg05314.html
http://www.ietf.org/mail-archive/web/ipfix/current/msg05315.html


Dan, what's the next step? Ideally an update to the IANA registry.


Thanks,
P.

-- 
Paul Aitken
Cisco Systems Ltd, Edinburgh, Scotland.
http://www.cisco.com/web/about/doing_business/legal/cri/index.html

From muenz@net.in.tum.de  Sat Mar 13 04:40:05 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id B1B4A3A6840 for <ipfix@core3.amsl.com>; Sat, 13 Mar 2010 04:40:05 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.249
X-Spam-Level: 
X-Spam-Status: No, score=-2.249 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id HES-OXcbu-3U for <ipfix@core3.amsl.com>; Sat, 13 Mar 2010 04:40:04 -0800 (PST)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 6383A3A67B1 for <ipfix@ietf.org>; Sat, 13 Mar 2010 04:40:04 -0800 (PST)
Received: from [192.168.2.152] (ppp-93-104-86-164.dynamic.mnet-online.de [93.104.86.164]) by mail.net.in.tum.de (Postfix) with ESMTPSA id CEFBD208B0F0; Sat, 13 Mar 2010 13:40:08 +0100 (CET)
Message-ID: <4B9B87BD.7020201@net.in.tum.de>
Date: Sat, 13 Mar 2010 13:40:29 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Paul Aitken <paitken@cisco.com>
References: <4B9AC2F8.1090504@cisco.com>
In-Reply-To: <4B9AC2F8.1090504@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms090505000902010102060306"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: "ipfix@ietf.org" <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX errata 2052 to 2059
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sat, 13 Mar 2010 12:40:05 -0000

This is a cryptographically signed message in MIME format.

--------------ms090505000902010102060306
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Paul,

Still, I have not heard any good arguments why we need to change it to=20
unsigned64. Looking at the examples in RFC5476, it seems that the=20
original intention was unsigned32. This is inline with=20
ipfixSelectionProcessSelectorIndex in IPFIX-MIB.

So, why do you prefer 64bit, which does not only require changes in=20
RFC5477, but in RFC5476 as well, and which is inconsistent with IPFIX-MIB=
?

See also my previous mail:
http://www.ietf.org/mail-archive/web/ipfix/current/msg05311.html

Regards,
Gerhard

Paul Aitken wrote:
> Dear all,
>=20
> Discussion of these errata ended ~ 2 weeks ago. Presumably nobody has=20
> any objections?
>=20
> http://www.rfc-editor.org/errata_search.php?rfc=3D5477
> http://www.rfc-editor.org/errata_search.php?rfc=3D5476
>=20
> http://www.ietf.org/mail-archive/web/ipfix/current/msg05314.html
> http://www.ietf.org/mail-archive/web/ipfix/current/msg05315.html
>=20
>=20
> Dan, what's the next step? Ideally an update to the IANA registry.
>=20
>=20
> Thanks,
> P.
>=20

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms090505000902010102060306
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms090505000902010102060306--

From muenz@net.in.tum.de  Mon Mar 15 03:06:22 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id DC01F3A68B9 for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 03:06:22 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.889
X-Spam-Level: 
X-Spam-Status: No, score=-0.889 tagged_above=-999 required=5 tests=[AWL=-1.240, BAYES_50=0.001, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id wXsCBp-gws3y for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 03:06:21 -0700 (PDT)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id B9C433A6863 for <ipfix@ietf.org>; Mon, 15 Mar 2010 03:06:16 -0700 (PDT)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 244EF208CD3D; Mon, 15 Mar 2010 11:06:22 +0100 (CET)
Message-ID: <4B9E06B4.6090203@net.in.tum.de>
Date: Mon, 15 Mar 2010 11:06:44 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Andrew Johnson <andrjohn@cisco.com>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com> <4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com> <4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com>
In-Reply-To: <4B9918A6.6010407@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms040606020501000200050208"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 15 Mar 2010 10:06:23 -0000

This is a cryptographically signed message in MIME format.

--------------ms040606020501000200050208
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Andrew,

Andrew Johnson wrote:
>> Nevertheless, I reworked the corresponding parts of the text with the =

>> following objectives:
>> - no reference to parameter descriptions in the MIB (as proposed by
>>   Brian)
>> - make adaptations to the state parameters adopted from the MIB where =
it
>>   is useful
>> - recommend re-initialization of counters at reconfiguration
>>
>> As an example, there exists YANG counter types which are set to zero a=
t=20
>> initialization (in contrast to MIB counters which start with an=20
>> arbitrary value). I propose to use these counter types.
>>
>> So, please comment the current text:
>>
>>
>> 4.2.1.  Selector Class
>>
>>     +---------------------------------------+
>>     | Selector                              |
>>     +---------------------------------------+      1 +----------------=
-+
>>     | name                                  |<>------+ SelectAll/     =
 |
>>     | selectorId {readOnly}                 |        | SampCountBased/=
 |
>>     | packetsObserved {readOnly}            |        | SampTimeBased/ =
 |
>>     | packetsDropped {readOnly}             |        | SampRandOutOfN/=
 |
>>     | selectorInitializationTime {readOnly} |        | SampUniProb/   =
 |
>>     |                                       |        | FilterMatch/   =
 |
>>     |                                       |        | FilterHash/    =
 |
>>     +---------------------------------------+        +----------------=
-+
>>
>>                          Figure 8: Selector class
>>
>> [snip]
>>
>>    As state parameters, the Selector class also contains the Selector
>>    statistics packetsObserved and packetsDropped as well as
>>    selectorInitializationTime:
>>
>>    packetsObserved:  The number of packets observed at the input of th=
e
>>       Selector since the point in time indicated by
>>       selectorDiscontinuityTime.  In contrast to
>>       ipfixSelectionProcessStatsPacketsObserved in the IPFIX MIB modul=
e
>>       [I-D.ietf-ipfix-mib], the counter is initialized with zero.
>>
>>    packetsDropped:  The number of packets discarded by the Selector
>>       since the point in time indicated by selectorDiscontinuityTime.
>>       In contrast to ipfixSelectionProcessStatsPacketsDropped in the
>>       IPFIX MIB module [I-D.ietf-ipfix-mib], the counter is initialize=
d
>>       with zero.
>>
>>    selectorInitializationTime:  The value of sysUpTime at the most
>>       recent occasion at which the Selector was re-initialized and at
>>       which the counters packetsObserved and packektsDropped were rese=
t
>>       to zero.  In contrast to
>>       ipfixSelectionProcessStatsDiscontinuityTime in the IPFIX MIB
>>       module [I-D.ietf-ipfix-mib], the zero value does not have any
>>       specific meaning.
>>
>>    The counter values of packetsObserved and packetsDropped MUST both =
be
>>    set to zero at the time indicated by selectorInitializationTime.  I=
t
>>    is RECOMMENDED to reset the counters after every reconfiguration of=

>>    the Selector.
>=20
> OK.  Clearing the stats on reconfiguration seems like a good idea.=20
> Perhaps we should make it clear that the statistics captured on a per=20
> Selector class basis are not the same as those that get exported.  The =

> exported ones are captured on a Selection Sequence basis and can be=20
> reset whenever the implementation feels the need.

To make it clearer, I will add an explicit note that the counters in the =

Selector class correspond to the counters in the IPFIX MIB.

Just like in the MIB specification, we do not specify how the counters=20
are related to the PSAMP statistics reports.

>> [snip]
>>
>> 4.3.  Cache Class
>>
>>      +-------------------------------------+
>>      | Cache                               |
>>      +-------------------------------------+         1 +-------------+=

>>      | name                                |<>---------| CacheLayout |=

>>      | cacheMode                           |           +-------------+=

>>      | maxRecords {opt.}                   |
>>      | activeTimeout {opt.} {Cache Modes   | 0..*
>>      |   "timeout" and "natural" only}     |---------------+
>>      | inactiveTimeout {opt.} {Cache Modes |               | 0..*
>>      |   "timeout" and "natural" only}     |               V
>>      | exportInterval {opt.} {Cache Mode   |      +------------------+=

>>      |   "permanent" only}                 |      | ExportingProcess |=

>>      | activeFlows {readOnly}              |      +------------------+=

>>      | inactiveFlows {readOnly}            |
>>      | cacheDataRecords {readOnly}         |
>>      | cacheInitializationTime {readOnly}  |
>>      +-------------------------------------+
>>
>>                           Figure 11: Cache class
>>
>> [snip]
>>
>>    As state parameters, the Cache class contains the Metering Process
>>    statistics activeFlows, inactiveFlows, and cacheDataRecords, as wel=
l
>>    as cacheInitializationTime:
>>
>>    activeFlows:  The number of Flows currently active in this cache.
>>       This value corresponds to ipfixMeteringProcessCacheActiveFlows i=
n
>>       the IPFIX MIB module [I-D.ietf-ipfix-mib].
>>
>>    inactiveFlows:  The number of Flows currently inactive in this cach=
e.
>>       This value corresponds to ipfixMeteringProcessCacheInactiveFlows=

>>       in the IPFIX MIB module [I-D.ietf-ipfix-mib].
>>
>>    cacheDataRecords:  The number of Data Records generated by this Cac=
he
>>       since the point in time indicated by cacheIntitializationTime.
>>       In contrast to ipfixMeteringProcessDataRecords in the IPFIX MIB
>>       module [I-D.ietf-ipfix-mib], the counter is initialized with zer=
o.
>>
>>    cacheInitializationTime:  The value of sysUpTime at the most recent=

>>       occasion at which the Cache was re-initialized and at which the
>>       counter cacheDataRecords was reset to zero.  In contrast to
>>       ipfixMeteringProcessDiscontinuityTime in the IPFIX MIB module
>>       [I-D.ietf-ipfix-mib], the zero value does not have any specific
>>       meaning.
>>
>>    The counter value of cacheDataRecords MUST be set to zero at the ti=
me
>>    indicated by cacheInitializationTime.  It is RECOMMENDED to reset t=
he
>>    counter after every reconfiguration of the Cache.
>=20
> Again resetting counters on reconfiguration seems reasonable to me,=20
> however does changing a timeout value really necessitate resetting the =

> counters?

"RECOMMEND" allows you to keep the counters. I do not think that it=20
makes sense to discuss any possible situation where not resetting the=20
counters might be appropriate.

>> Question:
>> Is it clear what inactive Flows are?
>> I can imagine what this number is supposed to by, but I'm not sure.
>=20
> No it isn't.  So if I have a cache with 1000 entries, some might be=20
> active, some might be free, ready to be populated with a new flow, and =
I=20
> suppose some might be considered ready for export, sitting in Limbo=20
> until the Exporting Process has the resources to send them.  Are these =

> the inactive flows?

This counter was adopted from the MIB. I'll send another e-mail to=20
discuss this together with other issues.

Regards,
Gerhard

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms040606020501000200050208
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms040606020501000200050208--

From muenz@net.in.tum.de  Mon Mar 15 03:54:40 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id A47B33A6B7E for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 03:54:40 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.13
X-Spam-Level: 
X-Spam-Status: No, score=-2.13 tagged_above=-999 required=5 tests=[AWL=0.119,  BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id aDCs1eRGaxxu for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 03:54:35 -0700 (PDT)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id C75AD3A6B7D for <ipfix@ietf.org>; Mon, 15 Mar 2010 03:50:06 -0700 (PDT)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 2D3F7208CD3D for <ipfix@ietf.org>; Mon, 15 Mar 2010 11:50:00 +0100 (CET)
Message-ID: <4B9E10EE.9050406@net.in.tum.de>
Date: Mon, 15 Mar 2010 11:50:22 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: IETF IPFIX Working Group <ipfix@ietf.org>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com> <4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com> <4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com>
In-Reply-To: <4B9918A6.6010407@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms030802080401080104060600"
X-Virus-Scanned: ClamAV using ClamSMTP
Subject: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIX MIB
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 15 Mar 2010 10:54:40 -0000

This is a cryptographically signed message in MIME format.

--------------ms030802080401080104060600
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Dear all,

I would appreciate some feedback regarding counter values which have=20
been adopted from IPFIX MIB.

In the IPFIX MIB, packet or record counters start at certain point in=20
time which is specified as "value of sysUpTime".
(sysUpTime is a timer in MIB-2 for "the time (in hundredths of a second) =

since the network management portion of the system was last=20
re-initialized.")

What we usually want to know is one of the following:
- absolute time when the packet/record counters were initialized, or
- duration of the current counting period.
To obtain these values, it is necessary to retrieve sysUpTime first from =

MIB-2, and then do some calculation.

In the current configuration data model, we have adopted the counters=20
and copied the definitions from the IPFIX MIB.

Since IPFIX-CONFIG is not located in the SNMP world, I propose the=20
following changes:

- Counters start with zero (instead of arbitrary values)
   =3D> yang type zero-based-counter64
- Time at which counters start is specified as an absolute time stamp
   (instead of sysUpTime value)
   =3D> yang type time-and-date
   (Alternatively, we could measure the duration of the counting period
   in timeticks or milliseconds.)

Are there any opinions/objections?


Then, I do not understand what the following value in IPFIX MIB is=20
supposed to be:

    ipfixMeteringProcessCacheInactiveFlows OBJECT-TYPE
        SYNTAX      Gauge32
        MAX-ACCESS  read-only
        STATUS      current
        DESCRIPTION
            "The number of Flows currently inactive at this cache."
        ::=3D { ipfixMeteringProcessStatsEntry 2 }


Well, I'm a coauthor of IPFIX MIB but I have to admit that it is not=20
clear to me what inactive flows are supposed to be.

Can anybody give an explanation?


Regards,
Gerhard

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms030802080401080104060600
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms030802080401080104060600--

From paitken@cisco.com  Mon Mar 15 05:24:22 2010
Return-Path: <paitken@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id DEBBB3A6D46 for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 05:24:22 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -4.139
X-Spam-Level: 
X-Spam-Status: No, score=-4.139 tagged_above=-999 required=5 tests=[AWL=-1.540, BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id sCGhiVQH3nOM for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 05:24:21 -0700 (PDT)
Received: from ams-iport-2.cisco.com (ams-iport-2.cisco.com [144.254.224.141]) by core3.amsl.com (Postfix) with ESMTP id 728BB3A6D48 for <ipfix@ietf.org>; Mon, 15 Mar 2010 04:50:18 -0700 (PDT)
Authentication-Results: ams-iport-2.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AhwBAAK8nUuQ/uCWe2dsb2JhbACabxUBAQsLJAYciCGWG5dRhHsE
X-IronPort-AV: E=Sophos;i="4.49,642,1262563200";  d="scan'208";a="4342694"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-2.cisco.com with ESMTP; 15 Mar 2010 11:16:50 +0000
Received: from [144.254.153.37] (dhcp-144-254-153-37.cisco.com [144.254.153.37]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2FBoOtt028689; Mon, 15 Mar 2010 11:50:24 GMT
Message-ID: <4B9E1F00.3080307@cisco.com>
Date: Mon, 15 Mar 2010 11:50:24 +0000
From: Paul Aitken <paitken@cisco.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-GB; rv:1.8.1.22) Gecko/20090605 SeaMonkey/1.1.17 (Ubuntu-1.1.17+nobinonly-0ubuntu0.9.04.1)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <4B9AC2F8.1090504@cisco.com> <4B9B87BD.7020201@net.in.tum.de>
In-Reply-To: <4B9B87BD.7020201@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: "ipfix@ietf.org" <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX errata 2052 to 2059
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 15 Mar 2010 12:24:23 -0000

Gerhard,

We agree that unsigned16 selectorID is too small.

Personally, I don't think field sizes should be specified in the info 
model at all, since sizes are specified in the templates. However, 
specifying sizes seems beneficial for collectors and necessary for the MIB.

So our options are to increase the selectorID size to unsigned32 or 
unsigned64.

u32 would help in the short to mid term. However, we can anticipate that 
even this will be too small in future when an observation domain may 
consist of many selectors or when selector Ids are encoded in blocks or 
groups.

If we settled for u32, we could anticipate a future request for u64.

Changing to u64 now seems to provide the greatest long-term flexibility.

Cheers,
P.

> Still, I have not heard any good arguments why we need to change it to 
> unsigned64. Looking at the examples in RFC5476, it seems that the 
> original intention was unsigned32. This is inline with 
> ipfixSelectionProcessSelectorIndex in IPFIX-MIB.
> 
> So, why do you prefer 64bit, which does not only require changes in 
> RFC5477, but in RFC5476 as well, and which is inconsistent with IPFIX-MIB?
> 
> See also my previous mail:
> http://www.ietf.org/mail-archive/web/ipfix/current/msg05311.html
> 
> Regards,
> Gerhard
> 
> Paul Aitken wrote:
>> Dear all,
>>
>> Discussion of these errata ended ~ 2 weeks ago. Presumably nobody has 
>> any objections?
>>
>> http://www.rfc-editor.org/errata_search.php?rfc=5477
>> http://www.rfc-editor.org/errata_search.php?rfc=5476
>>
>> http://www.ietf.org/mail-archive/web/ipfix/current/msg05314.html
>> http://www.ietf.org/mail-archive/web/ipfix/current/msg05315.html
>>
>>
>> Dan, what's the next step? Ideally an update to the IANA registry.
>>
>>
>> Thanks,
>> P.
>>
> 


-- 
Paul Aitken
Cisco Systems Ltd, Edinburgh, Scotland.
http://www.cisco.com/web/about/doing_business/legal/cri/index.html

From paitken@cisco.com  Mon Mar 15 06:17:30 2010
Return-Path: <paitken@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 8DB203A67A3 for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 06:17:30 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -7.882
X-Spam-Level: 
X-Spam-Status: No, score=-7.882 tagged_above=-999 required=5 tests=[AWL=2.717,  BAYES_00=-2.599, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id XQSr-o8qffip for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 06:17:28 -0700 (PDT)
Received: from ams-iport-1.cisco.com (ams-iport-1.cisco.com [144.254.224.140]) by core3.amsl.com (Postfix) with ESMTP id 825CB3A68B9 for <ipfix@ietf.org>; Mon, 15 Mar 2010 06:16:49 -0700 (PDT)
Authentication-Results: ams-iport-1.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AhsBACvQnUuQ/uCWe2dsb2JhbACacRUBAQsLJAYcny6XWYR7BA
X-IronPort-AV: E=Sophos;i="4.49,643,1262563200"; d="scan'208";a="58072947"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 15 Mar 2010 13:16:55 +0000
Received: from [144.254.153.37] (dhcp-144-254-153-37.cisco.com [144.254.153.37]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2FDGtB8022399; Mon, 15 Mar 2010 13:16:55 GMT
Message-ID: <4B9E3347.1050801@cisco.com>
Date: Mon, 15 Mar 2010 13:16:55 +0000
From: Paul Aitken <paitken@cisco.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-GB; rv:1.8.1.22) Gecko/20090605 SeaMonkey/1.1.17 (Ubuntu-1.1.17+nobinonly-0ubuntu0.9.04.1)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com>	<4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com>	<4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com>	<4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E10EE.9050406@net.in.tum.de>
In-Reply-To: <4B9E10EE.9050406@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIX MIB
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 15 Mar 2010 13:17:30 -0000

Gerhard,

> Then, I do not understand what the following value in IPFIX MIB is 
> supposed to be:
> 
>    ipfixMeteringProcessCacheInactiveFlows OBJECT-TYPE
>        SYNTAX      Gauge32
>        MAX-ACCESS  read-only
>        STATUS      current
>        DESCRIPTION
>            "The number of Flows currently inactive at this cache."
>        ::= { ipfixMeteringProcessStatsEntry 2 }
> 
> 
> Well, I'm a coauthor of IPFIX MIB but I have to admit that it is not 
> clear to me what inactive flows are supposed to be.
> 
> Can anybody give an explanation?

The definitions of ipfixMeteringProcessCacheActiveFlows and 
ipfixMeteringProcessCacheInactiveFlows are poor.

However, when a cache is a pre-allocated block of N entries (where each 
entry may represent a flow), then

N = (ipfixMeteringProcessCacheActiveFlows + 
ipfixMeteringProcessCacheInactiveFlows)

where

* ipfixMeteringProcessCacheActiveFlows is the number of flows currently 
in the cache (ie, the number of cache entries currently in use) and

* ipfixMeteringProcessCacheInactiveFlows is the number of further cache 
entries capable of holding new flows.


With a dynamic cache, ipfixMeteringProcessCacheInactiveFlows may 
indicate how much further the cache can grow, or it may be meaningless.


An alternative (and contradictory) explanation is that 
ipfixMeteringProcessCacheInactiveFlows is the number of flows in the 
cache which have exceeded the inactive timeout but have not yet been 
expired. However, given the transient nature of this statistic, it 
doesn't seem to be a very useful one.


Cheers,
P.

-- 
Paul Aitken
Cisco Systems Ltd, Edinburgh, Scotland.
http://www.cisco.com/web/about/doing_business/legal/cri/index.html

From bclaise@cisco.com  Mon Mar 15 06:20:00 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 3037E3A67A5 for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 06:20:00 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.298
X-Spam-Level: 
X-Spam-Status: No, score=-1.298 tagged_above=-999 required=5 tests=[AWL=-1.300, BAYES_50=0.001, HTML_MESSAGE=0.001]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 6NxZPnLk-ks6 for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 06:19:53 -0700 (PDT)
Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id 3060D3A697A for <ipfix@ietf.org>; Mon, 15 Mar 2010 06:18:53 -0700 (PDT)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2FDIwiC017981; Mon, 15 Mar 2010 14:18:58 +0100 (CET)
Received: from [10.55.43.57] (ams-bclaise-8718.cisco.com [10.55.43.57]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2FDIsvo013973; Mon, 15 Mar 2010 14:18:54 +0100 (CET)
Message-ID: <4B9E33BE.7070907@cisco.com>
Date: Mon, 15 Mar 2010 14:18:54 +0100
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: Brian Trammell <trammell@tik.ee.ethz.ch>
References: <4AF73525.8050009@net.in.tum.de> <4AF8F999.3000207@cisco.com> <F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>
In-Reply-To: <F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>
Content-Type: multipart/alternative; boundary="------------020109060105080704070801"
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 15 Mar 2010 13:20:00 -0000

This is a multi-part message in MIME format.
--------------020109060105080704070801
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Dear all,

We've been thinking about this one, and we see 3 solutions. We believe 
that the solution 3 is the way to go, as we show below.

_Solution 1. _
Consider that the semantic is out of scope for this document.
This is the easiest solution.  However, we understand it's not right: we 
would not do a complete job wrt to IPFIX structured data
As Gerhard was expressing, the collector has not clue how to treat the 
example of a BasicList of egress interfaces in a Flow Record.
     - Has every counted packet been sent on every egress interface?
       => multicast case, AND semantic
     - Has every counted packet been sent on any one of the egress 
interfaces?
      => load balancing case, OR semantic

_Soltution 2_
We could focus on the logical AND and OR semantics only , by defining 
semantic lists, such as andBasicList, andSubTemplateList, 
andSubTemplateMultiList, orBasicList, orSubTemplateList, and 
orSubTemplateMultiList
So 6 I.E.s in total, describing AND and OR semantics.
While it's solved the router and most mediation function needs today, we 
understand that this is not a complete solution.
Gerhard, in one of his old draft, draft-sommer-ipfix-mediator-ext-01, 
proposed ADTs for orderedList, orderedPair, and portRanges which allow 
the definition new IEs for port ranges etc.
Even if this solution 2 is extensible, in the long term, will lead to an 
explosion of IEs: 3 list types * the semantic type, where the semantic 
can be (and, or, orderedList, orderedPair, portRanges, random, etc...)

_Soltution 3
_We propose to add a semantic field in the 3 list types.
Something such as:

     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|0|               Field ID    |       Element Length            |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Semantic  |             BasicList Content ...                 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                           ...                                 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

This semantic field would be a new IANA registry, we could be populated 
initially with NONE, OR, AND, ORDERED, etc... Up to discussion.
The advantage of this solution is that it's extensible, and it doesn't 
need new IEs for each semantic.
If ever required (mainly in a mediation function), we can model  very 
complex semantic, eg, "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR 
linecard2", which could be the new observation point from a aggregated 
Flow Record



_Conclusion:_
After much debate internally, we believe that we should do the effort to 
include this semantic now, and not postpone the problem, which will lead 
to an explosion of IEs in the future.
Personally, I was initially against the solution 3 ... mostly due to the 
effort required to modify the complete specs.
We're now ready to modify the specifications in the IPFIX structured 
data, but we would like to get your feedback and agreement in advance as 
this is not a small piece of work.

Please comment.

Regards, Paul, Stan, Gowri, and Benoit.
> hi Benoit, Gerhard,
>
> In this case I'm strongly in favor of leaving semantics out of 
> structured data. Structured data defines containers. The semantics of 
> the containers as a whole and the elements change based upon the 
> information elements within the container and within the record 
> containing it. Wedging semantics into the structured data elements 1. 
> risks further proliferation of (potentially non-interoperable) ways to 
> represent the same thing, 2. gives us more ways to represent 
> nonsensical things (an OR basicList of MPLS stack 
> entries...means...what?), and 3. risks defining an inadequate semantic 
> representation mechanism (what about semantics for records not using 
> structured data? what about ordered versus unordered sets? what about 
> OR basicList vs AND basicList vs nested AND and OR basicLists vs just 
> three identical IEs?). If we really want an unambiguous semantic 
> framework for IPFIX (and here I'm not convinced either way) that's 
> best done on its own, addressing things at the information element and 
> record level. Doing it here confuses the issue.
>
> Regards,
>
> Brian
>
> On Nov 10, 2009, at 6:26 AM, Benoit Claise wrote:
>
>> Gerhard,
>>
>> Thanks for your email.
>> I have no strong feelings about the two solutions you proposed.
>>
>> From a pure router point of view,  I don't see any use cases for 
>> logical OR in exporting flow records.
>> However, from a IPFIX Mediator point of view, I see some use cases.
>> I mean that it requires an Intermediate Aggregation Process or 
>> Intermediate Correlation Process to express: I've seen this flow 
>> record OR that flow record.
>> Now, it's true that even routers will have mediation functions...
>>
>> I'm inclined to add orBasicList, orSubTemplateList, 
>> orSubTempalteMultiList to the draft (This is a small addition after 
>> all) and to express that, by default, a logical AND is assumed... 
>> specifically if the structured data is used in the IPFIX Mediation 
>> Protocol.
>>
>> I'm not convinced by the NOT in the context of structured data, as we 
>> don't even have a concept of NOT for a single information element!
>>
>> I would like to get some more feedback from others.
>>
>> Regards, Benoit.
>>
>>> Dear all,
>>>
>>> Regarding draft-ietf-ipfix-structured-data, I see the risk that the
>>> semantic of the exported structured data is not clear.
>>>
>>> How do you interpret the manifold occurrence of the same Information
>>> Element (basicList) or the same group of Information Elements
>>> (subTemplateList) in one record?
>>>
>>> What does it mean if basicList, subTemplateList, or 
>>> subTemplateMultiList
>>> is used for a Flow Key field? Or non-key field?
>>>
>>> Some Examples:
>>>
>>> - BasicList of egress interfaces in a Flow Record
>>>  How should a Flow Record be interpreted which contains a list of
>>>  egress interfaces and a packet counter?
>>>  Has every counted packet been sent on every egress interface?
>>>    => multicast case, AND semantic (see example in section 8.1)
>>>  Has every counted packet been sent on any one of the egress
>>>  interfaces?
>>>    => load balancing case, OR semantic
>>>  Can it be used as a Flow Key or not?
>>>
>>> - BasicList of destination ports in a Flow Record
>>>  As every packet has only one destination port, the only reasonable
>>>  interpretation is that the Flow contains packets having one of
>>>  the reported port numbers.
>>>    => OR semantic
>>>  This would be a non-key field.
>>>
>>>
>>> I think there are two solutions:
>>>
>>> 1. We decide that the semantic of list content is out of scope of
>>>   draft-ietf-ipfix-structured-data. We add a note to the draft that
>>>   the semantic must be clear from the context or the definition of the
>>>   Information Elements used within the lists.
>>>
>>> 2. We define semantic lists, such as
>>>   - andBasicList, andSubTemplateList, andSubTemplateMultiList
>>>   - orBasicList, orSubTemplateList, orSubTempalteMultiList
>>>   describing AND and OR semantic of the contained IEs/Templates,
>>>   respectively.
>>>
>>>
>>> As I wrote in an earlier mail, I see a good use case for 
>>> orBasicList. It
>>> could be used in the Selector Report Interpretation of Property Match
>>> Filtering to report a filter like "port 80 or port 443".
>>> http://www.ietf.org/mail-archive/web/ipfix/current/msg04856.html
>>> At the moment, the Selector Report Interpretation is limited to AND.
>>> However, if we also want to express a NOT, we still need another 
>>> solution...
>>>
>>> Regards,
>>> Gerhard
>>>
>>> _______________________________________________
>>> IPFIX mailing list
>>> IPFIX@ietf.org
>>> https://www.ietf.org/mailman/listinfo/ipfix
>>>
>>
>> _______________________________________________
>> IPFIX mailing list
>> IPFIX@ietf.org
>> https://www.ietf.org/mailman/listinfo/ipfix


--------------020109060105080704070801
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <meta content="text/html; charset=ISO-8859-1"
 http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Dear all,<br>
<br>
We've been thinking about this one, and we see 3 solutions. We
believe that the solution 3 is the way to go, as we show below.<br>
<br>
<u>Solution 1. </u><br>
Consider that the semantic is out of scope for this document. <br>
This is the easiest solution.&nbsp; However, we understand it's not right:
we would not do a complete job wrt to IPFIX structured data<br>
As Gerhard was expressing, the collector has not clue how to treat the
example of a BasicList of egress interfaces in a Flow Record. <br>
&nbsp;&nbsp;&nbsp; - Has every counted packet been sent on every egress interface?<br>
&nbsp; &nbsp;&nbsp;&nbsp; =&gt; multicast case, AND semantic<br>
&nbsp;&nbsp;&nbsp; - Has every counted packet been sent on any one of the egress
interfaces?<br>
&nbsp;&nbsp;&nbsp;&nbsp; =&gt; load balancing case, OR semantic<br>
<br>
<meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
<meta name="ProgId" content="PowerPoint.Slide">
<u>Soltution 2</u><br>
We could focus on the logical AND and OR semantics only
<meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
<meta name="ProgId" content="PowerPoint.Slide">
<meta name="Generator" content="Microsoft PowerPoint 12">
, by defining semantic lists,
such as andBasicList, andSubTemplateList, andSubTemplateMultiList,
orBasicList, orSubTemplateList, and orSubTemplateMultiList<br>
So 6 I.E.s in total, describing AND and OR semantics.<br>
While it's solved the router and most mediation function needs today,
we understand that this is not a complete solution.<br>
Gerhard, in one of his old draft, draft-sommer-ipfix-mediator-ext-01,
proposed ADTs for orderedList, orderedPair, and portRanges which allow
the definition new IEs for port ranges etc.<br>
Even if this solution 2 is extensible, in the long term, will lead to
an explosion of IEs: 3 list types * the semantic type, where the
semantic can be (and, or, orderedList, orderedPair, portRanges, random,
etc...)<br>
<br>
<meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
<meta name="ProgId" content="PowerPoint.Slide">
<u>Soltution 3<br>
</u>We propose to add a semantic field in the 3 list types.<br>
Something such as:<br>
<meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
<meta name="ProgId" content="PowerPoint.Slide">
<meta name="Generator" content="Microsoft PowerPoint 12">
<br>
<div
 style="line-height: 75%; margin-top: 9pt; margin-bottom: 0pt; margin-left: 0.32in; text-indent: -0.32in; text-align: left; direction: ltr; unicode-bidi: embed; vertical-align: baseline;">
<pre>&nbsp;&nbsp;&nbsp; +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|0| &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Field ID &nbsp;&nbsp;&nbsp;|&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Element Length&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp;&nbsp; |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Semantic  | &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;BasicList Content ... &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;...&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+</pre>
</div>
<div
 style="line-height: 75%; margin-top: 9pt; margin-bottom: 0pt; margin-left: 0.32in; text-indent: -0.32in; text-align: left; direction: ltr; unicode-bidi: embed; vertical-align: baseline;"><!--[if ppt]--></div>
This
semantic field would be a new IANA registry, we could be populated
initially with NONE, OR, AND, ORDERED, etc... Up to discussion.<br>
The advantage of this solution is that it's extensible, and it doesn't
need new IEs for each semantic.<br>
If ever required (mainly in a mediation function), we can model&nbsp; very
complex semantic, eg, "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR
linecard2", which could be the new observation point from a aggregated
Flow Record<br>
<br>
<br>
<br>
<u>Conclusion:</u><br>
After much debate internally, we believe that we should do the effort
to include this semantic now, and not postpone the problem, which will
lead to an explosion of IEs in the future.<br>
Personally, I was initially against the solution 3 ... mostly due to
the effort required to modify the complete specs.&nbsp; <br>
We're now ready to modify the specifications in the IPFIX structured
data, but we
would like to get your feedback and agreement in advance as this is not
a small piece
of work.<br>
<br>
Please comment.<br>
<br>
Regards, Paul, Stan, Gowri, and Benoit.
<blockquote
 cite="mid:F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch"
 type="cite">hi Benoit, Gerhard,
  <br>
  <br>
In this case I'm strongly in favor of leaving semantics out of
structured data. Structured data defines containers. The semantics of
the containers as a whole and the elements change based upon the
information elements within the container and within the record
containing it. Wedging semantics into the structured data elements 1.
risks further proliferation of (potentially non-interoperable) ways to
represent the same thing, 2. gives us more ways to represent
nonsensical things (an OR basicList of MPLS stack
entries...means...what?), and 3. risks defining an inadequate semantic
representation mechanism (what about semantics for records not using
structured data? what about ordered versus unordered sets? what about
OR basicList vs AND basicList vs nested AND and OR basicLists vs just
three identical IEs?). If we really want an unambiguous semantic
framework for IPFIX (and here I'm not convinced either way) that's best
done on its own, addressing things at the information element and
record level. Doing it here confuses the issue.
  <br>
  <br>
Regards,
  <br>
  <br>
Brian
  <br>
  <br>
On Nov 10, 2009, at 6:26 AM, Benoit Claise wrote:
  <br>
  <br>
  <blockquote type="cite">Gerhard,
    <br>
    <br>
Thanks for your email.
    <br>
I have no strong feelings about the two solutions you proposed.
    <br>
    <br>
>From a pure router point of view,&nbsp; I don't see any use cases for
logical OR in exporting flow records.
    <br>
However, from a IPFIX Mediator point of view, I see some use cases.
    <br>
I mean that it requires an Intermediate Aggregation Process or
Intermediate Correlation Process to express: I've seen this flow record
OR that flow record.
    <br>
Now, it's true that even routers will have mediation functions...
    <br>
    <br>
I'm inclined to add orBasicList, orSubTemplateList,
orSubTempalteMultiList to the draft (This is a small addition after
all) and to express that, by default, a logical AND is assumed...
specifically if the structured data is used in the IPFIX Mediation
Protocol.
    <br>
    <br>
I'm not convinced by the NOT in the context of structured data, as we
don't even have a concept of NOT for a single information element!
    <br>
    <br>
I would like to get some more feedback from others.
    <br>
    <br>
Regards, Benoit.
    <br>
    <br>
    <blockquote type="cite">Dear all,
      <br>
      <br>
Regarding draft-ietf-ipfix-structured-data, I see the risk that the
      <br>
semantic of the exported structured data is not clear.
      <br>
      <br>
How do you interpret the manifold occurrence of the same Information
      <br>
Element (basicList) or the same group of Information Elements
      <br>
(subTemplateList) in one record?
      <br>
      <br>
What does it mean if basicList, subTemplateList, or
subTemplateMultiList
      <br>
is used for a Flow Key field? Or non-key field?
      <br>
      <br>
Some Examples:
      <br>
      <br>
- BasicList of egress interfaces in a Flow Record
      <br>
&nbsp;How should a Flow Record be interpreted which contains a list of
      <br>
&nbsp;egress interfaces and a packet counter?
      <br>
&nbsp;Has every counted packet been sent on every egress interface?
      <br>
&nbsp;&nbsp; =&gt; multicast case, AND semantic (see example in section 8.1)
      <br>
&nbsp;Has every counted packet been sent on any one of the egress
      <br>
&nbsp;interfaces?
      <br>
&nbsp;&nbsp; =&gt; load balancing case, OR semantic
      <br>
&nbsp;Can it be used as a Flow Key or not?
      <br>
      <br>
- BasicList of destination ports in a Flow Record
      <br>
&nbsp;As every packet has only one destination port, the only reasonable
      <br>
&nbsp;interpretation is that the Flow contains packets having one of
      <br>
&nbsp;the reported port numbers.
      <br>
&nbsp;&nbsp; =&gt; OR semantic
      <br>
&nbsp;This would be a non-key field.
      <br>
      <br>
      <br>
I think there are two solutions:
      <br>
      <br>
1. We decide that the semantic of list content is out of scope of
      <br>
&nbsp; draft-ietf-ipfix-structured-data. We add a note to the draft that
      <br>
&nbsp; the semantic must be clear from the context or the definition of the
      <br>
&nbsp; Information Elements used within the lists.
      <br>
      <br>
2. We define semantic lists, such as
      <br>
&nbsp; - andBasicList, andSubTemplateList, andSubTemplateMultiList
      <br>
&nbsp; - orBasicList, orSubTemplateList, orSubTempalteMultiList
      <br>
&nbsp; describing AND and OR semantic of the contained IEs/Templates,
      <br>
&nbsp; respectively.
      <br>
      <br>
      <br>
As I wrote in an earlier mail, I see a good use case for orBasicList.
It
      <br>
could be used in the Selector Report Interpretation of Property Match
      <br>
Filtering to report a filter like "port 80 or port 443".
      <br>
<a class="moz-txt-link-freetext" href="http://www.ietf.org/mail-archive/web/ipfix/current/msg04856.html">http://www.ietf.org/mail-archive/web/ipfix/current/msg04856.html</a>
      <br>
At the moment, the Selector Report Interpretation is limited to AND.
      <br>
However, if we also want to express a NOT, we still need another
solution...
      <br>
      <br>
Regards,
      <br>
Gerhard
      <br>
      <br>
_______________________________________________
      <br>
IPFIX mailing list
      <br>
<a class="moz-txt-link-abbreviated" href="mailto:IPFIX@ietf.org">IPFIX@ietf.org</a>
      <br>
<a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/ipfix">https://www.ietf.org/mailman/listinfo/ipfix</a>
      <br>
      <br>
    </blockquote>
    <br>
_______________________________________________
    <br>
IPFIX mailing list
    <br>
<a class="moz-txt-link-abbreviated" href="mailto:IPFIX@ietf.org">IPFIX@ietf.org</a>
    <br>
<a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/ipfix">https://www.ietf.org/mailman/listinfo/ipfix</a>
    <br>
  </blockquote>
</blockquote>
<br>
</body>
</html>

--------------020109060105080704070801--

From andrjohn@cisco.com  Mon Mar 15 06:23:28 2010
Return-Path: <andrjohn@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id CB1293A6979 for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 06:23:28 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -10.598
X-Spam-Level: 
X-Spam-Status: No, score=-10.598 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id OBdjtw2nVrTy for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 06:23:25 -0700 (PDT)
Received: from ams-iport-1.cisco.com (ams-iport-1.cisco.com [144.254.224.140]) by core3.amsl.com (Postfix) with ESMTP id C25433A69BD for <ipfix@ietf.org>; Mon, 15 Mar 2010 06:22:11 -0700 (PDT)
Authentication-Results: ams-iport-1.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AhsBAFfRnUuQ/uCWe2dsb2JhbACacRUBAQsLJAYcnzGLA4xWhHsE
X-IronPort-AV: E=Sophos;i="4.49,643,1262563200"; d="scan'208,217";a="58073195"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 15 Mar 2010 13:22:17 +0000
Received: from cisco.com (mrwint.cisco.com [64.103.71.48]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2FDMHjT023670; Mon, 15 Mar 2010 13:22:17 GMT
Received: from [144.254.153.59] (dhcp-144-254-153-59.cisco.com [144.254.153.59]) by cisco.com (8.11.7p3+Sun/8.8.8) with ESMTP id o2FDMHX22433; Mon, 15 Mar 2010 13:22:17 GMT
Message-ID: <4B9E3498.4090103@cisco.com>
Date: Mon, 15 Mar 2010 13:22:32 +0000
From: Andrew Johnson <andrjohn@cisco.com>
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
MIME-Version: 1.0
To: Brian Trammell <trammell@tik.ee.ethz.ch>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B9257CE.6040409@cisco.com> <4B98E288.4040204@net.in.tum.de> <4B991389.2090209@cisco.com> <4B9A4B15.3010402@net.in.tum.de> <4B9A634C.4080503@cisco.com> <3BC9C9EF-4A1C-4A48-9AB6-E7CE5EF0C0BC@tik.ee.ethz.ch>
In-Reply-To: <3BC9C9EF-4A1C-4A48-9AB6-E7CE5EF0C0BC@tik.ee.ethz.ch>
Content-Type: multipart/alternative; boundary="------------050809090009080405040303"
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 15 Mar 2010 13:23:29 -0000

This is a multi-part message in MIME format.
--------------050809090009080405040303
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit



Brian Trammell wrote:
> hi Andrew, Gerhard,
>
> Let me step in here quickly, as I can easily see a place for a multicache MP.. the following is something I was asked to do once upon a time:
>
> Suppose we have an MP at an observation point in a security monitoring context. This MP is driving two separate processes: one which filters traffic according to some current set of criteria defining "bad" traffic, then displays in "real time" (~5s delay) this "bad" traffic volume (the fast stream), and one which prefilters and stores full flow data for retrospective analysis (the full stream). The fast stream gets used to drive a display and maybe turn on a blinkenlight somewhere, then it gets thrown away. The full stream has richer data about each flow and gets stored and used by the analysis tools for doing actual work.
>
> It seemed to me at the time that the right way to handle this situation is two caches in the same MP: one with very fast active timeouts (we don't care about termination conditions, we just want to see flow-level volumes), and one with reasonable timeouts (because the analyses are simpler when you don't have to stitch together all the flows from active timeouts.) Everything else -> observation point, selection process, even the EP and CP, are the same.
>
> Or is this two MPs in your reading of the architecture?
>   
I would treat then as two different Metering Processes.  I have no 
problem with 2 MPs sharing the same Observation Point and Exporting 
Process.  In your example one cache is receiving a filtered subset of 
traffic, so the Selection Process is not the same.

I don't see the advantage to having more than one cache per Metering 
Process.  I do see a disadvantage in the complexity it brings.  For 
example, if you were to have implemented the above system and the 
Metering Process Statistics Options and/or the Metering Process 
Reliability Statistics, would you count a packet in the 
"ignoredPacketTotalCount" if it successfully entered one cache and not 
the other?

I would say you should only report the loss if both caches failed to 
observe the packet.  From RFC 5470, section 6.4. Reporting Responsibilities:

     From time to time, an IPFIX Device may not be able to observe all
    the packets reaching one of its Observation Points.  This could
    occur if a Metering Process finds itself temporarily short of
    resources... In such situations, the IPFIX Device should attempt to
    count the number of packet losses that have occurred, and report
    them to its Collector(s).

Would it not be more useful to have separate statistics per cache?


Furthermore, the current suggestion is that the configuration draft 
could configure two caches, each populated from completely different 
Observation Points, Selection Process and Exporting to different 
destinations, yet they could be the same Metering Process.  If we 
configure four caches then we would expect between one and four Metering 
Processes to be configured, possibly grouping the caches in one of a 
hundred different ways.

The only reason for this seems to be that it would be convenient to have 
different caches share Selection configuration.  What I don't see is why 
a shared Selector configuration must result in a shared Selection 
Process and therefore a single Metering Process.

Cheers

Andrew


> Cheers,
>
> Brian
>
> On Mar 12, 2010, at 4:52 PM, Andrew Johnson wrote:
>
>   
>> Hi Gerhard
>>
>> Gerhard Muenz wrote:
>>     
>>> Hi Andrew, all,
>>> Andrew Johnson wrote:
>>>       
>>>> Gerhard Muenz wrote:
>>>>         
>>>>>>>> ======================================================================== 
>>>>>>>> Section 3:
>>>>>>>> ======================================================================== 
>>>>>>>> If a Selection Process is part of a Metering Process and a single
>>>>>>>> Selection Process can feed into more than one Cache, then it is
>>>>>>>> implied that a single Metering Process can have more than once cache.
>>>>>>>>                 
>>>>>>> Note: The text has slightly changed after Atsushi's review. Now, it explicitly says that a Metering Process may contain multiple Selection Processes and multiple Caches.
>>>>>>>               
>>>>>> If a Metering Process can have multiple caches and an IPFIX device can have multiple Metering Processes, how do we associate the caches with Metering Processes?  Can this configuration system only be used to configure a single Metering Process within a device?
>>>>>>             
>>>>> The model does not group Selection Processes and Caches into Metering Processes. This is done on purpose because there is not a single correct view.
>>>>>
>>>>> All Selection Process and all Caches of a Monitoring Device can be regarded as one Metering Process.
>>>>>
>>>>> Alternatively, Selection Processes and Caches which are not directly or indirectly linked to each other can be seen as belonging to different Metering Processes.
>>>>>           
>>>> OK, at this point I think I'm uncomfortable with the idea that a Metering Process can have more than one cache.  What is the reasoning behind this decision?
>>>>         
>>> It should be possible that the output of the same Selection Process enters two different Caches.
>>> Example:
>>>          +-> Cache 1 (PSAMP)
>>> OP -> SP -+
>>>          +-> Cache 2 (IPFIX)
>>> It is difficult to say that this corresponds to two MPs.
>>>       
>> Is the only argument for these two caches being in the same MP is so they can share a selection process?
>>
>> I say that the cache is the heart of a Metering Process and it should only have one!  I realise that it may not be explicitly stated in current IPFIX documents, but I think it's implied in the Architecture document and is the paradigm in which IPFIX and PSAMP have been developed.
>>
>> It seems that what you really object to is the PSAMP declaration that a selection process belongs to a single Metering Process, but as we've seen, the statistics held by a Metering Process need to be kept on a per-cache basis.
>>
>> I don't object to the configuration shorthand of using the same config across different Metering Processes, but with this model the stats are going to be quite different to those needed for PSAMP.  That's fine though!
>>
>>
>>     
>>>>>>> Furthermore, it says that only packets observed at a single Observation Point may enter a Selection Process. IMO, this results from the PSAMP definitions of Observed Packet Stream and Selection Process. Please tell me if this conclusion is wrong.
>>>>>>>               
>>>>>> A Selection process should only have a single feed, but can have multiple observers.  This results in the number of possible chains (Selection Sequences) being equal to the number of Selectors that directly feed a cache.
>>>>>>             
>>>>> Ok. Single feed means single Observation Point as specified in the draft.
>>>>>           
>>>> No, by "feed" I mean input stream.  Either from an Observation point or the output of a different Selection Process.
>>>>         
>>> Yes, this is what I wanted to say. Therefore, we have "0..1" in -05 (it was "0..*" in -04):
>>>                        +------------------+
>>>                        | ObservationPoint |
>>>                        +------------------+
>>>                             0..1 |
>>>                             ==== |
>>>                                  |
>>>                             0..* V
>>>                        +------------------+
>>>                        | SelectionProcess |
>>>                        +------------------+<-+
>>>                             0..* |  0..1 |   | 0..*
>>>                                  |  ==== +---+
>>>                             0..* V
>>>                        +------------------+
>>>                        | Cache            |
>>>                        +------------------+
>>>                             0..* |
>>>                                  |
>>>                             0..* V
>>>                        +------------------+
>>>                        | ExportingProcess |
>>>                        +------------------+
>>> Sorry, if this was not clear from the beginning.
>>>       
>> This has always been clear... I'm not sure what the original points were now.
>>
>>
>>     
>>>>>>>> ======================================================================== 
>>>>>>>> >From section 4.1:
>>>>>>>> ======================================================================== 
>>>>>>>> Doesn't the direction belong in the Observation Point class, not the
>>>>>>>> Interface or LC class.
>>>>>>>>                 
>>>>>>> If the current position of the direction parameter bothers you, we can move it.
>>>>>>>
>>>>>>> However, my reasoning is the following:
>>>>>>> - If it appears in the ObservationPoint class, it looks like the Observation Point is doing some filtering.
>>>>>>> - If it appears in the Interface/Linecard class, it looks like the filtering is already done at the interface or linecard.
>>>>>>>
>>>>>>> I thought that the interface/linecard is doing the filtering.
>>>>>>>               
>>>>>> I don't see it as filtering, just part of the definition of the Observation Point.  The Interface is most likely able to send and receive traffic, but the observation point could be on only the send or receive channel.  i.e. It's not a input only interface, but the Observation Point could only see input packets.
>>>>>>
>>>>>> Currently you have a 1:1 mapping of OP to interface.  If the direction is moved to the Observation Point then you could have two observation points on the same interface, one on the input side and one on the output.
>>>>>>             
>>>>> With the current model, there is no problem to configure two OPs on the same interface:
>>>>>
>>>>>     <observationPoint>
>>>>>       <name>interface ingress</name>
>>>>>       <interface>
>>>>>         <ifIndex>3</ifIndex>
>>>>>         <direction>ingress</direction>
>>>>>       </interface>
>>>>>       ...
>>>>>     </observationPoint>
>>>>>
>>>>>     <observationPoint>
>>>>>       <name>interface egress</name>
>>>>>       <interface>
>>>>>         <ifIndex>3</ifIndex>
>>>>>         <direction>egress</direction>
>>>>>       </interface>
>>>>>       ...
>>>>>     </observationPoint>
>>>>>           
>>>> OK, but my point was always that the interface is not defined by its direction.
>>>>
>>>> These two "interfaces" are actually two observation points on the same interface:
>>>>        <interface>
>>>>          <ifIndex>3</ifIndex>
>>>>          <direction>ingress</direction>
>>>>        </interface>
>>>>
>>>>        <interface>
>>>>          <ifIndex>3</ifIndex>
>>>>          <direction>ingress</direction>
>>>>        </interface>
>>>>         
>>> Ok. So it seems to you that <direction> belongs to the definition of the interface.
>>> If we moved <direction> to the Observation Point, we could remove the Interface/Linecard containers because there would be only one parameter left:
>>>    +-------------------------------------------------+
>>>    | ObservationPoint                                |
>>>    +-------------------------------------------------+
>>>    | name                                            |
>>>    | observationPointId {readOnly}                   |
>>>    | observationDomainId                             |
>>>    | ifIndex/ifName/entPhysicalIndex/entPhysicalName | 0..*
>>>    | direction                                       |----+
>>>    +-------------------------------------------------+    | 0..*
>>>                                                           V
>>>                                          +--------------------+
>>>                                          | SelectionProcess   |
>>>                                          +--------------------+
>>> Would you agree with this change?
>>>       
>> Yes, this seems fine.
>>
>>
>>     
>>>>>>> There is another small advantage: The current position of the parameter makes it easier to add vendor or implementation-specific packet sources for which the direction parameter has no meaning, for example:
>>>>>>>
>>>>>>> +------------------+
>>>>>>> | PcapFile         |
>>>>>>> +------------------+
>>>>>>> | filename         |
>>>>>>> +------------------+
>>>>>>>
>>>>>>>               
>>>>>>>> ======================================================================== 
>>>>>>>> >From Section 4.3.1
>>>>>>>> ======================================================================== 
>>>>>>>> Isn't is possible to generate a cache without any keys?
>>>>>>>> This would generate a cache with a single flow?  Do we allow this?
>>>>>>>>                 
>>>>>>> A the moment, there is no restriction which says that a flow key field must exist.
>>>>>>>
>>>>>>> Do you want to add such a restriction?
>>>>>>>               
>>>>>> A cache without keys will result in a single flow, which seems a bit daft.  I guess there is no pressing reason to add this restriction though.
>>>>>>
>>>>>>             
>>>>>>>> Would it be useful to have two lists, one for keys and one for
>>>>>>>> non-keys?
>>>>>>>>                 
>>>>>>> There is an argument for having a single list:
>>>>>>>
>>>>>>> A Cache implementation may generate records using the specified field order. So, the fields occur in the same order as specified in the configuration.
>>>>>>>               
>>>>>> If we're worried about implementation specifics can't they extend the YANG model?
>>>>>>             
>>>>> Yes, they can.
>>>>>
>>>>>           
>>>>>>> With separate lists for flow key fields and non-key fields, we cannot arbitrarily mix these two types of fields.
>>>>>>>
>>>>>>> Is there any advantage of having separate lists?
>>>>>>>               
>>>>>> I was just thinking about how you'd specify that at least one key was required.
>>>>>>             
>>>>> Caches of type "immediate" do not have Flow Keys. So, the restrictions would not apply to such Caches.
>>>>>           
>>>> Really?  Why not say that Caches of an "immediate" type have no non-keys?  For example, the counter fields should not be key fields, they make no sense as keys.  They also make no sense with the immediate cache.  It seems that an immediate cache is made up of entirely key fields.
>>>>         
>>> IMO, a Packet Report does not contain any Flow information, so it does not make sense to talk about Flow Keys.
>>> In -05, I restricted the <isFlowKey> parameter to Caches generating Flow Records because I think that it does not make sense to distinguish Flow Key fields and non-key fields if Packet Reports are generated.
>>>       
>> Ah I see.  I guess it makes sense to simply not say anything about whether it is a key or not.  So these fields are not keys, but neither are they non-keys.  The are simply collected fields.  Certain fields only make sense for flows and not for packet reports, which might be worth mentioning.
>>
>>
>>     
>>>>>>>> For a permanent cache, update counters should be used and for a
>>>>>>>> timeout cache the delta counters should be used.  Where is this
>>>>>>>> sort of thing enforced?
>>>>>>>>                 
>>>>>>> This is not enforced. I would not strictly enforce it because strange configurations might make sense in specific use cases.
>>>>>>>               
>>>>>> OK.
>>>>>>
>>>>>>             
>>>>>>> However, we could add a note that totalCounters are typically used for "permanent" and deltaCounters for "timeout".
>>>>>>>
>>>>>>> Could you provide a sentence or two?
>>>>>>>               
>>>>>> Perhaps a more general note that not all fields will be appropriate for all configurations. Is the Exporting Device free to change the fields
>>>>>> if it needs total counters for a permanent cache?
>>>>>>             
>>>>> No.
>>>>>
>>>>> I added a new paragraph at the end of 4.3.1. (CacheLayout Class):
>>>>>
>>>>>   Note that the use of Information Elements can be restricted to
>>>>>   certain Cache Modes or to Flow Key or non-key fields.  Such
>>>>>   restrictions may result from Information Element definitions or from
>>>>>   device-specific constraints.  According to Section 5, the Monitoring
>>>>>   Device MUST notify the user if a Cache field cannot be configured
>>>>>   with the given Information Element.
>>>>>
>>>>> Ok?
>>>>>           
>>>> Yes, seems reasonable.
>>>>
>>>>
>>>>         
>>>>>>>> ======================================================================== 
>>>>>>>> >From Section 4.4.1
>>>>>>>> ======================================================================== 
>>>>>>>>   exportMemberType:  Configures the export member type that corresponds
>>>>>>>>      to the ipfixTransportSessionGroupMemberType object in
>>>>>>>>      [I-D.ietf-ipfix-mib].  The following parameter values are
>>>>>>>>      specified by the configuration data model:
>>>>>>>>      *  primary: primary target of the Exporting Process
>>>>>>>>      *  secondary: secondary target of the Exporting Process used when
>>>>>>>>         the primary target is not reachable
>>>>>>>>      *  parallel: parallel exporting to all destinations and files of
>>>>>>>>         the Exporting Process
>>>>>>>>      *  loadBalancing: load-balancing between different destinations
>>>>>>>>         and files of the Exporting Process
>>>>>>>>      "parallel" is the default value if this parameter is not
>>>>>>>>      configured.  If one destination or file is configured as "primary"
>>>>>>>>
>>>>>>>> ------------------------------------------------------------------------ 
>>>>>>>> Comments
>>>>>>>> ------------------------------------------------------------------------ 
>>>>>>>> It's not clear to me that this parameter belongs to the Destination
>>>>>>>> class.  The number of restrictions on when you can use the various
>>>>>>>> types suggests that we could simplify things.
>>>>>>>>
>>>>>>>> A Cache can pass flow records to multiple Exporting Processes, so we
>>>>>>>> don't really need a parallel option.  This would mean that each
>>>>>>>> Exporting Process is responsible for sending one complete set of
>>>>>>>> records.  i.e. one record in, one record out.  It could do that, either
>>>>>>>> using a redundancy model or with a load-balancing one.
>>>>>>>>
>>>>>>>> In other words, the exportType would be an Exporting Process attribute
>>>>>>>> which would indicate whether the list of Destinations should be load-
>>>>>>>> balanced across or treated as Primary, Seondary, Tertiary, etc.
>>>>>>>>                 
>>>>>>> At the moment, the parameter is inline with the IPFIX-MIB:
>>>>>>>
>>>>>>> ipfixExportMemberType OBJECT-TYPE
>>>>>>>    SYNTAX      INTEGER {
>>>>>>>                    unknown(0),
>>>>>>>                    primary(1),
>>>>>>>                    secondary(2),
>>>>>>>                    parallel(3),
>>>>>>>                    loadBalancing(4)
>>>>>>>                }
>>>>>>>    MAX-ACCESS  read-only
>>>>>>>    STATUS      current
>>>>>>>    DESCRIPTION
>>>>>>>        "The type of a member Transport Session in a Transport
>>>>>>>        Session group (identified by the value of ipfixExportIndex,
>>>>>>>        ipfixObservationDomainId and ipfixMeteringProcessCacheId).
>>>>>>>        The following values are valid:
>>>>>>>
>>>>>>>        unknown(0)
>>>>>>>            This value MUST be used if the status of the group
>>>>>>>            membership cannot be detected by the equipment. This
>>>>>>>            value should be avoided as far as possible.
>>>>>>>
>>>>>>>        primary(1)
>>>>>>>            This value is used for a group member that is used as
>>>>>>>            the primary target of an Exporter. Other group members
>>>>>>>            (with the same ipfixExportIndex and
>>>>>>>            ipfixMeteringProcessCacheId) MUST NOT have the value
>>>>>>>            primary(1) but MUST have the value secondary(2).
>>>>>>>            This value MUST also be specified if the Exporter does
>>>>>>>            not support Transport Session grouping. In this case
>>>>>>>            the group contains only one Transport Session.
>>>>>>>
>>>>>>>        secondary(2)
>>>>>>>            This value is used for a group member that is used as a
>>>>>>>            secondary target of an Exporter. The Exporter will use
>>>>>>>            one of the targets specified as secondary(2) within the
>>>>>>>            same Transport Session group when the primary target is
>>>>>>>            not reachable.
>>>>>>>
>>>>>>>        parallel(3)
>>>>>>>            This value is used for a group member that is used for
>>>>>>>            duplicate exporting i.e., all group members identified> 
>>>>>>>               
>>>>>>>            by the ipfixExportIndex are exporting the same Records
>>>>>>>            in parallel. This implies that all group members MUST
>>>>>>>            have the same membertype parallel(3).
>>>>>>>
>>>>>>>        loadBalancing(4)
>>>>>>>            This value is used for a group member that is used
>>>>>>>            as one target for load-balancing. This means that a> 
>>>>>>>               
>>>>>>>            Record is sent to one of the group members in this
>>>>>>>            group identified by ipfixExportIndex.
>>>>>>>            This implies that all group members MUST have the same
>>>>>>>            membertype loadBalancing(4)."
>>>>>>>    ::= { ipfixExportEntry 2 }
>>>>>>>
>>>>>>>
>>>>>>> I'm open to better proposals.
>>>>>>>
>>>>>>> "parallel" and "loadBalancing" can be configured once with a single ExportingProcess class parameter.
>>>>>>>
>>>>>>> How would you configure "primary" and "secondary with a single Exporting Process parameter?
>>>>>>>               
>>>>> This issue is on my todo list.
>>>>>           
>>>> OK, it seems from a configuration point of view you have three modes. Parallel, loadBalancing and fallback.  In fallback mode you try the destinations in order until you get a connection.  The mode is a property of the Exporting Process and defines how to treat the list of destinations.
>>>>
>>>> i.e.
>>>>
>>>>   <Exporter>
>>>>     <name>Test 1</name>
>>>>     <mode>fallback</mode>
>>>>     <destination> ... </destination>
>>>>     <destination> ... </destination>
>>>>     <destination> ... </destination>
>>>>   </Exporter>
>>>>
>>>> So parallel mode sends to all three at once, load balancing takes it in turns and fallback treats the first as primary, the second and secondary and the third as tertiary.
>>>>         
>>> Yes, this sounds good!
>>>       
>> Great.
>>
>> Thanks
>>
>> Andrew
>> _______________________________________________
>> IPFIX mailing list
>> IPFIX@ietf.org
>> https://www.ietf.org/mailman/listinfo/ipfix
>>     
>
>
>   

--------------050809090009080405040303
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
  <title></title>
</head>
<body bgcolor="#ffffff" text="#000000">
<br>
<br>
Brian Trammell wrote:
<blockquote
 cite="mid:3BC9C9EF-4A1C-4A48-9AB6-E7CE5EF0C0BC@tik.ee.ethz.ch"
 type="cite">
  <pre wrap="">hi Andrew, Gerhard,

Let me step in here quickly, as I can easily see a place for a multicache MP.. the following is something I was asked to do once upon a time:

Suppose we have an MP at an observation point in a security monitoring context. This MP is driving two separate processes: one which filters traffic according to some current set of criteria defining "bad" traffic, then displays in "real time" (~5s delay) this "bad" traffic volume (the fast stream), and one which prefilters and stores full flow data for retrospective analysis (the full stream). The fast stream gets used to drive a display and maybe turn on a blinkenlight somewhere, then it gets thrown away. The full stream has richer data about each flow and gets stored and used by the analysis tools for doing actual work.

It seemed to me at the time that the right way to handle this situation is two caches in the same MP: one with very fast active timeouts (we don't care about termination conditions, we just want to see flow-level volumes), and one with reasonable timeouts (because the analyses are simpler when you don't have to stitch together all the flows from active timeouts.) Everything else -&gt; observation point, selection process, even the EP and CP, are the same.

Or is this two MPs in your reading of the architecture?
  </pre>
</blockquote>
I would treat then as two different Metering Processes.&nbsp; I have no
problem with 2 MPs sharing the same Observation Point and Exporting
Process.&nbsp; In your example one cache is receiving a filtered subset of
traffic, so the Selection Process is not the same.<br>
<br>
I don't see the advantage to having more than one cache per Metering
Process.&nbsp; I do see a disadvantage in the complexity it brings.&nbsp; For
example, if you were to have implemented the above system and the
Metering Process Statistics Options and/or the Metering Process
Reliability Statistics, would you count a packet in the
"ignoredPacketTotalCount" if it successfully entered one cache and not
the other?<br>
<br>
I would say you should only report the loss if both caches failed to
observe the packet.&nbsp; From RFC 5470, section 6.4. Reporting
Responsibilities:<br>
<blockquote>From time to time, an IPFIX Device may not be able to
observe all the packets reaching one of its Observation Points.&nbsp; This
could occur if a Metering Process finds itself temporarily short of
resources... In such situations, the IPFIX Device should attempt to
count the number of packet losses that have occurred, and report them
to its Collector(s).<br>
</blockquote>
Would it not be more useful to have separate statistics per cache?<br>
<br>
<br>
Furthermore, the current suggestion is that the configuration draft
could configure two caches, each populated from completely different
Observation Points, Selection Process and Exporting to different
destinations, yet they could be the same Metering Process.&nbsp; If we
configure four caches then we would expect between one and four
Metering Processes to be configured, possibly grouping the caches in
one of a hundred different ways.<br>
<br>
The only reason for this seems to be that it would be convenient to
have
different caches share Selection configuration.&nbsp; What I don't see is
why a shared Selector configuration must result in a shared Selection
Process and therefore a single Metering Process.<br>
<br>
Cheers<br>
<br>
Andrew<br>
<br>
<br>
<blockquote
 cite="mid:3BC9C9EF-4A1C-4A48-9AB6-E7CE5EF0C0BC@tik.ee.ethz.ch"
 type="cite">
  <pre wrap="">Cheers,

Brian

On Mar 12, 2010, at 4:52 PM, Andrew Johnson wrote:

  </pre>
  <blockquote type="cite">
    <pre wrap="">Hi Gerhard

Gerhard Muenz wrote:
    </pre>
    <blockquote type="cite">
      <pre wrap="">Hi Andrew, all,
Andrew Johnson wrote:
      </pre>
      <blockquote type="cite">
        <pre wrap="">Gerhard Muenz wrote:
        </pre>
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite">
              <blockquote type="cite">
                <pre wrap="">======================================================================== 
Section 3:
======================================================================== 
If a Selection Process is part of a Metering Process and a single
Selection Process can feed into more than one Cache, then it is
implied that a single Metering Process can have more than once cache.
                </pre>
              </blockquote>
              <pre wrap="">Note: The text has slightly changed after Atsushi's review. Now, it explicitly says that a Metering Process may contain multiple Selection Processes and multiple Caches.
              </pre>
            </blockquote>
            <pre wrap="">If a Metering Process can have multiple caches and an IPFIX device can have multiple Metering Processes, how do we associate the caches with Metering Processes?  Can this configuration system only be used to configure a single Metering Process within a device?
            </pre>
          </blockquote>
          <pre wrap="">The model does not group Selection Processes and Caches into Metering Processes. This is done on purpose because there is not a single correct view.

All Selection Process and all Caches of a Monitoring Device can be regarded as one Metering Process.

Alternatively, Selection Processes and Caches which are not directly or indirectly linked to each other can be seen as belonging to different Metering Processes.
          </pre>
        </blockquote>
        <pre wrap="">OK, at this point I think I'm uncomfortable with the idea that a Metering Process can have more than one cache.  What is the reasoning behind this decision?
        </pre>
      </blockquote>
      <pre wrap="">It should be possible that the output of the same Selection Process enters two different Caches.
Example:
         +-&gt; Cache 1 (PSAMP)
OP -&gt; SP -+
         +-&gt; Cache 2 (IPFIX)
It is difficult to say that this corresponds to two MPs.
      </pre>
    </blockquote>
    <pre wrap="">Is the only argument for these two caches being in the same MP is so they can share a selection process?

I say that the cache is the heart of a Metering Process and it should only have one!  I realise that it may not be explicitly stated in current IPFIX documents, but I think it's implied in the Architecture document and is the paradigm in which IPFIX and PSAMP have been developed.

It seems that what you really object to is the PSAMP declaration that a selection process belongs to a single Metering Process, but as we've seen, the statistics held by a Metering Process need to be kept on a per-cache basis.

I don't object to the configuration shorthand of using the same config across different Metering Processes, but with this model the stats are going to be quite different to those needed for PSAMP.  That's fine though!


    </pre>
    <blockquote type="cite">
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite">
              <pre wrap="">Furthermore, it says that only packets observed at a single Observation Point may enter a Selection Process. IMO, this results from the PSAMP definitions of Observed Packet Stream and Selection Process. Please tell me if this conclusion is wrong.
              </pre>
            </blockquote>
            <pre wrap="">A Selection process should only have a single feed, but can have multiple observers.  This results in the number of possible chains (Selection Sequences) being equal to the number of Selectors that directly feed a cache.
            </pre>
          </blockquote>
          <pre wrap="">Ok. Single feed means single Observation Point as specified in the draft.
          </pre>
        </blockquote>
        <pre wrap="">No, by "feed" I mean input stream.  Either from an Observation point or the output of a different Selection Process.
        </pre>
      </blockquote>
      <pre wrap="">Yes, this is what I wanted to say. Therefore, we have "0..1" in -05 (it was "0..*" in -04):
                       +------------------+
                       | ObservationPoint |
                       +------------------+
                            0..1 |
                            ==== |
                                 |
                            0..* V
                       +------------------+
                       | SelectionProcess |
                       +------------------+&lt;-+
                            0..* |  0..1 |   | 0..*
                                 |  ==== +---+
                            0..* V
                       +------------------+
                       | Cache            |
                       +------------------+
                            0..* |
                                 |
                            0..* V
                       +------------------+
                       | ExportingProcess |
                       +------------------+
Sorry, if this was not clear from the beginning.
      </pre>
    </blockquote>
    <pre wrap="">This has always been clear... I'm not sure what the original points were now.


    </pre>
    <blockquote type="cite">
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite">
              <blockquote type="cite">
                <pre wrap="">======================================================================== 
&gt;From section 4.1:
======================================================================== 
Doesn't the direction belong in the Observation Point class, not the
Interface or LC class.
                </pre>
              </blockquote>
              <pre wrap="">If the current position of the direction parameter bothers you, we can move it.

However, my reasoning is the following:
- If it appears in the ObservationPoint class, it looks like the Observation Point is doing some filtering.
- If it appears in the Interface/Linecard class, it looks like the filtering is already done at the interface or linecard.

I thought that the interface/linecard is doing the filtering.
              </pre>
            </blockquote>
            <pre wrap="">I don't see it as filtering, just part of the definition of the Observation Point.  The Interface is most likely able to send and receive traffic, but the observation point could be on only the send or receive channel.  i.e. It's not a input only interface, but the Observation Point could only see input packets.

Currently you have a 1:1 mapping of OP to interface.  If the direction is moved to the Observation Point then you could have two observation points on the same interface, one on the input side and one on the output.
            </pre>
          </blockquote>
          <pre wrap="">With the current model, there is no problem to configure two OPs on the same interface:

    &lt;observationPoint&gt;
      &lt;name&gt;interface ingress&lt;/name&gt;
      &lt;interface&gt;
        &lt;ifIndex&gt;3&lt;/ifIndex&gt;
        &lt;direction&gt;ingress&lt;/direction&gt;
      &lt;/interface&gt;
      ...
    &lt;/observationPoint&gt;

    &lt;observationPoint&gt;
      &lt;name&gt;interface egress&lt;/name&gt;
      &lt;interface&gt;
        &lt;ifIndex&gt;3&lt;/ifIndex&gt;
        &lt;direction&gt;egress&lt;/direction&gt;
      &lt;/interface&gt;
      ...
    &lt;/observationPoint&gt;
          </pre>
        </blockquote>
        <pre wrap="">OK, but my point was always that the interface is not defined by its direction.

These two "interfaces" are actually two observation points on the same interface:
       &lt;interface&gt;
         &lt;ifIndex&gt;3&lt;/ifIndex&gt;
         &lt;direction&gt;ingress&lt;/direction&gt;
       &lt;/interface&gt;

       &lt;interface&gt;
         &lt;ifIndex&gt;3&lt;/ifIndex&gt;
         &lt;direction&gt;ingress&lt;/direction&gt;
       &lt;/interface&gt;
        </pre>
      </blockquote>
      <pre wrap="">Ok. So it seems to you that &lt;direction&gt; belongs to the definition of the interface.
If we moved &lt;direction&gt; to the Observation Point, we could remove the Interface/Linecard containers because there would be only one parameter left:
   +-------------------------------------------------+
   | ObservationPoint                                |
   +-------------------------------------------------+
   | name                                            |
   | observationPointId {readOnly}                   |
   | observationDomainId                             |
   | ifIndex/ifName/entPhysicalIndex/entPhysicalName | 0..*
   | direction                                       |----+
   +-------------------------------------------------+    | 0..*
                                                          V
                                         +--------------------+
                                         | SelectionProcess   |
                                         +--------------------+
Would you agree with this change?
      </pre>
    </blockquote>
    <pre wrap="">Yes, this seems fine.


    </pre>
    <blockquote type="cite">
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite">
              <pre wrap="">There is another small advantage: The current position of the parameter makes it easier to add vendor or implementation-specific packet sources for which the direction parameter has no meaning, for example:

+------------------+
| PcapFile         |
+------------------+
| filename         |
+------------------+

              </pre>
              <blockquote type="cite">
                <pre wrap="">======================================================================== 
&gt;From Section 4.3.1
======================================================================== 
Isn't is possible to generate a cache without any keys?
This would generate a cache with a single flow?  Do we allow this?
                </pre>
              </blockquote>
              <pre wrap="">A the moment, there is no restriction which says that a flow key field must exist.

Do you want to add such a restriction?
              </pre>
            </blockquote>
            <pre wrap="">A cache without keys will result in a single flow, which seems a bit daft.  I guess there is no pressing reason to add this restriction though.

            </pre>
            <blockquote type="cite">
              <blockquote type="cite">
                <pre wrap="">Would it be useful to have two lists, one for keys and one for
non-keys?
                </pre>
              </blockquote>
              <pre wrap="">There is an argument for having a single list:

A Cache implementation may generate records using the specified field order. So, the fields occur in the same order as specified in the configuration.
              </pre>
            </blockquote>
            <pre wrap="">If we're worried about implementation specifics can't they extend the YANG model?
            </pre>
          </blockquote>
          <pre wrap="">Yes, they can.

          </pre>
          <blockquote type="cite">
            <blockquote type="cite">
              <pre wrap="">With separate lists for flow key fields and non-key fields, we cannot arbitrarily mix these two types of fields.

Is there any advantage of having separate lists?
              </pre>
            </blockquote>
            <pre wrap="">I was just thinking about how you'd specify that at least one key was required.
            </pre>
          </blockquote>
          <pre wrap="">Caches of type "immediate" do not have Flow Keys. So, the restrictions would not apply to such Caches.
          </pre>
        </blockquote>
        <pre wrap="">Really?  Why not say that Caches of an "immediate" type have no non-keys?  For example, the counter fields should not be key fields, they make no sense as keys.  They also make no sense with the immediate cache.  It seems that an immediate cache is made up of entirely key fields.
        </pre>
      </blockquote>
      <pre wrap="">IMO, a Packet Report does not contain any Flow information, so it does not make sense to talk about Flow Keys.
In -05, I restricted the &lt;isFlowKey&gt; parameter to Caches generating Flow Records because I think that it does not make sense to distinguish Flow Key fields and non-key fields if Packet Reports are generated.
      </pre>
    </blockquote>
    <pre wrap="">Ah I see.  I guess it makes sense to simply not say anything about whether it is a key or not.  So these fields are not keys, but neither are they non-keys.  The are simply collected fields.  Certain fields only make sense for flows and not for packet reports, which might be worth mentioning.


    </pre>
    <blockquote type="cite">
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite">
              <blockquote type="cite">
                <pre wrap="">For a permanent cache, update counters should be used and for a
timeout cache the delta counters should be used.  Where is this
sort of thing enforced?
                </pre>
              </blockquote>
              <pre wrap="">This is not enforced. I would not strictly enforce it because strange configurations might make sense in specific use cases.
              </pre>
            </blockquote>
            <pre wrap="">OK.

            </pre>
            <blockquote type="cite">
              <pre wrap="">However, we could add a note that totalCounters are typically used for "permanent" and deltaCounters for "timeout".

Could you provide a sentence or two?
              </pre>
            </blockquote>
            <pre wrap="">Perhaps a more general note that not all fields will be appropriate for all configurations. Is the Exporting Device free to change the fields
if it needs total counters for a permanent cache?
            </pre>
          </blockquote>
          <pre wrap="">No.

I added a new paragraph at the end of 4.3.1. (CacheLayout Class):

  Note that the use of Information Elements can be restricted to
  certain Cache Modes or to Flow Key or non-key fields.  Such
  restrictions may result from Information Element definitions or from
  device-specific constraints.  According to Section 5, the Monitoring
  Device MUST notify the user if a Cache field cannot be configured
  with the given Information Element.

Ok?
          </pre>
        </blockquote>
        <pre wrap="">Yes, seems reasonable.


        </pre>
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite">
              <blockquote type="cite">
                <pre wrap="">======================================================================== 
&gt;From Section 4.4.1
======================================================================== 
  exportMemberType:  Configures the export member type that corresponds
     to the ipfixTransportSessionGroupMemberType object in
     [I-D.ietf-ipfix-mib].  The following parameter values are
     specified by the configuration data model:
     *  primary: primary target of the Exporting Process
     *  secondary: secondary target of the Exporting Process used when
        the primary target is not reachable
     *  parallel: parallel exporting to all destinations and files of
        the Exporting Process
     *  loadBalancing: load-balancing between different destinations
        and files of the Exporting Process
     "parallel" is the default value if this parameter is not
     configured.  If one destination or file is configured as "primary"

------------------------------------------------------------------------ 
Comments
------------------------------------------------------------------------ 
It's not clear to me that this parameter belongs to the Destination
class.  The number of restrictions on when you can use the various
types suggests that we could simplify things.

A Cache can pass flow records to multiple Exporting Processes, so we
don't really need a parallel option.  This would mean that each
Exporting Process is responsible for sending one complete set of
records.  i.e. one record in, one record out.  It could do that, either
using a redundancy model or with a load-balancing one.

In other words, the exportType would be an Exporting Process attribute
which would indicate whether the list of Destinations should be load-
balanced across or treated as Primary, Seondary, Tertiary, etc.
                </pre>
              </blockquote>
              <pre wrap="">At the moment, the parameter is inline with the IPFIX-MIB:

ipfixExportMemberType OBJECT-TYPE
   SYNTAX      INTEGER {
                   unknown(0),
                   primary(1),
                   secondary(2),
                   parallel(3),
                   loadBalancing(4)
               }
   MAX-ACCESS  read-only
   STATUS      current
   DESCRIPTION
       "The type of a member Transport Session in a Transport
       Session group (identified by the value of ipfixExportIndex,
       ipfixObservationDomainId and ipfixMeteringProcessCacheId).
       The following values are valid:

       unknown(0)
           This value MUST be used if the status of the group
           membership cannot be detected by the equipment. This
           value should be avoided as far as possible.

       primary(1)
           This value is used for a group member that is used as
           the primary target of an Exporter. Other group members
           (with the same ipfixExportIndex and
           ipfixMeteringProcessCacheId) MUST NOT have the value
           primary(1) but MUST have the value secondary(2).
           This value MUST also be specified if the Exporter does
           not support Transport Session grouping. In this case
           the group contains only one Transport Session.

       secondary(2)
           This value is used for a group member that is used as a
           secondary target of an Exporter. The Exporter will use
           one of the targets specified as secondary(2) within the
           same Transport Session group when the primary target is
           not reachable.

       parallel(3)
           This value is used for a group member that is used for
           duplicate exporting i.e., all group members identified&gt; 
              </pre>
            </blockquote>
          </blockquote>
        </blockquote>
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite">
              <pre wrap="">           by the ipfixExportIndex are exporting the same Records
           in parallel. This implies that all group members MUST
           have the same membertype parallel(3).

       loadBalancing(4)
           This value is used for a group member that is used
           as one target for load-balancing. This means that a&gt; 
              </pre>
            </blockquote>
          </blockquote>
        </blockquote>
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite">
              <pre wrap="">           Record is sent to one of the group members in this
           group identified by ipfixExportIndex.
           This implies that all group members MUST have the same
           membertype loadBalancing(4)."
   ::= { ipfixExportEntry 2 }


I'm open to better proposals.

"parallel" and "loadBalancing" can be configured once with a single ExportingProcess class parameter.

How would you configure "primary" and "secondary with a single Exporting Process parameter?
              </pre>
            </blockquote>
          </blockquote>
          <pre wrap="">This issue is on my todo list.
          </pre>
        </blockquote>
        <pre wrap="">OK, it seems from a configuration point of view you have three modes. Parallel, loadBalancing and fallback.  In fallback mode you try the destinations in order until you get a connection.  The mode is a property of the Exporting Process and defines how to treat the list of destinations.

i.e.

  &lt;Exporter&gt;
    &lt;name&gt;Test 1&lt;/name&gt;
    &lt;mode&gt;fallback&lt;/mode&gt;
    &lt;destination&gt; ... &lt;/destination&gt;
    &lt;destination&gt; ... &lt;/destination&gt;
    &lt;destination&gt; ... &lt;/destination&gt;
  &lt;/Exporter&gt;

So parallel mode sends to all three at once, load balancing takes it in turns and fallback treats the first as primary, the second and secondary and the third as tertiary.
        </pre>
      </blockquote>
      <pre wrap="">Yes, this sounds good!
      </pre>
    </blockquote>
    <pre wrap="">Great.

Thanks

Andrew
_______________________________________________
IPFIX mailing list
<a class="moz-txt-link-abbreviated" href="mailto:IPFIX@ietf.org">IPFIX@ietf.org</a>
<a class="moz-txt-link-freetext"
 href="https://www.ietf.org/mailman/listinfo/ipfix">https://www.ietf.org/mailman/listinfo/ipfix</a>
    </pre>
  </blockquote>
  <pre wrap=""><!---->

  </pre>
</blockquote>
</body>
</html>

--------------050809090009080405040303--

From muenz@net.in.tum.de  Mon Mar 15 07:08:49 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 62ABD3A69BD for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 07:08:49 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.135
X-Spam-Level: 
X-Spam-Status: No, score=-2.135 tagged_above=-999 required=5 tests=[AWL=0.114,  BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id TtbRuSODyKbu for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 07:08:48 -0700 (PDT)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id E88C03A6A30 for <ipfix@ietf.org>; Mon, 15 Mar 2010 07:07:39 -0700 (PDT)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id D82BF200D3CA; Mon, 15 Mar 2010 15:07:45 +0100 (CET)
Message-ID: <4B9E3F48.4040004@net.in.tum.de>
Date: Mon, 15 Mar 2010 15:08:08 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Paul Aitken <paitken@cisco.com>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com>	<4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com>	<4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com>	<4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E10EE.9050406@net.in.tum.de> <4B9E3347.1050801@cisco.com>
In-Reply-To: <4B9E3347.1050801@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms060801020803060506010608"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIX MIB
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 15 Mar 2010 14:08:49 -0000

This is a cryptographically signed message in MIME format.

--------------ms060801020803060506010608
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Paul

Paul Aitken wrote:
> Gerhard,
>=20
>> Then, I do not understand what the following value in IPFIX MIB is=20
>> supposed to be:
>>
>>    ipfixMeteringProcessCacheInactiveFlows OBJECT-TYPE
>>        SYNTAX      Gauge32
>>        MAX-ACCESS  read-only
>>        STATUS      current
>>        DESCRIPTION
>>            "The number of Flows currently inactive at this cache."
>>        ::=3D { ipfixMeteringProcessStatsEntry 2 }
>>
>>
>> Well, I'm a coauthor of IPFIX MIB but I have to admit that it is not=20
>> clear to me what inactive flows are supposed to be.
>>
>> Can anybody give an explanation?
>=20
> The definitions of ipfixMeteringProcessCacheActiveFlows and=20
> ipfixMeteringProcessCacheInactiveFlows are poor.
>=20
> However, when a cache is a pre-allocated block of N entries (where each=
=20
> entry may represent a flow), then
>=20
> N =3D (ipfixMeteringProcessCacheActiveFlows +=20
> ipfixMeteringProcessCacheInactiveFlows)
>=20
> where
>=20
> * ipfixMeteringProcessCacheActiveFlows is the number of flows currently=
=20
> in the cache (ie, the number of cache entries currently in use) and
>=20
> * ipfixMeteringProcessCacheInactiveFlows is the number of further cache=
=20
> entries capable of holding new flows.

Ok, thank you for your interpretation.

Let's assume that it is correct.

So, since we do have

    maxRecords:  Maximum number of Data Records in the Cache.  If the
       Cache stores this number of Data Records, no additional record can=

       be created before any of the existing records is removed.  If not
       configured by the user, the Monitoring Device sets this parameter.=


(=3D N) in the configuration data model, we can omit inactiveFlows, right=
?

Cheers,
Gerhard


> With a dynamic cache, ipfixMeteringProcessCacheInactiveFlows may=20
> indicate how much further the cache can grow, or it may be meaningless.=

>=20
>=20
> An alternative (and contradictory) explanation is that=20
> ipfixMeteringProcessCacheInactiveFlows is the number of flows in the=20
> cache which have exceeded the inactive timeout but have not yet been=20
> expired. However, given the transient nature of this statistic, it=20
> doesn't seem to be a very useful one.
>=20
>=20
> Cheers,
> P.
>=20

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms060801020803060506010608
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms060801020803060506010608--

From paitken@cisco.com  Mon Mar 15 07:37:38 2010
Return-Path: <paitken@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id C293F3A6765 for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 07:37:38 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -4.27
X-Spam-Level: 
X-Spam-Status: No, score=-4.27 tagged_above=-999 required=5 tests=[AWL=-1.671,  BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id V0Uf1+rJBtXv for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 07:37:37 -0700 (PDT)
Received: from ams-iport-2.cisco.com (ams-iport-2.cisco.com [144.254.224.141]) by core3.amsl.com (Postfix) with ESMTP id 2BD193A65A6 for <ipfix@ietf.org>; Mon, 15 Mar 2010 07:37:37 -0700 (PDT)
Authentication-Results: ams-iport-2.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AhsBAK7inUuQ/uCWe2dsb2JhbACacRUBAQsLJAYcoGeXaYR7BA
X-IronPort-AV: E=Sophos;i="4.49,643,1262563200";  d="scan'208";a="4352257"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-2.cisco.com with ESMTP; 15 Mar 2010 14:04:09 +0000
Received: from [144.254.153.37] (dhcp-144-254-153-37.cisco.com [144.254.153.37]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2FEbhgH018794; Mon, 15 Mar 2010 14:37:43 GMT
Message-ID: <4B9E4637.40805@cisco.com>
Date: Mon, 15 Mar 2010 14:37:43 +0000
From: Paul Aitken <paitken@cisco.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-GB; rv:1.8.1.22) Gecko/20090605 SeaMonkey/1.1.17 (Ubuntu-1.1.17+nobinonly-0ubuntu0.9.04.1)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com>	<4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com>	<4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com>	<4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E10EE.9050406@net.in.tum.de> <4B9E3347.1050801@cisco.com> <4B9E3F48.4040004@net.in.tum.de>
In-Reply-To: <4B9E3F48.4040004@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIX MIB
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 15 Mar 2010 14:37:38 -0000

Gerhard,

>> The definitions of ipfixMeteringProcessCacheActiveFlows and 
>> ipfixMeteringProcessCacheInactiveFlows are poor.
>>
>> However, when a cache is a pre-allocated block of N entries (where 
>> each entry may represent a flow), then
>>
>> N = (ipfixMeteringProcessCacheActiveFlows + 
>> ipfixMeteringProcessCacheInactiveFlows)
>>
>> where
>>
>> * ipfixMeteringProcessCacheActiveFlows is the number of flows 
>> currently in the cache (ie, the number of cache entries currently in 
>> use) and
>>
>> * ipfixMeteringProcessCacheInactiveFlows is the number of further 
>> cache entries capable of holding new flows.
> 
> Ok, thank you for your interpretation.
> 
> Let's assume that it is correct.

I'd like to think so.


> So, since we do have
> 
>    maxRecords:  Maximum number of Data Records in the Cache.

Yuck. Data Records are produced by the exporting process, while a cache 
is maintained by the Metering Process. Data obtained from the cache may 
be transformed, either in whole or in part, and may be combined with 
other data, to form Data Records.


>       If the
>       Cache stores this number of Data Records, no additional record can
>       be created before any of the existing records is removed.  If not
>       configured by the user, the Monitoring Device sets this parameter.
> 
> (= N) in the configuration data model, we can omit inactiveFlows, right?

Yes. However, in the real world people like to know how full their cache 
is and how much space is left. Providing only "maxRecords" requires them 
to perform an (admittedly simple) calculation, whereas providing 
inactiveFlows makes it immediately obvious.

Is there any reason not to provide all three?

BTW, maxRecords would be the only configurable parameter, if 
configuration were possible.

P.

-- 
Paul Aitken
Cisco Systems Ltd, Edinburgh, Scotland.
http://www.cisco.com/web/about/doing_business/legal/cri/index.html

From muenz@net.in.tum.de  Mon Mar 15 07:52:22 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 34E2E3A6969 for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 07:52:22 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.14
X-Spam-Level: 
X-Spam-Status: No, score=-2.14 tagged_above=-999 required=5 tests=[AWL=0.109,  BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id eBXqBz5fm1LE for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 07:52:15 -0700 (PDT)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id E3F933A68C6 for <ipfix@ietf.org>; Mon, 15 Mar 2010 07:52:14 -0700 (PDT)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id DEEEF200D3CA; Mon, 15 Mar 2010 15:52:20 +0100 (CET)
Message-ID: <4B9E49BC.8010409@net.in.tum.de>
Date: Mon, 15 Mar 2010 15:52:44 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Paul Aitken <paitken@cisco.com>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com>	<4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com>	<4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com>	<4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E10EE.9050406@net.in.tum.de> <4B9E3347.1050801@cisco.com> <4B9E3F48.4040004@net.in.tum.de> <4B9E4637.40805@cisco.com>
In-Reply-To: <4B9E4637.40805@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms080205060305030302000505"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIX MIB
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 15 Mar 2010 14:52:22 -0000

This is a cryptographically signed message in MIME format.

--------------ms080205060305030302000505
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Paul,

Paul Aitken wrote:
> Gerhard,
>=20
>>> The definitions of ipfixMeteringProcessCacheActiveFlows and=20
>>> ipfixMeteringProcessCacheInactiveFlows are poor.
>>>
>>> However, when a cache is a pre-allocated block of N entries (where=20
>>> each entry may represent a flow), then
>>>
>>> N =3D (ipfixMeteringProcessCacheActiveFlows +=20
>>> ipfixMeteringProcessCacheInactiveFlows)
>>>
>>> where
>>>
>>> * ipfixMeteringProcessCacheActiveFlows is the number of flows=20
>>> currently in the cache (ie, the number of cache entries currently in =

>>> use) and
>>>
>>> * ipfixMeteringProcessCacheInactiveFlows is the number of further=20
>>> cache entries capable of holding new flows.
>> Ok, thank you for your interpretation.
>>
>> Let's assume that it is correct.
>=20
> I'd like to think so.
>=20
>=20
>> So, since we do have
>>
>>    maxRecords:  Maximum number of Data Records in the Cache.
>=20
> Yuck. Data Records are produced by the exporting process, while a cache=
=20
> is maintained by the Metering Process. Data obtained from the cache may=
=20
> be transformed, either in whole or in part, and may be combined with=20
> other data, to form Data Records.
>=20
>=20
>>       If the
>>       Cache stores this number of Data Records, no additional record c=
an
>>       be created before any of the existing records is removed.  If no=
t
>>       configured by the user, the Monitoring Device sets this paramete=
r.
>>
>> (=3D N) in the configuration data model, we can omit inactiveFlows, ri=
ght?
>=20
> Yes. However, in the real world people like to know how full their cach=
e=20
> is and how much space is left. Providing only "maxRecords" requires the=
m=20
> to perform an (admittedly simple) calculation, whereas providing=20
> inactiveFlows makes it immediately obvious.
>=20
> Is there any reason not to provide all three?

Ok.

> BTW, maxRecords would be the only configurable parameter, if=20
> configuration were possible.

Sure.

Regards,
Gerhard

>=20
> P.
>=20

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms080205060305030302000505
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature

MIAGCSqGSIb3DQEHAqCAMIACAQExCzAJBgUrDgMCGgUAMIAGCSqGSIb3DQEHAQAAoIIO5jCC
BEYwggOvoAMCAQICEGb9R+PCGeToms2Z3fU6yyQwDQYJKoZIhvcNAQEFBQAwXzELMAkGA1UE
BhMCVVMxFzAVBgNVBAoTDlZlcmlTaWduLCBJbmMuMTcwNQYDVQQLEy5DbGFzcyAxIFB1Ymxp
YyBQcmltYXJ5IENlcnRpZmljYXRpb24gQXV0aG9yaXR5MB4XDTA1MTAyODAwMDAwMFoXDTE1
MTAyNzIzNTk1OVowgd0xCzAJBgNVBAYTAlVTMRcwFQYDVQQKEw5WZXJpU2lnbiwgSW5jLjEf
MB0GA1UECxMWVmVyaVNpZ24gVHJ1c3QgTmV0d29yazE7MDkGA1UECxMyVGVybXMgb2YgdXNl
IGF0IGh0dHBzOi8vd3d3LnZlcmlzaWduLmNvbS9ycGEgKGMpMDUxHjAcBgNVBAsTFVBlcnNv
bmEgTm90IFZhbGlkYXRlZDE3MDUGA1UEAxMuVmVyaVNpZ24gQ2xhc3MgMSBJbmRpdmlkdWFs
IFN1YnNjcmliZXIgQ0EgLSBHMjCCASIwDQYJKoZIhvcNAQEBBQADggEPADCCAQoCggEBAMnf
rOfq+PgDFMQAktXBfjbCPO98chXLwKuMPRyVzm8eECw/AO2XJua2x+atQx0/pIdHR0w+VPhs
+Mf8sZ69MHC8l7EDBeqV8a1AxUR6SwWi8mD81zplYu//EHuiVrvFTnAt1qIfPO2wQuhejVch
rKaZ2RHp0hoHwHRHQgv8xTTq/ea6JNEdCBU3otdzzwFBL2OyOj++pRpu9MlKWz2VphW7NQIZ
+dTvvI8OcXZZu0u2Ptb8Whb01g6J8kn+bAztFenZiHWcec5gJ925rXXOL3OVekA6hXVJsLjf
aLyrzROChRFQo+A8C67AClPN1zBvhTJGG+RJEMJs4q8fef/btLUCAwEAAaOB/zCB/DASBgNV
HRMBAf8ECDAGAQH/AgEAMEQGA1UdIAQ9MDswOQYLYIZIAYb4RQEHFwEwKjAoBggrBgEFBQcC
ARYcaHR0cHM6Ly93d3cudmVyaXNpZ24uY29tL3JwYTALBgNVHQ8EBAMCAQYwEQYJYIZIAYb4
QgEBBAQDAgEGMC4GA1UdEQQnMCWkIzAhMR8wHQYDVQQDExZQcml2YXRlTGFiZWwzLTIwNDgt
MTU1MB0GA1UdDgQWBBQRfV4ZfTwE32ps1qKKGj8x2DuUUjAxBgNVHR8EKjAoMCagJKAihiBo
dHRwOi8vY3JsLnZlcmlzaWduLmNvbS9wY2ExLmNybDANBgkqhkiG9w0BAQUFAAOBgQA8o9oC
YzrEk6qrctPcrVA4HgyeFkqIt+7r2f8PjZWg1rv6aguuYYTYaEeJ70+ssh9JQZtJM3aTi55u
uUMcYL3C3Ioth8FFwBFyBBprJCpsb+f8BxMp0Hc6I+f1wYVoGb/GAVQgGa41gsxiPGEJxvTV
67APpp8zhZrTcY5Qj5ndYjCCBUowggQyoAMCAQICEF0kYdlVNx5E1qcmxun5IrswDQYJKoZI
hvcNAQEFBQAwgd0xCzAJBgNVBAYTAlVTMRcwFQYDVQQKEw5WZXJpU2lnbiwgSW5jLjEfMB0G
A1UECxMWVmVyaVNpZ24gVHJ1c3QgTmV0d29yazE7MDkGA1UECxMyVGVybXMgb2YgdXNlIGF0
IGh0dHBzOi8vd3d3LnZlcmlzaWduLmNvbS9ycGEgKGMpMDUxHjAcBgNVBAsTFVBlcnNvbmEg
Tm90IFZhbGlkYXRlZDE3MDUGA1UEAxMuVmVyaVNpZ24gQ2xhc3MgMSBJbmRpdmlkdWFsIFN1
YnNjcmliZXIgQ0EgLSBHMjAeFw0xMDAxMDIwMDAwMDBaFw0xMTAxMDIyMzU5NTlaMIIBEzEX
MBUGA1UEChMOVmVyaVNpZ24sIEluYy4xHzAdBgNVBAsTFlZlcmlTaWduIFRydXN0IE5ldHdv
cmsxRjBEBgNVBAsTPXd3dy52ZXJpc2lnbi5jb20vcmVwb3NpdG9yeS9SUEEgSW5jb3JwLiBi
eSBSZWYuLExJQUIuTFREKGMpOTgxHjAcBgNVBAsTFVBlcnNvbmEgTm90IFZhbGlkYXRlZDEz
MDEGA1UECxMqRGlnaXRhbCBJRCBDbGFzcyAxIC0gTmV0c2NhcGUgRnVsbCBTZXJ2aWNlMRYw
FAYDVQQDFA1HZXJoYXJkIE11ZW56MSIwIAYJKoZIhvcNAQkBFhNtdWVuekBuZXQuaW4udHVt
LmRlMIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAl/YDW9cnAXJjesGhwtGwj9JN
IHWXQ7QA18s+wGZNsQY0k4eKBeLTk/RAiC1hP+uQAXLoGQHQF/ljjQCQeRmnsN6w04iAzKVo
Ru+NP7Y6ipn8uztx+z0Y3w43CO+LWUudp3PIF3+Fm2zwsAFXj/IYy/8Kq6m9+nQgYO/D4HoN
DDR3//GepXghK0x93gR0GCkm8pgKINHNBD47/IZbDe/cIpAC9HaoxCeLiB1QaCDIYPIrJ0Vx
kk0crAA0tPaMDjdSQTmNRAn2P+7Wg7K+DLvauBHfxLs8PrsolI+XdwiBu6henK9nPxvtoHmm
mfxcdslKLbS9Xz2rHxi6PVVgv1ubWwIDAQABo4HMMIHJMAkGA1UdEwQCMAAwRAYDVR0gBD0w
OzA5BgtghkgBhvhFAQcXATAqMCgGCCsGAQUFBwIBFhxodHRwczovL3d3dy52ZXJpc2lnbi5j
b20vcnBhMAsGA1UdDwQEAwIFoDAdBgNVHSUEFjAUBggrBgEFBQcDBAYIKwYBBQUHAwIwSgYD
VR0fBEMwQTA/oD2gO4Y5aHR0cDovL0luZEMxRGlnaXRhbElELWNybC52ZXJpc2lnbi5jb20v
SW5kQzFEaWdpdGFsSUQuY3JsMA0GCSqGSIb3DQEBBQUAA4IBAQBPU7OeWNo2LRzsdu/RuuAk
L5cj/RzeZ0xv9q4K/eosy9dRu45XACpPvb0fjc3u/dsY4m3jDqzUsB02IrrnF9jpEo10BJPw
b/RZ7NaU7jC/gEknL3bSXiD2PVwL4ZI5ChH5+TJUboWhVh+FpOwOfP/yKk7wQM/46iXXzRIz
enk0rcJp71k4aQgQTML3/8MDeD6TUipuQwQMY9n6T6DsNd/6JpGa+x9ZX4HIEG6+EWwQ85G5
vGfDBaq3dFrksHE75u7b7pf75wm9HIFU2SDHP2tymTHJt/XM9sqP1URIC57s+uZV0/d5KRc1
uPcJ0FAnAiC9o7ybPIh27shiq0FvOEwfMIIFSjCCBDKgAwIBAgIQXSRh2VU3HkTWpybG6fki
uzANBgkqhkiG9w0BAQUFADCB3TELMAkGA1UEBhMCVVMxFzAVBgNVBAoTDlZlcmlTaWduLCBJ
bmMuMR8wHQYDVQQLExZWZXJpU2lnbiBUcnVzdCBOZXR3b3JrMTswOQYDVQQLEzJUZXJtcyBv
ZiB1c2UgYXQgaHR0cHM6Ly93d3cudmVyaXNpZ24uY29tL3JwYSAoYykwNTEeMBwGA1UECxMV
UGVyc29uYSBOb3QgVmFsaWRhdGVkMTcwNQYDVQQDEy5WZXJpU2lnbiBDbGFzcyAxIEluZGl2
aWR1YWwgU3Vic2NyaWJlciBDQSAtIEcyMB4XDTEwMDEwMjAwMDAwMFoXDTExMDEwMjIzNTk1
OVowggETMRcwFQYDVQQKEw5WZXJpU2lnbiwgSW5jLjEfMB0GA1UECxMWVmVyaVNpZ24gVHJ1
c3QgTmV0d29yazFGMEQGA1UECxM9d3d3LnZlcmlzaWduLmNvbS9yZXBvc2l0b3J5L1JQQSBJ
bmNvcnAuIGJ5IFJlZi4sTElBQi5MVEQoYyk5ODEeMBwGA1UECxMVUGVyc29uYSBOb3QgVmFs
aWRhdGVkMTMwMQYDVQQLEypEaWdpdGFsIElEIENsYXNzIDEgLSBOZXRzY2FwZSBGdWxsIFNl
cnZpY2UxFjAUBgNVBAMUDUdlcmhhcmQgTXVlbnoxIjAgBgkqhkiG9w0BCQEWE211ZW56QG5l
dC5pbi50dW0uZGUwggEiMA0GCSqGSIb3DQEBAQUAA4IBDwAwggEKAoIBAQCX9gNb1ycBcmN6
waHC0bCP0k0gdZdDtADXyz7AZk2xBjSTh4oF4tOT9ECILWE/65ABcugZAdAX+WONAJB5Gaew
3rDTiIDMpWhG740/tjqKmfy7O3H7PRjfDjcI74tZS52nc8gXf4WbbPCwAVeP8hjL/wqrqb36
dCBg78Pgeg0MNHf/8Z6leCErTH3eBHQYKSbymAog0c0EPjv8hlsN79wikAL0dqjEJ4uIHVBo
IMhg8isnRXGSTRysADS09owON1JBOY1ECfY/7taDsr4Mu9q4Ed/Euzw+uyiUj5d3CIG7qF6c
r2c/G+2geaaZ/Fx2yUottL1fPasfGLo9VWC/W5tbAgMBAAGjgcwwgckwCQYDVR0TBAIwADBE
BgNVHSAEPTA7MDkGC2CGSAGG+EUBBxcBMCowKAYIKwYBBQUHAgEWHGh0dHBzOi8vd3d3LnZl
cmlzaWduLmNvbS9ycGEwCwYDVR0PBAQDAgWgMB0GA1UdJQQWMBQGCCsGAQUFBwMEBggrBgEF
BQcDAjBKBgNVHR8EQzBBMD+gPaA7hjlodHRwOi8vSW5kQzFEaWdpdGFsSUQtY3JsLnZlcmlz
aWduLmNvbS9JbmRDMURpZ2l0YWxJRC5jcmwwDQYJKoZIhvcNAQEFBQADggEBAE9Ts55Y2jYt
HOx279G64CQvlyP9HN5nTG/2rgr96izL11G7jlcAKk+9vR+Nze792xjibeMOrNSwHTYiuucX
2OkSjXQEk/Bv9Fns1pTuML+ASScvdtJeIPY9XAvhkjkKEfn5MlRuhaFWH4Wk7A58//IqTvBA
z/jqJdfNEjN6eTStwmnvWThpCBBMwvf/wwN4PpNSKm5DBAxj2fpPoOw13/omkZr7H1lfgcgQ
br4RbBDzkbm8Z8MFqrd0WuSwcTvm7tvul/vnCb0cgVTZIMc/a3KZMcm39cz2yo/VREgLnuz6
5lXT93kpFzW49wnQUCcCIL2jvJs8iHbuyGKrQW84TB8xggTsMIIE6AIBATCB8jCB3TELMAkG
A1UEBhMCVVMxFzAVBgNVBAoTDlZlcmlTaWduLCBJbmMuMR8wHQYDVQQLExZWZXJpU2lnbiBU
cnVzdCBOZXR3b3JrMTswOQYDVQQLEzJUZXJtcyBvZiB1c2UgYXQgaHR0cHM6Ly93d3cudmVy
aXNpZ24uY29tL3JwYSAoYykwNTEeMBwGA1UECxMVUGVyc29uYSBOb3QgVmFsaWRhdGVkMTcw
NQYDVQQDEy5WZXJpU2lnbiBDbGFzcyAxIEluZGl2aWR1YWwgU3Vic2NyaWJlciBDQSAtIEcy
AhBdJGHZVTceRNanJsbp+SK7MAkGBSsOAwIaBQCgggLOMBgGCSqGSIb3DQEJAzELBgkqhkiG
9w0BBwEwHAYJKoZIhvcNAQkFMQ8XDTEwMDMxNTE0NTI0NFowIwYJKoZIhvcNAQkEMRYEFC0Q
8j9XQOyB8TGdzCutaRFnp85fMF8GCSqGSIb3DQEJDzFSMFAwCwYJYIZIAWUDBAECMAoGCCqG
SIb3DQMHMA4GCCqGSIb3DQMCAgIAgDANBggqhkiG9w0DAgIBQDAHBgUrDgMCBzANBggqhkiG
9w0DAgIBKDCCAQMGCSsGAQQBgjcQBDGB9TCB8jCB3TELMAkGA1UEBhMCVVMxFzAVBgNVBAoT
DlZlcmlTaWduLCBJbmMuMR8wHQYDVQQLExZWZXJpU2lnbiBUcnVzdCBOZXR3b3JrMTswOQYD
VQQLEzJUZXJtcyBvZiB1c2UgYXQgaHR0cHM6Ly93d3cudmVyaXNpZ24uY29tL3JwYSAoYykw
NTEeMBwGA1UECxMVUGVyc29uYSBOb3QgVmFsaWRhdGVkMTcwNQYDVQQDEy5WZXJpU2lnbiBD
bGFzcyAxIEluZGl2aWR1YWwgU3Vic2NyaWJlciBDQSAtIEcyAhBdJGHZVTceRNanJsbp+SK7
MIIBBQYLKoZIhvcNAQkQAgsxgfWggfIwgd0xCzAJBgNVBAYTAlVTMRcwFQYDVQQKEw5WZXJp
U2lnbiwgSW5jLjEfMB0GA1UECxMWVmVyaVNpZ24gVHJ1c3QgTmV0d29yazE7MDkGA1UECxMy
VGVybXMgb2YgdXNlIGF0IGh0dHBzOi8vd3d3LnZlcmlzaWduLmNvbS9ycGEgKGMpMDUxHjAc
BgNVBAsTFVBlcnNvbmEgTm90IFZhbGlkYXRlZDE3MDUGA1UEAxMuVmVyaVNpZ24gQ2xhc3Mg
MSBJbmRpdmlkdWFsIFN1YnNjcmliZXIgQ0EgLSBHMgIQXSRh2VU3HkTWpybG6fkiuzANBgkq
hkiG9w0BAQEFAASCAQBoYY5P64Lv8DoFdm6hHzCk58WV+F0Ng8ei0jjWT+BM0AdiS6jzj9tx
m2G6Ky5Db3y+WR41gJv0OOxkjucxtE8Un+Vy+jbyCJjLcMqRYczGGtjrzCmRFUJGUG5nUy/S
lCvhFsG+/q8RiwbthLEpRp/HFhVliqBMP9pGTrOlSIMJTSLtr3c4kFdAMDfvPjzIzPh2k6fV
sOpuBoDjEDK3twUT3H+zJGVCrpSkr0mSkeiB73H0Z/I9Dnoxk1q8vSDsiJIavjBRLvYVnDWZ
xmxTiq607iPP7Bh/IiOJK8iuFZd4GNBexwaUzCKMiD6ffP3yPz+I5wIam1pH2fPhRSncSOPe
AAAAAAAA
--------------ms080205060305030302000505--

From n.brownlee@auckland.ac.nz  Mon Mar 15 15:32:05 2010
Return-Path: <n.brownlee@auckland.ac.nz>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id E115D3A6A7F for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 15:32:05 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.092
X-Spam-Level: 
X-Spam-Status: No, score=-5.092 tagged_above=-999 required=5 tests=[AWL=1.093,  BAYES_40=-0.185, GB_I_INVITATION=-2, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 50PELS9jeIfb for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 15:32:05 -0700 (PDT)
Received: from mailhost.auckland.ac.nz (larry.its.auckland.ac.nz [130.216.12.34]) by core3.amsl.com (Postfix) with ESMTP id 17A863A681F for <ipfix@ietf.org>; Mon, 15 Mar 2010 15:32:04 -0700 (PDT)
Received: from localhost (localhost.localdomain [127.0.0.1]) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 3A4F24149FC for <ipfix@ietf.org>; Tue, 16 Mar 2010 11:32:11 +1300 (NZDT)
X-Virus-Scanned: by amavisd-new at mailhost.auckland.ac.nz
Received: from mailhost.auckland.ac.nz ([127.0.0.1]) by localhost (larry.its.auckland.ac.nz [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id rBjHgYbKvEyf for <ipfix@ietf.org>; Tue, 16 Mar 2010 11:32:11 +1300 (NZDT)
Received: from [10.255.240.236] (unknown [192.5.67.11]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mailhost.auckland.ac.nz (Postfix) with ESMTP id BD7554149E2 for <ipfix@ietf.org>; Tue, 16 Mar 2010 11:32:10 +1300 (NZDT)
Message-ID: <4B9EB563.5090600@auckland.ac.nz>
Date: Mon, 15 Mar 2010 15:32:03 -0700
From: Nevil Brownlee <n.brownlee@auckland.ac.nz>
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
MIME-Version: 1.0
To: IPFIX list <ipfix@ietf.org>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Subject: [IPFIX] netconf/yang discussion at Anaheim
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 15 Mar 2010 22:32:06 -0000

Forwarded on behalf of Carl Moberg  <calle@tail-f.com>  ...

Cheers, Nevil


This is an open invitation for sharing and discussing implementation
and operational experiences of NETCONF and YANG on Sunday, March 21
2010 from 12:00noon to 3:00pm in the Carmel room at the Anaheim Hilton.

We are particularily looking for input to the following open questions:
- Any experience from implementing NETCONF and YANG?
- What can be expected from NETCONF and YANG implementations?
- Any operational pros and cons of NETCONF and YANG?
- What can be done to make NETCONF and YANG more useful in
    operator environments?

Lunch (pizza and drinks) will be available.

-- 
---------------------------------------------------------------------
  Nevil Brownlee                    Computer Science Department | ITS
  Phone: +64 9 373 7599 x88941             The University of Auckland
  FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand


From trammell@tik.ee.ethz.ch  Mon Mar 15 15:59:02 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id E6EBA3A67F9 for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 15:59:02 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.555
X-Spam-Level: 
X-Spam-Status: No, score=-6.555 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, DATE_IN_PAST_03_06=0.044, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id y8XBm5sW25MY for <ipfix@core3.amsl.com>; Mon, 15 Mar 2010 15:59:01 -0700 (PDT)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id 95E533A680E for <ipfix@ietf.org>; Mon, 15 Mar 2010 15:58:54 -0700 (PDT)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id 65BF0D93A5; Mon, 15 Mar 2010 23:59:00 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id v4sM0RnlR6g3; Mon, 15 Mar 2010 23:59:00 +0100 (MET)
Received: from [10.0.1.5] (unknown [75.85.164.240]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id BDD8CD938F; Mon, 15 Mar 2010 23:58:58 +0100 (MET)
Mime-Version: 1.0 (Apple Message framework v1077)
Content-Type: text/plain; charset=us-ascii
From: Brian Trammell <trammell@tik.ee.ethz.ch>
In-Reply-To: <4B9E33BE.7070907@cisco.com>
Date: Mon, 15 Mar 2010 11:01:28 -0700
Content-Transfer-Encoding: quoted-printable
Message-Id: <7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch>
References: <4AF73525.8050009@net.in.tum.de> <4AF8F999.3000207@cisco.com> <F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch> <4B9E33BE.7070907@cisco.com>
To: Benoit Claise <bclaise@cisco.com>
X-Mailer: Apple Mail (2.1077)
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 15 Mar 2010 22:59:03 -0000

hi, Benoit, all...

Agreed that we should do something about this (i.e., that solution 1 is =
no solution.); that said, a few comments in no particular order:

1. In considering adding explicit semantics to structured data, we as a =
WG are taking on the task of defining semantics for IPFIX as a whole. =
Semantics, as I understand them, are in IPFIX largely contextual and =
template dependent, but in almost all cases I can think of it seems like =
these are implicitly "AND" semnantics (this flow has source IP A _AND_ =
destination IP B _AND_...). We will need to make an explicit statement =
on this. We will need to determine whether these implicit semantics are =
a property of the protocol (in which case Structured Data is really a =
protocol-level extension), or a property of each information element (in =
which case all 5103 IEs have implicit AND semantics; question 2: will we =
need to add semantics to the IANA registry in this case?). We will need =
to be quite careful about this. It's not as simple as defining semantics =
within structured data then calling it done.

2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid argument =
against Solution 2, as it's trying to solve a somewhat different and =
more limited problem than structured data. Solution 2 _might_ cause a =
problem for this draft, but certainly not the other way around (unless =
we as a WG want to subsume that work into this draft, which would =
probably require rechartering...). Also, I don't think we need semantics =
for all of the list types, just basicList (Illustrative question: How do =
I meaningfully interpret two "or" subTemplateMultiLists with disjoint IE =
sets? Two "and" subTemplateMultiLists? Nestings thereof?)... So, we =
don't really have an explosion to deal with if we do Solution 2 =
correctly: andBasicList, orBasicList, xorBasicList, notBasicList; Four =
IEs, nestable, done. What can't we do with those four IEs? In this case, =
we could even step back and say that semantics outside these four are =
within the protocol explicitly _undefined_, and to be interpreted within =
the context of each Template.

3. If we really _do_ want ranges and so on (which, again, we'd need to =
get WG consensus on; this is explicitly out of scope in my reading of =
the present charter), then we could do them in the scope of Solution 3. =
However, this seems a little not-quite-fleshed-out-enough for me to say =
whether I like it or not. Could you present an example of how you would =
use the proposed Semantic field to model your example? "(eth1 OR eth2) =
AND (NOT (eth3 OR eth4)) OR linecard2"

(FWIW, I would do this with my proposal to Solution 2 as follows:)

(orBasicList (andBasicList (orBasicList ingressInterface eth1 eth2) =
(notBasicList (orBasicList ingressInterface eth3 eth4)) (andBasicList =
lineCardID 2))

Best regards,

Brian

On Mar 15, 2010, at 6:18 AM, Benoit Claise wrote:

> Dear all,
>=20
> We've been thinking about this one, and we see 3 solutions. We believe =
that the solution 3 is the way to go, as we show below.
>=20
> Solution 1.=20
> Consider that the semantic is out of scope for this document.=20
> This is the easiest solution.  However, we understand it's not right: =
we would not do a complete job wrt to IPFIX structured data
> As Gerhard was expressing, the collector has not clue how to treat the =
example of a BasicList of egress interfaces in a Flow Record.=20
>     - Has every counted packet been sent on every egress interface?
>       =3D> multicast case, AND semantic
>     - Has every counted packet been sent on any one of the egress =
interfaces?
>      =3D> load balancing case, OR semantic
>=20
> Soltution 2
> We could focus on the logical AND and OR semantics only , by defining =
semantic lists, such as andBasicList, andSubTemplateList, =
andSubTemplateMultiList, orBasicList, orSubTemplateList, and =
orSubTemplateMultiList
> So 6 I.E.s in total, describing AND and OR semantics.
> While it's solved the router and most mediation function needs today, =
we understand that this is not a complete solution.
> Gerhard, in one of his old draft, draft-sommer-ipfix-mediator-ext-01, =
proposed ADTs for orderedList, orderedPair, and portRanges which allow =
the definition new IEs for port ranges etc.
> Even if this solution 2 is extensible, in the long term, will lead to =
an explosion of IEs: 3 list types * the semantic type, where the =
semantic can be (and, or, orderedList, orderedPair, portRanges, random, =
etc...)
>=20
> Soltution 3
> We propose to add a semantic field in the 3 list types.
> Something such as:
>=20
>     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> |0|               Field ID    |       Element Length            |
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> | Semantic  |             BasicList Content ...                 |
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> |                           ...                                 |
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>=20
> This semantic field would be a new IANA registry, we could be =
populated initially with NONE, OR, AND, ORDERED, etc... Up to =
discussion.
> The advantage of this solution is that it's extensible, and it doesn't =
need new IEs for each semantic.
> If ever required (mainly in a mediation function), we can model  very =
complex semantic, eg, "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR =
linecard2", which could be the new observation point from a aggregated =
Flow Record
>=20
>=20
>=20
> Conclusion:
> After much debate internally, we believe that we should do the effort =
to include this semantic now, and not postpone the problem, which will =
lead to an explosion of IEs in the future.
> Personally, I was initially against the solution 3 ... mostly due to =
the effort required to modify the complete specs. =20
> We're now ready to modify the specifications in the IPFIX structured =
data, but we would like to get your feedback and agreement in advance as =
this is not a small piece of work.
>=20
> Please comment.
>=20
> Regards, Paul, Stan, Gowri, and Benoit.
>> hi Benoit, Gerhard,=20
>>=20
>> In this case I'm strongly in favor of leaving semantics out of =
structured data. Structured data defines containers. The semantics of =
the containers as a whole and the elements change based upon the =
information elements within the container and within the record =
containing it. Wedging semantics into the structured data elements 1. =
risks further proliferation of (potentially non-interoperable) ways to =
represent the same thing, 2. gives us more ways to represent nonsensical =
things (an OR basicList of MPLS stack entries...means...what?), and 3. =
risks defining an inadequate semantic representation mechanism (what =
about semantics for records not using structured data? what about =
ordered versus unordered sets? what about OR basicList vs AND basicList =
vs nested AND and OR basicLists vs just three identical IEs?). If we =
really want an unambiguous semantic framework for IPFIX (and here I'm =
not convinced either way) that's best done on its own, addressing things =
at the information element and record level. Doing it here confuses the =
issue.=20
>>=20
>> Regards,=20
>>=20
>> Brian=20
>>=20
>> On Nov 10, 2009, at 6:26 AM, Benoit Claise wrote:=20
>>=20
>>> Gerhard,=20
>>>=20
>>> Thanks for your email.=20
>>> I have no strong feelings about the two solutions you proposed.=20
>>>=20
>>> >=46rom a pure router point of view,  I don't see any use cases for =
logical OR in exporting flow records.=20
>>> However, from a IPFIX Mediator point of view, I see some use cases.=20=

>>> I mean that it requires an Intermediate Aggregation Process or =
Intermediate Correlation Process to express: I've seen this flow record =
OR that flow record.=20
>>> Now, it's true that even routers will have mediation functions...=20
>>>=20
>>> I'm inclined to add orBasicList, orSubTemplateList, =
orSubTempalteMultiList to the draft (This is a small addition after all) =
and to express that, by default, a logical AND is assumed... =
specifically if the structured data is used in the IPFIX Mediation =
Protocol.=20
>>>=20
>>> I'm not convinced by the NOT in the context of structured data, as =
we don't even have a concept of NOT for a single information element!=20
>>>=20
>>> I would like to get some more feedback from others.=20
>>>=20
>>> Regards, Benoit.=20
>>>=20
>>>> Dear all,=20
>>>>=20
>>>> Regarding draft-ietf-ipfix-structured-data, I see the risk that the=20=

>>>> semantic of the exported structured data is not clear.=20
>>>>=20
>>>> How do you interpret the manifold occurrence of the same =
Information=20
>>>> Element (basicList) or the same group of Information Elements=20
>>>> (subTemplateList) in one record?=20
>>>>=20
>>>> What does it mean if basicList, subTemplateList, or =
subTemplateMultiList=20
>>>> is used for a Flow Key field? Or non-key field?=20
>>>>=20
>>>> Some Examples:=20
>>>>=20
>>>> - BasicList of egress interfaces in a Flow Record=20
>>>>  How should a Flow Record be interpreted which contains a list of=20=

>>>>  egress interfaces and a packet counter?=20
>>>>  Has every counted packet been sent on every egress interface?=20
>>>>    =3D> multicast case, AND semantic (see example in section 8.1)=20=

>>>>  Has every counted packet been sent on any one of the egress=20
>>>>  interfaces?=20
>>>>    =3D> load balancing case, OR semantic=20
>>>>  Can it be used as a Flow Key or not?=20
>>>>=20
>>>> - BasicList of destination ports in a Flow Record=20
>>>>  As every packet has only one destination port, the only reasonable=20=

>>>>  interpretation is that the Flow contains packets having one of=20
>>>>  the reported port numbers.=20
>>>>    =3D> OR semantic=20
>>>>  This would be a non-key field.=20
>>>>=20
>>>>=20
>>>> I think there are two solutions:=20
>>>>=20
>>>> 1. We decide that the semantic of list content is out of scope of=20=

>>>>   draft-ietf-ipfix-structured-data. We add a note to the draft that=20=

>>>>   the semantic must be clear from the context or the definition of =
the=20
>>>>   Information Elements used within the lists.=20
>>>>=20
>>>> 2. We define semantic lists, such as=20
>>>>   - andBasicList, andSubTemplateList, andSubTemplateMultiList=20
>>>>   - orBasicList, orSubTemplateList, orSubTempalteMultiList=20
>>>>   describing AND and OR semantic of the contained IEs/Templates,=20
>>>>   respectively.=20
>>>>=20
>>>>=20
>>>> As I wrote in an earlier mail, I see a good use case for =
orBasicList. It=20
>>>> could be used in the Selector Report Interpretation of Property =
Match=20
>>>> Filtering to report a filter like "port 80 or port 443".=20
>>>> http://www.ietf.org/mail-archive/web/ipfix/current/msg04856.html=20
>>>> At the moment, the Selector Report Interpretation is limited to =
AND.=20
>>>> However, if we also want to express a NOT, we still need another =
solution...=20
>>>>=20
>>>> Regards,=20
>>>> Gerhard=20
>>>>=20
>>>> _______________________________________________=20
>>>> IPFIX mailing list=20
>>>> IPFIX@ietf.org=20
>>>> https://www.ietf.org/mailman/listinfo/ipfix=20
>>>>=20
>>>=20
>>> _______________________________________________=20
>>> IPFIX mailing list=20
>>> IPFIX@ietf.org=20
>>> https://www.ietf.org/mailman/listinfo/ipfix=20
>=20


From bclaise@cisco.com  Tue Mar 16 02:40:38 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 1A9E63A6A2A for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 02:40:38 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.468
X-Spam-Level: 
X-Spam-Status: No, score=-2.468 tagged_above=-999 required=5 tests=[AWL=0.130,  BAYES_00=-2.599, HTML_MESSAGE=0.001]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id G-Kz6YBHrT6m for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 02:40:35 -0700 (PDT)
Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id 9BBDB3A6A28 for <ipfix@ietf.org>; Tue, 16 Mar 2010 02:40:26 -0700 (PDT)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2G9eX5R004468; Tue, 16 Mar 2010 10:40:33 +0100 (CET)
Received: from [10.55.43.57] (ams-bclaise-8718.cisco.com [10.55.43.57]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2G9eUXo000265; Tue, 16 Mar 2010 10:40:31 +0100 (CET)
Message-ID: <4B9F520E.8060507@cisco.com>
Date: Tue, 16 Mar 2010 10:40:30 +0100
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: Brian Trammell <trammell@tik.ee.ethz.ch>
References: <4AF73525.8050009@net.in.tum.de> <4AF8F999.3000207@cisco.com>	<F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>	<4B9E33BE.7070907@cisco.com> <7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch>
In-Reply-To: <7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch>
Content-Type: multipart/alternative; boundary="------------040705010903090204000402"
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 16 Mar 2010 09:40:38 -0000

This is a multi-part message in MIME format.
--------------040705010903090204000402
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable

Hi Brian,

Thanks for your feedback.
> hi, Benoit, all...
>
> Agreed that we should do something about this (i.e., that solution 1 is=
 no solution.); that said, a few comments in no particular order:
>
> 1. In considering adding explicit semantics to structured data, we as a=
 WG are taking on the task of defining semantics for IPFIX as a whole. Se=
mantics, as I understand them, are in IPFIX largely contextual and templa=
te dependent, but in almost all cases I can think of it seems like these =
are implicitly "AND" semnantics (this flow has source IP A _AND_ destinat=
ion IP B _AND_...).
Agreed.
> We will need to make an explicit statement on this.
Not sure why.
> We will need to determine whether these implicit semantics are a proper=
ty of the protocol (in which case Structured Data is really a protocol-le=
vel extension), or a property of each information element (in which case =
all 5103 IEs have implicit AND semantics; question 2: will we need to add=
 semantics to the IANA registry in this case?). We will need to be quite =
careful about this. It's not as simple as defining semantics within struc=
tured data then calling it done.
>   =20
I'm not sure why each individual IE should have a semantic ... in the=20
IANA registry.
As you wrote, "in almost all cases I can think of it seems like these=20
are implicitly "AND" semnantics", so the case we try to solve is when=20
there are multiple instances of a single IE. We know that RFC 5101=20
foresaw that case " The Collector MUST support the use of Templates=20
containing multiple occurrences of the similar Information Elements",=20
but the idea is not to change RFC5101.  If we put some semantic in the=20
IPFIX structured data, that would solve the vast majority of our cases.=20
Also, we could say: if you want some semantic when exporting multiple=20
IEs, then you SHOULD use IPFIX structured data.

Also, the default value for the semantic field in the IPFIX structured=20
data SHOULD be "NONE", to express that flow record doesn't include any=20
semantic.... like in RFC5101. You might draw your own conclusion, maybe=20
because you know your network, maybe because you have configured the=20
exporter, but then it's your decision.
The way I see the proposed solution is: in IPFIX structured data, you=20
MAY use the semantic field as a way to express the relationship between=20
IEs within the structure.
> 2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid argument=
 against Solution 2, as it's trying to solve a somewhat different and mor=
e limited problem than structured data. Solution 2 _might_ cause a proble=
m for this draft, but certainly not the other way around (unless we as a =
WG want to subsume that work into this draft, which would probably requir=
e rechartering...). Also, I don't think we need semantics for all of the =
list types, just basicList (Illustrative question: How do I meaningfully =
interpret two "or" subTemplateMultiLists with disjoint IE sets? Two "and"=
 subTemplateMultiLists? Nestings thereof?)... So, we don't really have an=
 explosion to deal with if we do Solution 2 correctly: andBasicList, orBa=
sicList, xorBasicList, notBasicList; Four IEs, nestable, done. What can't=
 we do with those four IEs? In this case, we could even step back and say=
 that semantics outside these four are within the protocol explicitly _un=
defined_, and to be interpreted within the context of each Template.
>   =20
If I translate some more what I wrote "While it's solved the router and=20
most mediation function needs today", this would be "Me, myself, and I=20
don't need more that OR and AND _now _;-)".
However, who am I to tell that others don't need it now... and that the=20
logical solution is to use the IPFIX structured data
Furthermore, if we think a little bit longer term, the next big step in=20
IPFIX is the mediation function. In my company, every features want to=20
export his own data with NetFlow/IPFIX... up to the point where a CPE=20
would not have enough bandwidth across the WAN to export all the=20
"management" information. So we'll need more and more of aggregated flow =

records (both in time and space) even in the router. Again, the logical=20
solution will be to use the IPFIX structured data. At this point, we=20
will most probably need something else than OR and AND, i.e. RANGE,=20
ORDERRED, etc...

An example of subTemplateMultiLists with disjoint IE sets, let's imagine =

that you have to export an aggregated observation point, composed of=20
multiple template records
     template 1: exporterIPaddress
     template 2: exporterIPaddress, basicList of interfaces
     template 3: exporterIPaddress, LC


> 3. If we really _do_ want ranges and so on (which, again, we'd need to =
get WG consensus on; this is explicitly out of scope in my reading of the=
 present charter), then we could do them in the scope of Solution 3.
Btw, acknowledging that one day we will have to solve this is good=20
enough for solution 3. I mean, we don't have to populate the semantic=20
IANA now... even if that would be more efficient.
> However, this seems a little not-quite-fleshed-out-enough for me to say=
 whether I like it or not. Could you present an example of how you would =
use the proposed Semantic field to model your example? "(eth1 OR eth2) AN=
D (NOT (eth3 OR eth4)) OR linecard2"
>
> (FWIW, I would do this with my proposal to Solution 2 as follows:)
>
> (orBasicList (andBasicList (orBasicList ingressInterface eth1 eth2) (no=
tBasicList (orBasicList ingressInterface eth3 eth4)) (andBasicList lineCa=
rdID 2))
>   =20
(BasicList, OR, (basicList, AND, (basicList, OR, eth1, eth2), (basicList,=
 NOR, eth3, eth4)), (basicList, NONE, linecard2)))

Thanks for your valuable feedback, as always.

Regards, Benoit.

> Best regards,
>
> Brian
>
> On Mar 15, 2010, at 6:18 AM, Benoit Claise wrote:
>
>   =20
>> Dear all,
>>
>> We've been thinking about this one, and we see 3 solutions. We believe=
 that the solution 3 is the way to go, as we show below.
>>
>> Solution 1.
>> Consider that the semantic is out of scope for this document.
>> This is the easiest solution.  However, we understand it's not right: =
we would not do a complete job wrt to IPFIX structured data
>> As Gerhard was expressing, the collector has not clue how to treat the=
 example of a BasicList of egress interfaces in a Flow Record.
>>      - Has every counted packet been sent on every egress interface?
>>        =3D>  multicast case, AND semantic
>>      - Has every counted packet been sent on any one of the egress int=
erfaces?
>>       =3D>  load balancing case, OR semantic
>>
>> Soltution 2
>> We could focus on the logical AND and OR semantics only , by defining =
semantic lists, such as andBasicList, andSubTemplateList, andSubTemplateM=
ultiList, orBasicList, orSubTemplateList, and orSubTemplateMultiList
>> So 6 I.E.s in total, describing AND and OR semantics.
>> While it's solved the router and most mediation function needs today, =
we understand that this is not a complete solution.
>> Gerhard, in one of his old draft, draft-sommer-ipfix-mediator-ext-01, =
proposed ADTs for orderedList, orderedPair, and portRanges which allow th=
e definition new IEs for port ranges etc.
>> Even if this solution 2 is extensible, in the long term, will lead to =
an explosion of IEs: 3 list types * the semantic type, where the semantic=
 can be (and, or, orderedList, orderedPair, portRanges, random, etc...)
>>
>> Soltution 3
>> We propose to add a semantic field in the 3 list types.
>> Something such as:
>>
>>      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+=

>> |0|               Field ID    |       Element Length            |
>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>> | Semantic  |             BasicList Content ...                 |
>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>> |                           ...                                 |
>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>
>> This semantic field would be a new IANA registry, we could be populate=
d initially with NONE, OR, AND, ORDERED, etc... Up to discussion.
>> The advantage of this solution is that it's extensible, and it doesn't=
 need new IEs for each semantic.
>> If ever required (mainly in a mediation function), we can model  very =
complex semantic, eg, "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR linecar=
d2", which could be the new observation point from a aggregated Flow Reco=
rd
>>
>>
>>
>> Conclusion:
>> After much debate internally, we believe that we should do the effort =
to include this semantic now, and not postpone the problem, which will le=
ad to an explosion of IEs in the future.
>> Personally, I was initially against the solution 3 ... mostly due to t=
he effort required to modify the complete specs.
>> We're now ready to modify the specifications in the IPFIX structured d=
ata, but we would like to get your feedback and agreement in advance as t=
his is not a small piece of work.
>>
>> Please comment.
>>
>> Regards, Paul, Stan, Gowri, and Benoit.
>>     =20
>>> hi Benoit, Gerhard,
>>>
>>> In this case I'm strongly in favor of leaving semantics out of struct=
ured data. Structured data defines containers. The semantics of the conta=
iners as a whole and the elements change based upon the information eleme=
nts within the container and within the record containing it. Wedging sem=
antics into the structured data elements 1. risks further proliferation o=
f (potentially non-interoperable) ways to represent the same thing, 2. gi=
ves us more ways to represent nonsensical things (an OR basicList of MPLS=
 stack entries...means...what?), and 3. risks defining an inadequate sema=
ntic representation mechanism (what about semantics for records not using=
 structured data? what about ordered versus unordered sets? what about OR=
 basicList vs AND basicList vs nested AND and OR basicLists vs just three=
 identical IEs?). If we really want an unambiguous semantic framework for=
 IPFIX (and here I'm not convinced either way) that's best done on its ow=
n, addressing things at the information element and record level. Doing i=
t here confuses the issue.
>>>
>>> Regards,
>>>
>>> Brian
>>>
>>> On Nov 10, 2009, at 6:26 AM, Benoit Claise wrote:
>>>
>>>       =20
>>>> Gerhard,
>>>>
>>>> Thanks for your email.
>>>> I have no strong feelings about the two solutions you proposed.
>>>>
>>>> > From a pure router point of view,  I don't see any use cases for l=
ogical OR in exporting flow records.
>>>> However, from a IPFIX Mediator point of view, I see some use cases.
>>>> I mean that it requires an Intermediate Aggregation Process or Inter=
mediate Correlation Process to express: I've seen this flow record OR tha=
t flow record.
>>>> Now, it's true that even routers will have mediation functions...
>>>>
>>>> I'm inclined to add orBasicList, orSubTemplateList, orSubTempalteMul=
tiList to the draft (This is a small addition after all) and to express t=
hat, by default, a logical AND is assumed... specifically if the structur=
ed data is used in the IPFIX Mediation Protocol.
>>>>
>>>> I'm not convinced by the NOT in the context of structured data, as w=
e don't even have a concept of NOT for a single information element!
>>>>
>>>> I would like to get some more feedback from others.
>>>>
>>>> Regards, Benoit.
>>>>
>>>>         =20
>>>>> Dear all,
>>>>>
>>>>> Regarding draft-ietf-ipfix-structured-data, I see the risk that the=

>>>>> semantic of the exported structured data is not clear.
>>>>>
>>>>> How do you interpret the manifold occurrence of the same Informatio=
n
>>>>> Element (basicList) or the same group of Information Elements
>>>>> (subTemplateList) in one record?
>>>>>
>>>>> What does it mean if basicList, subTemplateList, or subTemplateMult=
iList
>>>>> is used for a Flow Key field? Or non-key field?
>>>>>
>>>>> Some Examples:
>>>>>
>>>>> - BasicList of egress interfaces in a Flow Record
>>>>>   How should a Flow Record be interpreted which contains a list of
>>>>>   egress interfaces and a packet counter?
>>>>>   Has every counted packet been sent on every egress interface?
>>>>>     =3D>  multicast case, AND semantic (see example in section 8.1)=

>>>>>   Has every counted packet been sent on any one of the egress
>>>>>   interfaces?
>>>>>     =3D>  load balancing case, OR semantic
>>>>>   Can it be used as a Flow Key or not?
>>>>>
>>>>> - BasicList of destination ports in a Flow Record
>>>>>   As every packet has only one destination port, the only reasonabl=
e
>>>>>   interpretation is that the Flow contains packets having one of
>>>>>   the reported port numbers.
>>>>>     =3D>  OR semantic
>>>>>   This would be a non-key field.
>>>>>
>>>>>
>>>>> I think there are two solutions:
>>>>>
>>>>> 1. We decide that the semantic of list content is out of scope of
>>>>>    draft-ietf-ipfix-structured-data. We add a note to the draft tha=
t
>>>>>    the semantic must be clear from the context or the definition of=
 the
>>>>>    Information Elements used within the lists.
>>>>>
>>>>> 2. We define semantic lists, such as
>>>>>    - andBasicList, andSubTemplateList, andSubTemplateMultiList
>>>>>    - orBasicList, orSubTemplateList, orSubTempalteMultiList
>>>>>    describing AND and OR semantic of the contained IEs/Templates,
>>>>>    respectively.
>>>>>
>>>>>
>>>>> As I wrote in an earlier mail, I see a good use case for orBasicLis=
t. It
>>>>> could be used in the Selector Report Interpretation of Property Mat=
ch
>>>>> Filtering to report a filter like "port 80 or port 443".
>>>>> http://www.ietf.org/mail-archive/web/ipfix/current/msg04856.html
>>>>> At the moment, the Selector Report Interpretation is limited to AND=
=2E
>>>>> However, if we also want to express a NOT, we still need another so=
lution...
>>>>>
>>>>> Regards,
>>>>> Gerhard
>>>>>
>>>>> _______________________________________________
>>>>> IPFIX mailing list
>>>>> IPFIX@ietf.org
>>>>> https://www.ietf.org/mailman/listinfo/ipfix
>>>>>
>>>>>           =20
>>>> _______________________________________________
>>>> IPFIX mailing list
>>>> IPFIX@ietf.org
>>>> https://www.ietf.org/mailman/listinfo/ipfix
>>>>         =20
>>     =20
>   =20


--------------040705010903090204000402
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <meta content="text/html; charset=ISO-8859-1"
 http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Hi Brian,<br>
<br>
Thanks for your feedback.<br>
<blockquote
 cite="mid:7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch"
 type="cite">
  <pre wrap="">hi, Benoit, all...

Agreed that we should do something about this (i.e., that solution 1 is no solution.); that said, a few comments in no particular order:

1. In considering adding explicit semantics to structured data, we as a WG are taking on the task of defining semantics for IPFIX as a whole. Semantics, as I understand them, are in IPFIX largely contextual and template dependent, but in almost all cases I can think of it seems like these are implicitly "AND" semnantics (this flow has source IP A _AND_ destination IP B _AND_...). </pre>
</blockquote>
Agreed.<br>
<blockquote
 cite="mid:7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch"
 type="cite">
  <pre wrap="">We will need to make an explicit statement on this. </pre>
</blockquote>
Not sure why.<br>
<blockquote
 cite="mid:7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch"
 type="cite">
  <pre wrap="">We will need to determine whether these implicit semantics are a property of the protocol (in which case Structured Data is really a protocol-level extension), or a property of each information element (in which case all 5103 IEs have implicit AND semantics; question 2: will we need to add semantics to the IANA registry in this case?). We will need to be quite careful about this. It's not as simple as defining semantics within structured data then calling it done.
  </pre>
</blockquote>
I'm not sure why each individual IE should have a semantic ... in the
IANA registry.<br>
As you wrote, "in almost all cases I can think of it seems like these
are implicitly "AND" semnantics", so the case we try to solve is when
there are multiple instances of a single IE. We know that RFC 5101
foresaw that case " The Collector MUST support the use of Templates
containing multiple occurrences of the similar Information Elements",
but the idea is not to change RFC5101.&nbsp; If we put some semantic in the
IPFIX structured data, that would solve the vast majority of our cases.
Also, we could say: if you want some semantic when exporting multiple
IEs, then you SHOULD use IPFIX structured data.<br>
<br>
Also, the default value for the semantic field in the IPFIX structured
data SHOULD be "NONE", to express that flow record doesn't include any
semantic.... like in RFC5101. You might draw your own conclusion, maybe
because you know your network, maybe because you have configured the
exporter, but then it's your decision. <br>
The way I see the proposed solution is: in IPFIX structured data, you
MAY use the semantic field as a way to express the relationship between
IEs within the structure. <br>
<blockquote
 cite="mid:7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch"
 type="cite">
  <pre wrap="">
2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid argument against Solution 2, as it's trying to solve a somewhat different and more limited problem than structured data. Solution 2 _might_ cause a problem for this draft, but certainly not the other way around (unless we as a WG want to subsume that work into this draft, which would probably require rechartering...). Also, I don't think we need semantics for all of the list types, just basicList (Illustrative question: How do I meaningfully interpret two "or" subTemplateMultiLists with disjoint IE sets? Two "and" subTemplateMultiLists? Nestings thereof?)... So, we don't really have an explosion to deal with if we do Solution 2 correctly: andBasicList, orBasicList, xorBasicList, notBasicList; Four IEs, nestable, done. What can't we do with those four IEs? In this case, we could even step back and say that semantics outside these four are within the protocol explicitly _undefined_, and to be interpreted within the
 context of each Template.
  </pre>
</blockquote>
If I translate some more what I wrote "While it's solved the router and
most mediation function needs today", this would be "Me, myself, and I
don't need more that OR and AND <u>now </u>;-)".<br>
However, who am I to tell that others don't need it now... and that the
logical solution is to use the IPFIX structured data<br>
Furthermore, if we think a little bit longer term, the next big step in
IPFIX is the mediation function. In my company, every features want to
export his own data with NetFlow/IPFIX... up to the point where a CPE
would not have enough bandwidth across the WAN to export all the
"management" information. So we'll need more and more of aggregated
flow records (both in time and space) even in the router. Again, the
logical solution will be to use the IPFIX structured data. At this
point, we will most probably need something else than OR and AND, i.e.
RANGE, ORDERRED, etc...<br>
<br>
An example of subTemplateMultiLists with disjoint IE sets, let's
imagine that you have to export an aggregated observation point,
composed of multiple template records<br>
&nbsp;&nbsp;&nbsp; template 1: exporterIPaddress<br>
&nbsp;&nbsp;&nbsp; template 2: exporterIPaddress, basicList of interfaces<br>
&nbsp;&nbsp;&nbsp; template 3: exporterIPaddress, LC<br>
&nbsp;&nbsp;&nbsp; <br>
<br>
<blockquote
 cite="mid:7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch"
 type="cite">
  <pre wrap="">
3. If we really _do_ want ranges and so on (which, again, we'd need to get WG consensus on; this is explicitly out of scope in my reading of the present charter), then we could do them in the scope of Solution 3. </pre>
</blockquote>
Btw, acknowledging that one day we will have to solve this is good
enough for solution 3. I mean, we don't have to populate the semantic
IANA now... even if that would be more efficient.<br>
<blockquote
 cite="mid:7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch"
 type="cite">
  <pre wrap="">However, this seems a little not-quite-fleshed-out-enough for me to say whether I like it or not. Could you present an example of how you would use the proposed Semantic field to model your example? "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR linecard2"

(FWIW, I would do this with my proposal to Solution 2 as follows:)

(orBasicList (andBasicList (orBasicList ingressInterface eth1 eth2) (notBasicList (orBasicList ingressInterface eth3 eth4)) (andBasicList lineCardID 2))
  </pre>
</blockquote>
<pre wrap="">
(BasicList, OR, (basicList, AND, (basicList, OR, eth1, eth2), (basicList, NOR, eth3, eth4)), (basicList, NONE, linecard2)))

Thanks for your valuable feedback, as always.

Regards, Benoit.
</pre>
<blockquote
 cite="mid:7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch"
 type="cite">
  <pre wrap="">
Best regards,

Brian

On Mar 15, 2010, at 6:18 AM, Benoit Claise wrote:

  </pre>
  <blockquote type="cite">
    <pre wrap="">Dear all,

We've been thinking about this one, and we see 3 solutions. We believe that the solution 3 is the way to go, as we show below.

Solution 1. 
Consider that the semantic is out of scope for this document. 
This is the easiest solution.  However, we understand it's not right: we would not do a complete job wrt to IPFIX structured data
As Gerhard was expressing, the collector has not clue how to treat the example of a BasicList of egress interfaces in a Flow Record. 
    - Has every counted packet been sent on every egress interface?
      =&gt; multicast case, AND semantic
    - Has every counted packet been sent on any one of the egress interfaces?
     =&gt; load balancing case, OR semantic

Soltution 2
We could focus on the logical AND and OR semantics only , by defining semantic lists, such as andBasicList, andSubTemplateList, andSubTemplateMultiList, orBasicList, orSubTemplateList, and orSubTemplateMultiList
So 6 I.E.s in total, describing AND and OR semantics.
While it's solved the router and most mediation function needs today, we understand that this is not a complete solution.
Gerhard, in one of his old draft, draft-sommer-ipfix-mediator-ext-01, proposed ADTs for orderedList, orderedPair, and portRanges which allow the definition new IEs for port ranges etc.
Even if this solution 2 is extensible, in the long term, will lead to an explosion of IEs: 3 list types * the semantic type, where the semantic can be (and, or, orderedList, orderedPair, portRanges, random, etc...)

Soltution 3
We propose to add a semantic field in the 3 list types.
Something such as:

    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|0|               Field ID    |       Element Length            |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Semantic  |             BasicList Content ...                 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                           ...                                 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

This semantic field would be a new IANA registry, we could be populated initially with NONE, OR, AND, ORDERED, etc... Up to discussion.
The advantage of this solution is that it's extensible, and it doesn't need new IEs for each semantic.
If ever required (mainly in a mediation function), we can model  very complex semantic, eg, "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR linecard2", which could be the new observation point from a aggregated Flow Record



Conclusion:
After much debate internally, we believe that we should do the effort to include this semantic now, and not postpone the problem, which will lead to an explosion of IEs in the future.
Personally, I was initially against the solution 3 ... mostly due to the effort required to modify the complete specs.  
We're now ready to modify the specifications in the IPFIX structured data, but we would like to get your feedback and agreement in advance as this is not a small piece of work.

Please comment.

Regards, Paul, Stan, Gowri, and Benoit.
    </pre>
    <blockquote type="cite">
      <pre wrap="">hi Benoit, Gerhard, 

In this case I'm strongly in favor of leaving semantics out of structured data. Structured data defines containers. The semantics of the containers as a whole and the elements change based upon the information elements within the container and within the record containing it. Wedging semantics into the structured data elements 1. risks further proliferation of (potentially non-interoperable) ways to represent the same thing, 2. gives us more ways to represent nonsensical things (an OR basicList of MPLS stack entries...means...what?), and 3. risks defining an inadequate semantic representation mechanism (what about semantics for records not using structured data? what about ordered versus unordered sets? what about OR basicList vs AND basicList vs nested AND and OR basicLists vs just three identical IEs?). If we really want an unambiguous semantic framework for IPFIX (and here I'm not convinced either way) that's best done on its own, addressing things at the information elem
ent and record level. Doing it here confuses the issue. 

Regards, 

Brian 

On Nov 10, 2009, at 6:26 AM, Benoit Claise wrote: 

      </pre>
      <blockquote type="cite">
        <pre wrap="">Gerhard, 

Thanks for your email. 
I have no strong feelings about the two solutions you proposed. 

&gt;From a pure router point of view,  I don't see any use cases for logical OR in exporting flow records. 
However, from a IPFIX Mediator point of view, I see some use cases. 
I mean that it requires an Intermediate Aggregation Process or Intermediate Correlation Process to express: I've seen this flow record OR that flow record. 
Now, it's true that even routers will have mediation functions... 

I'm inclined to add orBasicList, orSubTemplateList, orSubTempalteMultiList to the draft (This is a small addition after all) and to express that, by default, a logical AND is assumed... specifically if the structured data is used in the IPFIX Mediation Protocol. 

I'm not convinced by the NOT in the context of structured data, as we don't even have a concept of NOT for a single information element! 

I would like to get some more feedback from others. 

Regards, Benoit. 

        </pre>
        <blockquote type="cite">
          <pre wrap="">Dear all, 

Regarding draft-ietf-ipfix-structured-data, I see the risk that the 
semantic of the exported structured data is not clear. 

How do you interpret the manifold occurrence of the same Information 
Element (basicList) or the same group of Information Elements 
(subTemplateList) in one record? 

What does it mean if basicList, subTemplateList, or subTemplateMultiList 
is used for a Flow Key field? Or non-key field? 

Some Examples: 

- BasicList of egress interfaces in a Flow Record 
 How should a Flow Record be interpreted which contains a list of 
 egress interfaces and a packet counter? 
 Has every counted packet been sent on every egress interface? 
   =&gt; multicast case, AND semantic (see example in section 8.1) 
 Has every counted packet been sent on any one of the egress 
 interfaces? 
   =&gt; load balancing case, OR semantic 
 Can it be used as a Flow Key or not? 

- BasicList of destination ports in a Flow Record 
 As every packet has only one destination port, the only reasonable 
 interpretation is that the Flow contains packets having one of 
 the reported port numbers. 
   =&gt; OR semantic 
 This would be a non-key field. 


I think there are two solutions: 

1. We decide that the semantic of list content is out of scope of 
  draft-ietf-ipfix-structured-data. We add a note to the draft that 
  the semantic must be clear from the context or the definition of the 
  Information Elements used within the lists. 

2. We define semantic lists, such as 
  - andBasicList, andSubTemplateList, andSubTemplateMultiList 
  - orBasicList, orSubTemplateList, orSubTempalteMultiList 
  describing AND and OR semantic of the contained IEs/Templates, 
  respectively. 


As I wrote in an earlier mail, I see a good use case for orBasicList. It 
could be used in the Selector Report Interpretation of Property Match 
Filtering to report a filter like "port 80 or port 443". 
<a class="moz-txt-link-freetext" href="http://www.ietf.org/mail-archive/web/ipfix/current/msg04856.html">http://www.ietf.org/mail-archive/web/ipfix/current/msg04856.html</a> 
At the moment, the Selector Report Interpretation is limited to AND. 
However, if we also want to express a NOT, we still need another solution... 

Regards, 
Gerhard 

_______________________________________________ 
IPFIX mailing list 
<a class="moz-txt-link-abbreviated" href="mailto:IPFIX@ietf.org">IPFIX@ietf.org</a> 
<a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/ipfix">https://www.ietf.org/mailman/listinfo/ipfix</a> 

          </pre>
        </blockquote>
        <pre wrap="">
_______________________________________________ 
IPFIX mailing list 
<a class="moz-txt-link-abbreviated" href="mailto:IPFIX@ietf.org">IPFIX@ietf.org</a> 
<a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/ipfix">https://www.ietf.org/mailman/listinfo/ipfix</a> 
        </pre>
      </blockquote>
    </blockquote>
    <pre wrap="">
    </pre>
  </blockquote>
  <pre wrap="">
  </pre>
</blockquote>
<br>
</body>
</html>

--------------040705010903090204000402--

From Thomas.Dietz@nw.neclab.eu  Tue Mar 16 06:37:30 2010
Return-Path: <Thomas.Dietz@nw.neclab.eu>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 8EA113A6A5C for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 06:37:30 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.855
X-Spam-Level: 
X-Spam-Status: No, score=-1.855 tagged_above=-999 required=5 tests=[AWL=-0.745, BAYES_05=-1.11]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id nJZPCr8LZCoY for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 06:37:29 -0700 (PDT)
Received: from smtp0.neclab.eu (smtp0.neclab.eu [195.37.70.41]) by core3.amsl.com (Postfix) with ESMTP id ECD703A6987 for <ipfix@ietf.org>; Tue, 16 Mar 2010 06:36:43 -0700 (PDT)
Received: from localhost (localhost.localdomain [127.0.0.1]) by smtp0.neclab.eu (Postfix) with ESMTP id 514522C020507; Tue, 16 Mar 2010 14:36:51 +0100 (CET)
X-Virus-Scanned: Amavisd on Debian GNU/Linux (atlas2.office)
Received: from smtp0.neclab.eu ([127.0.0.1]) by localhost (atlas2.office [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id COOveyxVFaps; Tue, 16 Mar 2010 14:36:51 +0100 (CET)
Received: from VENUS.office (mx2.office [192.168.24.15]) by smtp0.neclab.eu (Postfix) with ESMTP id 30BEA2C020505; Tue, 16 Mar 2010 14:36:41 +0100 (CET)
Content-class: urn:content-classes:message
MIME-Version: 1.0
X-MimeOLE: Produced By Microsoft Exchange V6.5
Date: Tue, 16 Mar 2010 14:36:40 +0100
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=SHA1; boundary="----=_NextPart_000_0113_01CAC516.17700490"
Message-ID: <547F018265F92642B577B986577D671C012A2D27@VENUS.office>
In-reply-to: <4B9E10EE.9050406@net.in.tum.de>
X-MS-Has-Attach: yes
X-MS-TNEF-Correlator: 
Thread-Topic: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIXMIB
Thread-Index: AcrELfuUyimrPJQjSfqzCIT0KhKeFwA3nGrA
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com><4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com><4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com><4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E10EE.9050406@net.in.tum.de>
From: "Thomas Dietz" <Thomas.Dietz@nw.neclab.eu>
To: "Gerhard Muenz" <muenz@net.in.tum.de>, "IETF IPFIX Working Group" <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIXMIB
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 16 Mar 2010 13:37:30 -0000

This is a multi-part message in MIME format.

------=_NextPart_000_0113_01CAC516.17700490
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Dear Gerhard,

find my replies inline.

--=20
Thomas Dietz
NEC Europe Ltd., NEC Laboratories, Network Research Division, 69115
Heidelberg, Germany

NEC Europe Limited, Registered in England 2832014
Registered Office: NEC House, 1 Victoria Road, London W3 6BL

> -----Original Message-----
> From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] On Behalf
> Of Gerhard Muenz
> Sent: Montag, 15. M=E4rz 2010 11:50
> To: IETF IPFIX Working Group
> Subject: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from
> IPFIXMIB
>=20
>=20
> Dear all,
>=20
> I would appreciate some feedback regarding counter values which have
> been adopted from IPFIX MIB.
>=20
> In the IPFIX MIB, packet or record counters start at certain point in
> time which is specified as "value of sysUpTime".
> (sysUpTime is a timer in MIB-2 for "the time (in hundredths of a
> second)
> since the network management portion of the system was last
> re-initialized.")
>=20
> What we usually want to know is one of the following:
> - absolute time when the packet/record counters were initialized, or
> - duration of the current counting period.
> To obtain these values, it is necessary to retrieve sysUpTime first
> from
> MIB-2, and then do some calculation.
>=20
> In the current configuration data model, we have adopted the counters
> and copied the definitions from the IPFIX MIB.
>=20
> Since IPFIX-CONFIG is not located in the SNMP world, I propose the
> following changes:
>=20
> - Counters start with zero (instead of arbitrary values)
>    =3D> yang type zero-based-counter64
> - Time at which counters start is specified as an absolute time stamp
>    (instead of sysUpTime value)
>    =3D> yang type time-and-date
>    (Alternatively, we could measure the duration of the counting =
period
>    in timeticks or milliseconds.)
>=20
> Are there any opinions/objections?
>=20

Fine with me. You have to make sure though that nobody mixes up the =
values
from IPFIX CONFIG and IPFIX MIB somehow. If you make it an absolute time
stamp you must make sure somehow that you really have the correct time. =
This
may be difficult on device reboot and within network environments that =
don't
give you a timeserver that keeps a correct time.

Anyway I doubt that you profit from making it an absolute time stamp =
since
counters are only meaningful if you take the difference between two =
counter
readouts. So you also have to take the difference of the times between =
last
readout of the counter and current readout to know how long it took to =
count
the number of whatsoever.

Best Regards,

Thomas

>=20
> Then, I do not understand what the following value in IPFIX MIB is
> supposed to be:
>=20
>     ipfixMeteringProcessCacheInactiveFlows OBJECT-TYPE
>         SYNTAX      Gauge32
>         MAX-ACCESS  read-only
>         STATUS      current
>         DESCRIPTION
>             "The number of Flows currently inactive at this cache."
>         ::=3D { ipfixMeteringProcessStatsEntry 2 }
>=20
>=20
> Well, I'm a coauthor of IPFIX MIB but I have to admit that it is not
> clear to me what inactive flows are supposed to be.
>=20
> Can anybody give an explanation?
>=20
>=20
> Regards,
> Gerhard
>=20
> --
> Dipl.-Ing. Gerhard M=FCnz
> Chair for Network Architectures and Services (I8)
> Technische Universit=E4t M=FCnchen - Department of Informatics
> Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
> Phone:  +49 89 289-18008       Fax: +49 89 289-18033
> E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz
>=20


------=_NextPart_000_0113_01CAC516.17700490
Content-Type: application/x-pkcs7-signature;
	name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment;
	filename="smime.p7s"
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------=_NextPart_000_0113_01CAC516.17700490--

From Thomas.Dietz@nw.neclab.eu  Tue Mar 16 06:39:27 2010
Return-Path: <Thomas.Dietz@nw.neclab.eu>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 6D85B3A67A1 for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 06:39:27 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.227
X-Spam-Level: 
X-Spam-Status: No, score=-2.227 tagged_above=-999 required=5 tests=[AWL=0.372,  BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id aky1bCN954DP for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 06:39:26 -0700 (PDT)
Received: from smtp0.neclab.eu (smtp0.neclab.eu [195.37.70.41]) by core3.amsl.com (Postfix) with ESMTP id 191503A67A4 for <ipfix@ietf.org>; Tue, 16 Mar 2010 06:39:24 -0700 (PDT)
Received: from localhost (localhost.localdomain [127.0.0.1]) by smtp0.neclab.eu (Postfix) with ESMTP id 516622C017B2E; Tue, 16 Mar 2010 14:39:32 +0100 (CET)
X-Virus-Scanned: Amavisd on Debian GNU/Linux (atlas2.office)
Received: from smtp0.neclab.eu ([127.0.0.1]) by localhost (atlas2.office [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id iDdyYB3jR+tJ; Tue, 16 Mar 2010 14:39:32 +0100 (CET)
Received: from VENUS.office (mx2.office [192.168.24.15]) by smtp0.neclab.eu (Postfix) with ESMTP id 2E4652C020508; Tue, 16 Mar 2010 14:39:17 +0100 (CET)
Content-class: urn:content-classes:message
MIME-Version: 1.0
X-MimeOLE: Produced By Microsoft Exchange V6.5
Date: Tue, 16 Mar 2010 14:39:16 +0100
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=SHA1; boundary="----=_NextPart_000_0117_01CAC516.74692280"
Message-ID: <547F018265F92642B577B986577D671C012A2D29@VENUS.office>
In-reply-to: <4B9E1F00.3080307@cisco.com>
X-MS-Has-Attach: yes
X-MS-TNEF-Correlator: 
Thread-Topic: [IPFIX] IPFIX errata 2052 to 2059
Thread-Index: AcrEOoNLHTiiXAVwQWKCZMwdmgE92AA00Hgg
References: <4B9AC2F8.1090504@cisco.com> <4B9B87BD.7020201@net.in.tum.de> <4B9E1F00.3080307@cisco.com>
From: "Thomas Dietz" <Thomas.Dietz@nw.neclab.eu>
To: "Paul Aitken" <paitken@cisco.com>, "Gerhard Muenz" <muenz@net.in.tum.de>
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] IPFIX errata 2052 to 2059
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 16 Mar 2010 13:39:27 -0000

This is a multi-part message in MIME format.

------=_NextPart_000_0117_01CAC516.74692280
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Dear all,

the only thing I want to add is that we cannot change easily the size of =
the
Id's in the MIB anyway because Unsigned32 is the biggest value you can =
get
in SNMP. There is no Unsigned64 defined for it.

Best Regards,

Thomas

--=20
Thomas Dietz                 E-mail: Thomas.Dietz@nw.neclab.eu
NEC Europe Ltd.              Phone:  +49 6221 4342-128
NEC Laboratories Europe      Fax:    +49 6221 4342-155
Network Research Division
Kurfuersten-Anlage 36
69115 Heidelberg, Germany    http://www.nw.neclab.eu

NEC Europe Limited           Registered in England 2832014
Registered Office: NEC House, 1 Victoria Road, London W3 6BL

> -----Original Message-----
> From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] On Behalf
> Of Paul Aitken
> Sent: Montag, 15. M=E4rz 2010 12:50
> To: Gerhard Muenz
> Cc: ipfix@ietf.org
> Subject: Re: [IPFIX] IPFIX errata 2052 to 2059
>=20
> Gerhard,
>=20
> We agree that unsigned16 selectorID is too small.
>=20
> Personally, I don't think field sizes should be specified in the info
> model at all, since sizes are specified in the templates. However,
> specifying sizes seems beneficial for collectors and necessary for the
> MIB.
>=20
> So our options are to increase the selectorID size to unsigned32 or
> unsigned64.
>=20
> u32 would help in the short to mid term. However, we can anticipate
> that
> even this will be too small in future when an observation domain may
> consist of many selectors or when selector Ids are encoded in blocks =
or
> groups.
>=20
> If we settled for u32, we could anticipate a future request for u64.
>=20
> Changing to u64 now seems to provide the greatest long-term
> flexibility.
>=20
> Cheers,
> P.
>=20
> > Still, I have not heard any good arguments why we need to change it
> to
> > unsigned64. Looking at the examples in RFC5476, it seems that the
> > original intention was unsigned32. This is inline with
> > ipfixSelectionProcessSelectorIndex in IPFIX-MIB.
> >
> > So, why do you prefer 64bit, which does not only require changes in
> > RFC5477, but in RFC5476 as well, and which is inconsistent with
> IPFIX-MIB?
> >
> > See also my previous mail:
> > http://www.ietf.org/mail-archive/web/ipfix/current/msg05311.html
> >
> > Regards,
> > Gerhard
> >
> > Paul Aitken wrote:
> >> Dear all,
> >>
> >> Discussion of these errata ended ~ 2 weeks ago. Presumably nobody
> has
> >> any objections?
> >>
> >> http://www.rfc-editor.org/errata_search.php?rfc=3D5477
> >> http://www.rfc-editor.org/errata_search.php?rfc=3D5476
> >>
> >> http://www.ietf.org/mail-archive/web/ipfix/current/msg05314.html
> >> http://www.ietf.org/mail-archive/web/ipfix/current/msg05315.html
> >>
> >>
> >> Dan, what's the next step? Ideally an update to the IANA registry.
> >>
> >>
> >> Thanks,
> >> P.
> >>
> >
>=20
>=20
> --
> Paul Aitken
> Cisco Systems Ltd, Edinburgh, Scotland.
> http://www.cisco.com/web/about/doing_business/legal/cri/index.html
> _______________________________________________
> IPFIX mailing list
> IPFIX@ietf.org
> https://www.ietf.org/mailman/listinfo/ipfix

------=_NextPart_000_0117_01CAC516.74692280
Content-Type: application/x-pkcs7-signature;
	name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment;
	filename="smime.p7s"

MIAGCSqGSIb3DQEHAqCAMIACAQExCzAJBgUrDgMCGgUAMIAGCSqGSIb3DQEHAQAAoIISNzCCA58w
ggKHoAMCAQICASYwDQYJKoZIhvcNAQEFBQAwcTELMAkGA1UEBhMCREUxHDAaBgNVBAoTE0RldXRz
Y2hlIFRlbGVrb20gQUcxHzAdBgNVBAsTFlQtVGVsZVNlYyBUcnVzdCBDZW50ZXIxIzAhBgNVBAMT
GkRldXRzY2hlIFRlbGVrb20gUm9vdCBDQSAyMB4XDTk5MDcwOTEyMTEwMFoXDTE5MDcwOTIzNTkw
MFowcTELMAkGA1UEBhMCREUxHDAaBgNVBAoTE0RldXRzY2hlIFRlbGVrb20gQUcxHzAdBgNVBAsT
FlQtVGVsZVNlYyBUcnVzdCBDZW50ZXIxIzAhBgNVBAMTGkRldXRzY2hlIFRlbGVrb20gUm9vdCBD
QSAyMIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAqwujNeCLKRSxFIWvPBDkOW81XUqu
3ephjZVJ9G9koxpgZqSpQCKE2dSl5XiTDmgBrblNXDrO07ioQkDfz6O6gllqkhusHJraCCslJ/lp
I0fx4Ossepv1EwLQfjR8wp48AFmr9doM9TI8K6xQ2tbD3oOUyqgMmTIOCEhWW2r72uFYWAFJX3JB
PBUGAY5draq4k7TNnuun6GotUjTbOu9cdVHa2/Mx+e5xmDLEVBVEDPmbVe2t3xgIoKOGiknuUwWP
GUzV3lh5m9JqHEKrxdWnz2gPluThYZh2YciRfNY+AOKRUIfhnQrmrZfSHcY6fcu82gM01Y5bAfVq
B7cWtm5KfwIDAQABo0IwQDAdBgNVHQ4EFgQUMcN5G7r1U9cX4Il6LRdsCrMrnTMwDwYDVR0TBAgw
BgEB/wIBBTAOBgNVHQ8BAf8EBAMCAQYwDQYJKoZIhvcNAQEFBQADggEBAJRkWa05ZOcp6xP+WsOL
E1fIBCTwdHfAYONn++mJpoO/loJ8btTDPe+egG67KbSYerE7VOs5F0d+Go4L/B8xWTEEss4X8yzH
YjZV4iLYiVW0mEiqZPrWHDbYRHhaWiM6V5f1ejBPrp9qTEsrjqAD4z7gqdTSe9KzqOJyPK2e/4BZ
5JtFtPY7sM05GZgy5eohYZDkMSGONLH3LzVKhRDa54o3Ib5ZY+DyhYgxU9RUFIVwefQuBncndS8f
uIr5/sW62Dbkg+znZbe/Y1rzRq+BlDfUQYzWI9Yez/VoG0Rjolq6pzVZoeVwBZsOI1eZlAptujlj
KIaS8xiE2PvRzwVWZFcwggQhMIIDCaADAgECAgIAxzANBgkqhkiG9w0BAQUFADBxMQswCQYDVQQG
EwJERTEcMBoGA1UEChMTRGV1dHNjaGUgVGVsZWtvbSBBRzEfMB0GA1UECxMWVC1UZWxlU2VjIFRy
dXN0IENlbnRlcjEjMCEGA1UEAxMaRGV1dHNjaGUgVGVsZWtvbSBSb290IENBIDIwHhcNMDYxMjE5
MTAyOTAwWhcNMTkwNjMwMjM1OTAwWjBaMQswCQYDVQQGEwJERTETMBEGA1UEChMKREZOLVZlcmVp
bjEQMA4GA1UECxMHREZOLVBLSTEkMCIGA1UEAxMbREZOLVZlcmVpbiBQQ0EgR2xvYmFsIC0gRzAx
MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEA6ZvDZ4X5Da71jVTDllA1PWLpbkztlNcA
W5UidNQg6zSP1uzAMQQLmYHiphTSUqAoI4SLdIkEXlvg4njBeMsWyyg1OXstkEXQ7aAAeny/Sg4b
AMOG6VwrMRF7DPOCJEOMHDiLamgAmu7cT3ir0sYTm3at7t4m6O8Br3QPwQmi9mvOvdPNFDBP9eXj
pMhim4IaAycwDQJlYE3t0QkjKpY1WCfTdsZxtpAdxO3/NYZ9bzOz2w/FEcKKg6GUXUFr2NIQ9Uz9
ylGs2b3vkoO72uuLFlZWQ8/h1RM9ph8nMM1JVNvJEzSacXXFbOqnC5j5IZ0nrz6jOTlIaoytyZn7
wxLyvQIDAQABo4HZMIHWMHAGA1UdHwRpMGcwZaBjoGGGX2h0dHA6Ly9wa2kudGVsZXNlYy5kZS9j
Z2ktYmluL3NlcnZpY2UvYWZfRG93bmxvYWRBUkwuY3JsPy1jcmxfZm9ybWF0PVhfNTA5Ji1pc3N1
ZXI9RFRfUk9PVF9DQV8yMB0GA1UdDgQWBBRJt8bP6D0ff+pEexMp9/EKcD7eZDAfBgNVHSMEGDAW
gBQxw3kbuvVT1xfgiXotF2wKsyudMzAOBgNVHQ8BAf8EBAMCAQYwEgYDVR0TAQH/BAgwBgEB/wIB
AjANBgkqhkiG9w0BAQUFAAOCAQEAO+Fad8BIF9ypGOyBr1qJ8L0okqbKWRgScOwo8ueuf5Ys5/Jd
GTH2Eyt0vb2Asrn3Z8k5onk74RER7mt4kTN+O18mJ3VTZY4zY+7Pc8OwkiNJIVB1I6EfGOKUhT0/
M+l3II2iveahhSlA9j9zMlgNCWum2oVswD+7jWZkViROrg0/MjUBW+mMgtlyWU+xhoXxdIVW5cP4
XPON7kezUwVw5+VNimmDKOETCYaeXsjqWB4MH/mk1FoEaP0oPosCtli19qEsN1cAZ6sjaI1jpe+Z
a1z9S1b2q0CHNNQRkmzsh8UKCwczcrRvDB1ULNhRx8y/MNNDcvEyv4zOSWOoAPfyHDCCBS8wggQX
oAMCAQICBA0hCkcwDQYJKoZIhvcNAQEFBQAwWjELMAkGA1UEBhMCREUxEzARBgNVBAoTCkRGTi1W
ZXJlaW4xEDAOBgNVBAsTB0RGTi1QS0kxJDAiBgNVBAMTG0RGTi1WZXJlaW4gUENBIEdsb2JhbCAt
IEcwMTAeFw0wODEwMjQwODUyMDhaFw0xOTA2MzAwMDAwMDBaMIGQMQswCQYDVQQGEwJERTEYMBYG
A1UEChMPTkVDIEV1cm9wZSBMdGQuMSAwHgYDVQQLExdORUMgTGFib3JhdG9yaWVzIEV1cm9wZTES
MBAGA1UEAxMJTkVDTEFCLUNBMTEwLwYJKoZIhvcNAQkBFiJ6ZXJ0aWZpemllcnVuZ3NzdGVsbGVA
bncubmVjbGFiLmV1MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAnvIVsbURqjIOcbnf
ruYkRceWOZpyvM2ebnYpbd1cP+zdWm6yR7HSO9ppOe1ZZFIasArqQpedPEFvcncSG94FRuW3ND4r
Rcq08mbpUTmpWmXfdYlJpQezbsOHCWR74NXoEEbK6TPZIMFpJr6dzQDAxnRc7UOgO6JQ1V42Z39B
PhIbPIWz64t8svafxbORmxulJn7F5zDLDcR1AEGyn+L9b645AGwapoKNh7cSQFTqdb6kGyPQjLWf
tv09dvmBDKesrcyLZXuDWJ1LMeizSjUEygdSszNXD3gePgJaVaZDS3o923W5gAyPCTSxpAFj8XJ+
/7Ap5jJwYhjJgJ8khFR7JQIDAQABo4IBxDCCAcAwEgYDVR0TAQH/BAgwBgEB/wIBATALBgNVHQ8E
BAMCAQYwHQYDVR0OBBYEFE8ch3od4C+Z9r4VqtE1nQ5K5ro2MB8GA1UdIwQYMBaAFEm3xs/oPR9/
6kR7Eyn38QpwPt5kMC0GA1UdEQQmMCSBInplcnRpZml6aWVydW5nc3N0ZWxsZUBudy5uZWNsYWIu
ZXUwgYgGA1UdHwSBgDB+MD2gO6A5hjdodHRwOi8vY2RwMS5wY2EuZGZuLmRlL2dsb2JhbC1yb290
LWNhL3B1Yi9jcmwvY2FjcmwuY3JsMD2gO6A5hjdodHRwOi8vY2RwMi5wY2EuZGZuLmRlL2dsb2Jh
bC1yb290LWNhL3B1Yi9jcmwvY2FjcmwuY3JsMIGiBggrBgEFBQcBAQSBlTCBkjBHBggrBgEFBQcw
AoY7aHR0cDovL2NkcDEucGNhLmRmbi5kZS9nbG9iYWwtcm9vdC1jYS9wdWIvY2FjZXJ0L2NhY2Vy
dC5jcnQwRwYIKwYBBQUHMAKGO2h0dHA6Ly9jZHAyLnBjYS5kZm4uZGUvZ2xvYmFsLXJvb3QtY2Ev
cHViL2NhY2VydC9jYWNlcnQuY3J0MA0GCSqGSIb3DQEBBQUAA4IBAQBsMQ+dD0mmi48dgDU4R6Q/
eXcY0zQcHNp1Vu1s3kwO0WahWB0tiqVBodvfTAWG44xuw0I7NXSTwQ68FU5P0GIQnvRFZyVJlDjr
SNlLYxjqfmgb+KVm67o383cZIuDakEE0f29kULIEn2fg/HsDiBAXTsb4I19XaN0TXLI+PMhU+GDp
sGCJrydeugEV7qi15q8yymjSAsYgnrc2wJuXpyQ9r3qCtP6aedAPSHqOT8ga1dLT2YRZFs3vNm7T
HSr5JJymWMbfpD6OcbRTnNAjSMDHwJxgRBAflA6WzDVm7fk4jiWyLvJwWTMk19t8QLiKG7D0nYvj
cEUYyiOSE+SFUTEdMIIFODCCBCCgAwIBAgIEDTI0WzANBgkqhkiG9w0BAQUFADCBkDELMAkGA1UE
BhMCREUxGDAWBgNVBAoTD05FQyBFdXJvcGUgTHRkLjEgMB4GA1UECxMXTkVDIExhYm9yYXRvcmll
cyBFdXJvcGUxEjAQBgNVBAMTCU5FQ0xBQi1DQTExMC8GCSqGSIb3DQEJARYiemVydGlmaXppZXJ1
bmdzc3RlbGxlQG53Lm5lY2xhYi5ldTAeFw0wODExMDYwOTIwMTFaFw0xMTExMDYwOTIwMTFaMGAx
CzAJBgNVBAYTAkRFMRgwFgYDVQQKEw9ORUMgRXVyb3BlIEx0ZC4xIDAeBgNVBAsTF05FQyBMYWJv
cmF0b3JpZXMgRXVyb3BlMRUwEwYDVQQDEwxUaG9tYXMgRGlldHowggEiMA0GCSqGSIb3DQEBAQUA
A4IBDwAwggEKAoIBAQCniLuqlMflMs7ag3idESVRwfZ9nrdIyUmq5Tget4k9APGSPo2GZ1f1hr8y
MuJIGowc/HzS1laVSICclOXju1xW93Tm+Vco5gwkRqHXY3Rmda0r2jlb8niVie4qXQgXPzunFRqk
yNmbwYep3oD5Kq0GfB6EuZ7X6cYH5A7erAMeeQPhoDQDIfR79lRHlcMtanJZyORYNQONPEa+L4AF
v5f3nAsmY7ZLJ72wX7X8BtcF6Vdkr0T2b1YrPl8Qir7e0TKY9Rf0q5DKu3QdLnXk0Rb+63V/4vLS
PZ3XQVdwHzLiOhIZJVVMJE7TmJI0DFeDFn99O/F/Le/sJOJjNO4n6KMLAgMBAAGjggHHMIIBwzAJ
BgNVHRMEAjAAMAsGA1UdDwQEAwIF4DApBgNVHSUEIjAgBggrBgEFBQcDAgYIKwYBBQUHAwQGCisG
AQQBgjcUAgIwHQYDVR0OBBYEFLkbmpJJpIXhIR9JFttGC+KHW5g5MB8GA1UdIwQYMBaAFE8ch3od
4C+Z9r4VqtE1nQ5K5ro2MCQGA1UdEQQdMBuBGVRob21hcy5EaWV0ekBudy5uZWNsYWIuZXUwfQYD
VR0fBHYwdDA4oDagNIYyaHR0cDovL2NkcDEucGNhLmRmbi5kZS9uZWNsYWItY2EvcHViL2NybC9j
YWNybC5jcmwwOKA2oDSGMmh0dHA6Ly9jZHAyLnBjYS5kZm4uZGUvbmVjbGFiLWNhL3B1Yi9jcmwv
Y2FjcmwuY3JsMIGYBggrBgEFBQcBAQSBizCBiDBCBggrBgEFBQcwAoY2aHR0cDovL2NkcDEucGNh
LmRmbi5kZS9uZWNsYWItY2EvcHViL2NhY2VydC9jYWNlcnQuY3J0MEIGCCsGAQUFBzAChjZodHRw
Oi8vY2RwMi5wY2EuZGZuLmRlL25lY2xhYi1jYS9wdWIvY2FjZXJ0L2NhY2VydC5jcnQwDQYJKoZI
hvcNAQEFBQADggEBAEbKwRBhNxAzsH6kAYRBoIziyI2IY2QnZGeiGitOfkLKcFNIBH3ZfQXH4j20
4BP34Vzzob17EgsbLbNkMlxh2M7tenXH9vA4MiN5yPPBKoR7SfI6mIaIr+7kewOizJ8D5dvpdfP5
HbAUTZVSYHqixgBWJyNp7sXWVQtOo+eOv8qKUeiodClanKuCnGD42Bl6EQR486dlXvOronEikRYX
Xg6gFrhO+DonUYluMZpNdabnybKozchSdHcceOKd0JFMmyiSNG944ystXbR7QZqfNSh/Pbyc5QRp
Z1vdFZLFh907iS3KxueDYKDPWmlsPt/QnmXmF4A9/5OrmVS1C45k5rcxggQ4MIIENAIBATCBmTCB
kDELMAkGA1UEBhMCREUxGDAWBgNVBAoTD05FQyBFdXJvcGUgTHRkLjEgMB4GA1UECxMXTkVDIExh
Ym9yYXRvcmllcyBFdXJvcGUxEjAQBgNVBAMTCU5FQ0xBQi1DQTExMC8GCSqGSIb3DQEJARYiemVy
dGlmaXppZXJ1bmdzc3RlbGxlQG53Lm5lY2xhYi5ldQIEDTI0WzAJBgUrDgMCGgUAoIICczAYBgkq
hkiG9w0BCQMxCwYJKoZIhvcNAQcBMBwGCSqGSIb3DQEJBTEPFw0xMDAzMTYxMzM5MTZaMCMGCSqG
SIb3DQEJBDEWBBQRf7XC58/LTJxLQocJrSH2JQv/ZzCBqgYJKwYBBAGCNxAEMYGcMIGZMIGQMQsw
CQYDVQQGEwJERTEYMBYGA1UEChMPTkVDIEV1cm9wZSBMdGQuMSAwHgYDVQQLExdORUMgTGFib3Jh
dG9yaWVzIEV1cm9wZTESMBAGA1UEAxMJTkVDTEFCLUNBMTEwLwYJKoZIhvcNAQkBFiJ6ZXJ0aWZp
emllcnVuZ3NzdGVsbGVAbncubmVjbGFiLmV1AgQNMjRbMIGsBgsqhkiG9w0BCRACCzGBnKCBmTCB
kDELMAkGA1UEBhMCREUxGDAWBgNVBAoTD05FQyBFdXJvcGUgTHRkLjEgMB4GA1UECxMXTkVDIExh
Ym9yYXRvcmllcyBFdXJvcGUxEjAQBgNVBAMTCU5FQ0xBQi1DQTExMC8GCSqGSIb3DQEJARYiemVy
dGlmaXppZXJ1bmdzc3RlbGxlQG53Lm5lY2xhYi5ldQIEDTI0WzCBtwYJKoZIhvcNAQkPMYGpMIGm
MAsGCWCGSAFlAwQBKjALBglghkgBZQMEARYwCgYIKoZIhvcNAwcwCwYJYIZIAWUDBAECMA4GCCqG
SIb3DQMCAgIAgDANBggqhkiG9w0DAgIBQDAHBgUrDgMCBzANBggqhkiG9w0DAgIBKDAHBgUrDgMC
GjALBglghkgBZQMEAgMwCwYJYIZIAWUDBAICMAsGCWCGSAFlAwQCATAKBggqhkiG9w0CBTANBgkq
hkiG9w0BAQEFAASCAQAr0HZsZ6LhksI2DbXWSXY+Xur70g3+vHK9uzRhEJvtAhKDNcK/5I2VAKjr
YqXJj3AD4kviIQgUeLm72SQpYAKbb/JhRWUns0e1+Xf93jl5QFX6zU3Ghb2pppYeQaLEwu6XF0oE
wzRPoDS4Ev20n4laqghCqlKvU/qxvnHLhVSyBwuddnYzFAVpaOzTGJm4/OiRoo8+EXnZZE1kt4Ua
TFu7bxzcAXummQEtmuR2JCoaf34CinSzfB0eA8kubGITsRClG3QslIKADNxJ5lBmjAhVDlSkq2Vh
Tke9r5vJcqYPb+9rdnjdxeItVqeLofjIDOGZGs0BETgR3v8swBzLUZD6AAAAAAAA

------=_NextPart_000_0117_01CAC516.74692280--

From muenz@net.in.tum.de  Tue Mar 16 07:27:41 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 2C6B63A698E for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 07:27:41 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.145
X-Spam-Level: 
X-Spam-Status: No, score=-2.145 tagged_above=-999 required=5 tests=[AWL=0.104,  BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id YxY24gSmXccC for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 07:27:40 -0700 (PDT)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 86B333A6984 for <ipfix@ietf.org>; Tue, 16 Mar 2010 07:27:37 -0700 (PDT)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id DA7552010C77; Tue, 16 Mar 2010 15:27:39 +0100 (CET)
Message-ID: <4B9F9574.1080606@net.in.tum.de>
Date: Tue, 16 Mar 2010 15:28:04 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Thomas Dietz <Thomas.Dietz@nw.neclab.eu>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com><4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com><4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com><4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E10EE.9050406@net.in.tum.de> <547F018265F92642B577B986577D671C012A2D27@VENUS.office>
In-Reply-To: <547F018265F92642B577B986577D671C012A2D27@VENUS.office>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms040609000302060808040900"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIXMIB
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 16 Mar 2010 14:27:41 -0000

This is a cryptographically signed message in MIME format.

--------------ms040609000302060808040900
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Thomas,

>> -----Original Message-----
>> > From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] On Beha=
lf
>> > Of Gerhard Muenz
>> > Sent: Montag, 15. M=E4rz 2010 11:50
>> > To: IETF IPFIX Working Group
>> > Subject: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from=

>> > IPFIXMIB
>> >=20
>> >=20
>> > Dear all,
>> >=20
>> > I would appreciate some feedback regarding counter values which have=

>> > been adopted from IPFIX MIB.
>> >=20
>> > In the IPFIX MIB, packet or record counters start at certain point i=
n
>> > time which is specified as "value of sysUpTime".
>> > (sysUpTime is a timer in MIB-2 for "the time (in hundredths of a
>> > second)
>> > since the network management portion of the system was last
>> > re-initialized.")
>> >=20
>> > What we usually want to know is one of the following:
>> > - absolute time when the packet/record counters were initialized, or=

>> > - duration of the current counting period.
>> > To obtain these values, it is necessary to retrieve sysUpTime first
>> > from
>> > MIB-2, and then do some calculation.
>> >=20
>> > In the current configuration data model, we have adopted the counter=
s
>> > and copied the definitions from the IPFIX MIB.
>> >=20
>> > Since IPFIX-CONFIG is not located in the SNMP world, I propose the
>> > following changes:
>> >=20
>> > - Counters start with zero (instead of arbitrary values)
>> >    =3D> yang type zero-based-counter64
>> > - Time at which counters start is specified as an absolute time stam=
p
>> >    (instead of sysUpTime value)
>> >    =3D> yang type time-and-date
>> >    (Alternatively, we could measure the duration of the counting per=
iod
>> >    in timeticks or milliseconds.)
>> >=20
>> > Are there any opinions/objections?
>> >=20
>=20
> Fine with me. You have to make sure though that nobody mixes up the val=
ues
> from IPFIX CONFIG and IPFIX MIB somehow.=20

Yes. The time parameters will be called differently (not discontinuity=20
time as in the MIB) to avoid confusion.

> If you make it an absolute time stamp you must make sure somehow that=20
> you really have the correct time. This
> may be difficult on device reboot and within network environments that =
don't
> give you a timeserver that keeps a correct time.

I am aware of this problem. However, in the IPFIX Message header, we have=
:

    Export Time

       Time, in seconds, since 0000 UTC Jan 1, 1970, at which the IPFIX
       Message Header leaves the Exporter.

So, at least the Exporter must know the absolute time anyway.

> Anyway I doubt that you profit from making it an absolute time stamp si=
nce
> counters are only meaningful if you take the difference between two cou=
nter
> readouts. So you also have to take the difference of the times between =
last
> readout of the counter and current readout to know how long it took to =
count
> the number of whatsoever.

As we use zero-based counters in IPFIX-CONFIG, you know the counters=20
from the indicated point in time.

Thanks and regards,
Gerhard

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms040609000302060808040900
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms040609000302060808040900--

From Thomas.Dietz@nw.neclab.eu  Tue Mar 16 07:44:56 2010
Return-Path: <Thomas.Dietz@nw.neclab.eu>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 00C093A6842 for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 07:44:56 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.351
X-Spam-Level: 
X-Spam-Status: No, score=-2.351 tagged_above=-999 required=5 tests=[AWL=0.248,  BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id uNT3-d-+2uC7 for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 07:44:54 -0700 (PDT)
Received: from smtp0.neclab.eu (smtp0.neclab.eu [195.37.70.41]) by core3.amsl.com (Postfix) with ESMTP id C37893A698E for <ipfix@ietf.org>; Tue, 16 Mar 2010 07:43:48 -0700 (PDT)
Received: from localhost (localhost.localdomain [127.0.0.1]) by smtp0.neclab.eu (Postfix) with ESMTP id 2BA3A2C020505; Tue, 16 Mar 2010 15:43:57 +0100 (CET)
X-Virus-Scanned: Amavisd on Debian GNU/Linux (atlas2.office)
Received: from smtp0.neclab.eu ([127.0.0.1]) by localhost (atlas2.office [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 58q3cO6r8vzI; Tue, 16 Mar 2010 15:43:57 +0100 (CET)
Received: from VENUS.office (mx2.office [192.168.24.15]) by smtp0.neclab.eu (Postfix) with ESMTP id 0BD3A2C0204F9; Tue, 16 Mar 2010 15:43:47 +0100 (CET)
Content-class: urn:content-classes:message
MIME-Version: 1.0
X-MimeOLE: Produced By Microsoft Exchange V6.5
Date: Tue, 16 Mar 2010 15:43:46 +0100
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=SHA1; boundary="----=_NextPart_000_0129_01CAC51F.76EE30A0"
Message-ID: <547F018265F92642B577B986577D671C012A2D5B@VENUS.office>
In-reply-to: <4B9F9574.1080606@net.in.tum.de>
X-MS-Has-Attach: yes
X-MS-TNEF-Correlator: 
Thread-Topic: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIXMIB
Thread-Index: AcrFFNjQaegr0qdkST+hT1oXddXrQwAAbAjQ
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com><4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com><4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com><4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E10EE.9050406@net.in.tum.de> <547F018265F92642B577B986577D671C012A2D27@VENUS.office> <4B9F9574.1080606@net.in.tum.de>
From: "Thomas Dietz" <Thomas.Dietz@nw.neclab.eu>
To: "Gerhard Muenz" <muenz@net.in.tum.de>
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIXMIB
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 16 Mar 2010 14:44:56 -0000

This is a multi-part message in MIME format.

------=_NextPart_000_0129_01CAC51F.76EE30A0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Hi Gerhard,

--=20
Thomas Dietz                 E-mail: Thomas.Dietz@nw.neclab.eu
NEC Europe Ltd.              Phone:  +49 6221 4342-128
NEC Laboratories Europe      Fax:    +49 6221 4342-155
Network Research Division
Kurfuersten-Anlage 36
69115 Heidelberg, Germany    http://www.nw.neclab.eu

NEC Europe Limited           Registered in England 2832014
Registered Office: NEC House, 1 Victoria Road, London W3 6BL

> -----Original Message-----
> From: Gerhard Muenz [mailto:muenz@net.in.tum.de]
> Sent: Dienstag, 16. M=E4rz 2010 15:28
> To: Thomas Dietz
> Cc: IETF IPFIX Working Group
> Subject: Re: [IPFIX] IPFIX-CONFIG: Questions about counters adopted
> from IPFIXMIB
>=20
>=20
> Hi Thomas,
>=20
> >> -----Original Message-----
> >> > From: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] On
> Behalf
> >> > Of Gerhard Muenz
> >> > Sent: Montag, 15. M=E4rz 2010 11:50
> >> > To: IETF IPFIX Working Group
> >> > Subject: [IPFIX] IPFIX-CONFIG: Questions about counters adopted
> from
> >> > IPFIXMIB
> >> >
> >> >
> >> > Dear all,
> >> >
> >> > I would appreciate some feedback regarding counter values which
> have
> >> > been adopted from IPFIX MIB.
> >> >
> >> > In the IPFIX MIB, packet or record counters start at certain =
point
> in
> >> > time which is specified as "value of sysUpTime".
> >> > (sysUpTime is a timer in MIB-2 for "the time (in hundredths of a
> >> > second)
> >> > since the network management portion of the system was last
> >> > re-initialized.")
> >> >
> >> > What we usually want to know is one of the following:
> >> > - absolute time when the packet/record counters were initialized,
> or
> >> > - duration of the current counting period.
> >> > To obtain these values, it is necessary to retrieve sysUpTime
> first
> >> > from
> >> > MIB-2, and then do some calculation.
> >> >
> >> > In the current configuration data model, we have adopted the
> counters
> >> > and copied the definitions from the IPFIX MIB.
> >> >
> >> > Since IPFIX-CONFIG is not located in the SNMP world, I propose =
the
> >> > following changes:
> >> >
> >> > - Counters start with zero (instead of arbitrary values)
> >> >    =3D> yang type zero-based-counter64
> >> > - Time at which counters start is specified as an absolute time
> stamp
> >> >    (instead of sysUpTime value)
> >> >    =3D> yang type time-and-date
> >> >    (Alternatively, we could measure the duration of the counting
> period
> >> >    in timeticks or milliseconds.)
> >> >
> >> > Are there any opinions/objections?
> >> >
> >
> > Fine with me. You have to make sure though that nobody mixes up the
> values
> > from IPFIX CONFIG and IPFIX MIB somehow.
>=20
> Yes. The time parameters will be called differently (not discontinuity
> time as in the MIB) to avoid confusion.
>=20

Be careful here: the discontinuity time is the time when anything went =
wrong
and you cannot rely on the value of the counter anymore because it may =
have
missed some packets or whatsoever. This is not the time the counter was =
last
updated or started!

By the way this is also something you may need?


> > If you make it an absolute time stamp you must make sure somehow =
that
> > you really have the correct time. This
> > may be difficult on device reboot and within network environments
> that don't
> > give you a timeserver that keeps a correct time.
>=20
> I am aware of this problem. However, in the IPFIX Message header, we
> have:
>=20
>     Export Time
>=20
>        Time, in seconds, since 0000 UTC Jan 1, 1970, at which the =
IPFIX
>        Message Header leaves the Exporter.
>=20
> So, at least the Exporter must know the absolute time anyway.
>=20
> > Anyway I doubt that you profit from making it an absolute time stamp
> since
> > counters are only meaningful if you take the difference between two
> counter
> > readouts. So you also have to take the difference of the times
> between last
> > readout of the counter and current readout to know how long it took
> to count
> > the number of whatsoever.
>=20
> As we use zero-based counters in IPFIX-CONFIG, you know the counters
> from the indicated point in time.
>=20

Fine but what do you do if your counter wraps?

Regards,

Thomas

> Thanks and regards,
> Gerhard
>=20
> --
> Dipl.-Ing. Gerhard M=FCnz
> Chair for Network Architectures and Services (I8)
> Technische Universit=E4t M=FCnchen - Department of Informatics
> Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
> Phone:  +49 89 289-18008       Fax: +49 89 289-18033
> E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz
>=20


------=_NextPart_000_0129_01CAC51F.76EE30A0
Content-Type: application/x-pkcs7-signature;
	name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment;
	filename="smime.p7s"
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------=_NextPart_000_0129_01CAC51F.76EE30A0--

From muenz@net.in.tum.de  Tue Mar 16 08:31:07 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 6CEF23A6B4B for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 08:31:07 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.149
X-Spam-Level: 
X-Spam-Status: No, score=-2.149 tagged_above=-999 required=5 tests=[AWL=0.100,  BAYES_00=-2.599, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id y99Tkt0zujsH for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 08:31:03 -0700 (PDT)
Received: from mail-out2.informatik.tu-muenchen.de (mail-out2.informatik.tu-muenchen.de [131.159.0.36]) by core3.amsl.com (Postfix) with ESMTP id E12C23A6B9F for <ipfix@ietf.org>; Tue, 16 Mar 2010 08:31:00 -0700 (PDT)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 7F8ED2010C77; Tue, 16 Mar 2010 16:31:05 +0100 (CET)
Message-ID: <4B9FA451.8080802@net.in.tum.de>
Date: Tue, 16 Mar 2010 16:31:29 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Thomas Dietz <Thomas.Dietz@nw.neclab.eu>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com><4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com><4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com><4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E10EE.9050406@net.in.tum.de> <547F018265F92642B577B986577D671C012A2D27@VENUS.office> <4B9F9574.1080606@net.in.tum.de> <547F018265F92642B577B986577D671C012A2D5B@VENUS.office>
In-Reply-To: <547F018265F92642B577B986577D671C012A2D5B@VENUS.office>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms010106030406060109010605"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIXMIB
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 16 Mar 2010 15:31:07 -0000

This is a cryptographically signed message in MIME format.

--------------ms010106030406060109010605
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable

Hi Thomas,

> Be careful here: the discontinuity time is the time when anything went =
wrong
> and you cannot rely on the value of the counter anymore because it may =
have
> missed some packets or whatsoever. This is not the time the counter was=
 last
> updated or started!
>=20
> By the way this is also something you may need?
 >
> Fine but what do you do if your counter wraps?

Good points.

My original point was that I would like to get away from sysUpTime=20
because it is not sure from where you can get it with Netconf.

Discontinuity times like in the MIB are ok, but then they should contain =

absolute time stamps.

Maybe it is the easiest to only make this little change and keep the=20
definitions of the other counters as they are in the MIB.

What do you think?

Regards,
Gerhard

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms010106030406060109010605
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms010106030406060109010605--

From trammell@tik.ee.ethz.ch  Tue Mar 16 11:22:23 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 25A413A6A7D for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 11:22:23 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.577
X-Spam-Level: 
X-Spam-Status: No, score=-6.577 tagged_above=-999 required=5 tests=[AWL=0.022,  BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id rtgg9skX9yh7 for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 11:22:19 -0700 (PDT)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id 34DD93A69C2 for <ipfix@ietf.org>; Tue, 16 Mar 2010 11:20:58 -0700 (PDT)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id C07C8D939E; Tue, 16 Mar 2010 19:21:05 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id d1EzU25rgQyE; Tue, 16 Mar 2010 19:21:05 +0100 (MET)
Received: from [10.0.1.5] (unknown [75.85.164.240]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id A4A0CD9375; Tue, 16 Mar 2010 19:21:03 +0100 (MET)
Mime-Version: 1.0 (Apple Message framework v1077)
Content-Type: text/plain; charset=us-ascii
From: Brian Trammell <trammell@tik.ee.ethz.ch>
In-Reply-To: <4B9F520E.8060507@cisco.com>
Date: Tue, 16 Mar 2010 09:48:28 -0700
Content-Transfer-Encoding: quoted-printable
Message-Id: <DB6E59D9-B373-4919-BB58-00EB26014564@tik.ee.ethz.ch>
References: <4AF73525.8050009@net.in.tum.de> <4AF8F999.3000207@cisco.com>	<F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>	<4B9E33BE.7070907@cisco.com> <7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch> <4B9F520E.8060507@cisco.com>
To: Benoit Claise <bclaise@cisco.com>
X-Mailer: Apple Mail (2.1077)
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 16 Mar 2010 18:22:23 -0000

Hi, Benoit,

Replies inline...

On Mar 16, 2010, at 2:40 AM, Benoit Claise wrote:

> Hi Brian,
>=20
> Thanks for your feedback.
>> hi, Benoit, all...
>>=20
>> Agreed that we should do something about this (i.e., that solution 1 =
is no solution.); that said, a few comments in no particular order:
>>=20
>> 1. In considering adding explicit semantics to structured data, we as =
a WG are taking on the task of defining semantics for IPFIX as a whole. =
Semantics, as I understand them, are in IPFIX largely contextual and =
template dependent, but in almost all cases I can think of it seems like =
these are implicitly "AND" semnantics (this flow has source IP A _AND_ =
destination IP B _AND_...).=20
>>=20
> Agreed.
>> We will need to make an explicit statement on this.=20
> Not sure why.
>> We will need to determine whether these implicit semantics are a =
property of the protocol (in which case Structured Data is really a =
protocol-level extension), or a property of each information element (in =
which case all 5103 IEs have implicit AND semantics; question 2: will we =
need to add semantics to the IANA registry in this case?). We will need =
to be quite careful about this. It's not as simple as defining semantics =
within structured data then calling it done.
>>  =20
>>=20
> I'm not sure why each individual IE should have a semantic ... in the =
IANA registry.

I'm not either. On reflection it seems like overkill. I'm just saying =
that if we, as a WG, are moving from stating that semantics are =
explicitly out of scope to defining them as in-scope, we need to have a =
consistent approach, and answers to all the questions that arise when we =
consider moving the protocol from a simple framing mechanism to a =
framing mechanism with some logic behind it, so we answer them once, so =
that all future efforts having to do with semantics are consistent. =
This, I acknowledge, is an argument in favor of solution 3...)

One very simple new question that arises here, to illustrate my point: =
Is it legal to export a record that has sourceIPAddress X AND NOT =
sourceIPAddress X?=20

> As you wrote, "in almost all cases I can think of it seems like these =
are implicitly "AND" semnantics", so the case we try to solve is when =
there are multiple instances of a single IE. We know that RFC 5101 =
foresaw that case " The Collector MUST support the use of Templates =
containing multiple occurrences of the similar Information Elements", =
but the idea is not to change RFC5101.  If we put some semantic in the =
IPFIX structured data, that would solve the vast majority of our cases. =
Also, we could say: if you want some semantic when exporting multiple =
IEs, then you SHOULD use IPFIX structured data.
>=20
> Also, the default value for the semantic field in the IPFIX structured =
data SHOULD be "NONE", to express that flow record doesn't include any =
semantic.... like in RFC5101. You might draw your own conclusion, maybe =
because you know your network, maybe because you have configured the =
exporter, but then it's your decision.=20
> The way I see the proposed solution is: in IPFIX structured data, you =
MAY use the semantic field as a way to express the relationship between =
IEs within the structure.=20
>> 2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid =
argument against Solution 2, as it's trying to solve a somewhat =
different and more limited problem than structured data. Solution 2 =
_might_ cause a problem for this draft, but certainly not the other way =
around (unless we as a WG want to subsume that work into this draft, =
which would probably require rechartering...). Also, I don't think we =
need semantics for all of the list types, just basicList (Illustrative =
question: How do I meaningfully interpret two "or" subTemplateMultiLists =
with disjoint IE sets? Two "and" subTemplateMultiLists? Nestings =
thereof?)... So, we don't really have an explosion to deal with if we do =
Solution 2 correctly: andBasicList, orBasicList, xorBasicList, =
notBasicList; Four IEs, nestable, done. What can't we do with those four =
IEs? In this case, we could even step back and say that semantics =
outside these four are within the protocol explicitly _undefined_, and =
to be interpreted within the
>>  context of each Template.
>>  =20
>>=20
> If I translate some more what I wrote "While it's solved the router =
and most mediation function needs today", this would be "Me, myself, and =
I don't need more that OR and AND now ;-)".
> However, who am I to tell that others don't need it now... and that =
the logical solution is to use the IPFIX structured data
> Furthermore, if we think a little bit longer term, the next big step =
in IPFIX is the mediation function. In my company, every features want =
to export his own data with NetFlow/IPFIX... up to the point where a CPE =
would not have enough bandwidth across the WAN to export all the =
"management" information. So we'll need more and more of aggregated flow =
records (both in time and space) even in the router. Again, the logical =
solution will be to use the IPFIX structured data. At this point, we =
will most probably need something else than OR and AND, i.e. RANGE, =
ORDERRED, etc...
>=20
> An example of subTemplateMultiLists with disjoint IE sets, let's =
imagine that you have to export an aggregated observation point, =
composed of multiple template records
>     template 1: exporterIPaddress
>     template 2: exporterIPaddress, basicList of interfaces
>     template 3: exporterIPaddress, LC

and then you'd want to OR these... Okay, makes sense...

>=20
>> 3. If we really _do_ want ranges and so on (which, again, we'd need =
to get WG consensus on; this is explicitly out of scope in my reading of =
the present charter), then we could do them in the scope of Solution 3.=20=

> Btw, acknowledging that one day we will have to solve this is good =
enough for solution 3. I mean, we don't have to populate the semantic =
IANA now... even if that would be more efficient.
>> However, this seems a little not-quite-fleshed-out-enough for me to =
say whether I like it or not. Could you present an example of how you =
would use the proposed Semantic field to model your example? "(eth1 OR =
eth2) AND (NOT (eth3 OR eth4)) OR linecard2"
>>=20
>> (FWIW, I would do this with my proposal to Solution 2 as follows:)
>>=20
>> (orBasicList (andBasicList (orBasicList ingressInterface eth1 eth2) =
(notBasicList (orBasicList ingressInterface eth3 eth4)) (andBasicList =
lineCardID 2))
>>  =20
>>=20
> (BasicList, OR, (basicList, AND, (basicList, OR, eth1, eth2), =
(basicList, NOR, eth3, eth4)), (basicList, NONE, linecard2)))

Hm. Okay. This makes sense. A couple of questions then, about solution =
3:

1. Would we want to try and bitfield this, to define the semantics we =
_know_ we have, then leave the rest of it reserved? Something like:

MSb                      LSb
+---+------+---------------+
| ! | multi| reserved      |
+---+------+---------------+

! =3D negate sense of semantics if 1 (this is the NOT flag)
multi (multiplicity) =3D 00: undefined, 01: or/oneOrMore, 10: =
xor/exactlyOne, 11: and/exactlyAll
reserved =3D place for adding other bells and whistles like RANGE and so =
on in the future.

2. Are we sure we want to do this in one byte? If we do bitfielding as =
above this gives us 32 possible extensions, which seems like _way_ more =
than enough, but does stick another odd offset in there, which slows =
things down on machines that need aligned access. Probably one byte is =
okay and we let the implementation use paddingOctets and set padding to =
fix this...

3. Would it make sense maybe to have two sets of structured data =
elements, one with the semantics byte, and one without (which is then =
explicitly undefined)? Then exporters who don't need it don't have to =
bother sticking an extra odd-aligned zero in the stream for every list.

I'm sure I'l have more questions, but none come to mind now... But it =
seems like we're converging on the least-unnecessarily-complex solution =
here, which is good. :)

Best regards,

Brian

>> On Mar 15, 2010, at 6:18 AM, Benoit Claise wrote:
>>=20
>>  =20
>>=20
>>> Dear all,
>>>=20
>>> We've been thinking about this one, and we see 3 solutions. We =
believe that the solution 3 is the way to go, as we show below.
>>>=20
>>> Solution 1.=20
>>> Consider that the semantic is out of scope for this document.=20
>>> This is the easiest solution.  However, we understand it's not =
right: we would not do a complete job wrt to IPFIX structured data
>>> As Gerhard was expressing, the collector has not clue how to treat =
the example of a BasicList of egress interfaces in a Flow Record.=20
>>>     - Has every counted packet been sent on every egress interface?
>>>       =3D> multicast case, AND semantic
>>>     - Has every counted packet been sent on any one of the egress =
interfaces?
>>>      =3D> load balancing case, OR semantic
>>>=20
>>> Soltution 2
>>> We could focus on the logical AND and OR semantics only , by =
defining semantic lists, such as andBasicList, andSubTemplateList, =
andSubTemplateMultiList, orBasicList, orSubTemplateList, and =
orSubTemplateMultiList
>>> So 6 I.E.s in total, describing AND and OR semantics.
>>> While it's solved the router and most mediation function needs =
today, we understand that this is not a complete solution.
>>> Gerhard, in one of his old draft, =
draft-sommer-ipfix-mediator-ext-01, proposed ADTs for orderedList, =
orderedPair, and portRanges which allow the definition new IEs for port =
ranges etc.
>>> Even if this solution 2 is extensible, in the long term, will lead =
to an explosion of IEs: 3 list types * the semantic type, where the =
semantic can be (and, or, orderedList, orderedPair, portRanges, random, =
etc...)
>>>=20
>>> Soltution 3
>>> We propose to add a semantic field in the 3 list types.
>>> Something such as:
>>>=20
>>>     =
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>> |0|               Field ID    |       Element Length            |
>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>> | Semantic  |             BasicList Content ...                 |
>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>> |                           ...                                 |
>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>=20
>>> This semantic field would be a new IANA registry, we could be =
populated initially with NONE, OR, AND, ORDERED, etc... Up to =
discussion.
>>> The advantage of this solution is that it's extensible, and it =
doesn't need new IEs for each semantic.
>>> If ever required (mainly in a mediation function), we can model  =
very complex semantic, eg, "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR =
linecard2", which could be the new observation point from a aggregated =
Flow Record
>>>=20
>>>=20
>>>=20
>>> Conclusion:
>>> After much debate internally, we believe that we should do the =
effort to include this semantic now, and not postpone the problem, which =
will lead to an explosion of IEs in the future.
>>> Personally, I was initially against the solution 3 ... mostly due to =
the effort required to modify the complete specs. =20
>>> We're now ready to modify the specifications in the IPFIX structured =
data, but we would like to get your feedback and agreement in advance as =
this is not a small piece of work.
>>>=20
>>> Please comment.
>>>=20
>>> Regards, Paul, Stan, Gowri, and Benoit.
>>>    =20
>>>=20
>>>> hi Benoit, Gerhard,=20
>>>>=20
>>>> In this case I'm strongly in favor of leaving semantics out of =
structured data. Structured data defines containers. The semantics of =
the containers as a whole and the elements change based upon the =
information elements within the container and within the record =
containing it. Wedging semantics into the structured data elements 1. =
risks further proliferation of (potentially non-interoperable) ways to =
represent the same thing, 2. gives us more ways to represent nonsensical =
things (an OR basicList of MPLS stack entries...means...what?), and 3. =
risks defining an inadequate semantic representation mechanism (what =
about semantics for records not using structured data? what about =
ordered versus unordered sets? what about OR basicList vs AND basicList =
vs nested AND and OR basicLists vs just three identical IEs?). If we =
really want an unambiguous semantic framework for IPFIX (and here I'm =
not convinced either way) that's best done on its own, addressing things =
at the information elem
>>>> ent and record level. Doing it here confuses the issue.=20
>>>>=20
>>>> Regards,=20
>>>>=20
>>>> Brian=20
>>>>=20
>>>> On Nov 10, 2009, at 6:26 AM, Benoit Claise wrote:=20
>>>>=20
>>>>      =20
>>>>=20
>>>>> Gerhard,=20
>>>>>=20
>>>>> Thanks for your email.=20
>>>>> I have no strong feelings about the two solutions you proposed.=20
>>>>>=20
>>>>> >=46rom a pure router point of view,  I don't see any use cases =
for logical OR in exporting flow records.=20
>>>>> However, from a IPFIX Mediator point of view, I see some use =
cases.=20
>>>>> I mean that it requires an Intermediate Aggregation Process or =
Intermediate Correlation Process to express: I've seen this flow record =
OR that flow record.=20
>>>>> Now, it's true that even routers will have mediation functions...=20=

>>>>>=20
>>>>> I'm inclined to add orBasicList, orSubTemplateList, =
orSubTempalteMultiList to the draft (This is a small addition after all) =
and to express that, by default, a logical AND is assumed... =
specifically if the structured data is used in the IPFIX Mediation =
Protocol.=20
>>>>>=20
>>>>> I'm not convinced by the NOT in the context of structured data, as =
we don't even have a concept of NOT for a single information element!=20
>>>>>=20
>>>>> I would like to get some more feedback from others.=20
>>>>>=20
>>>>> Regards, Benoit.=20
>>>>>=20
>>>>>        =20
>>>>>=20
>>>>>> Dear all,=20
>>>>>>=20
>>>>>> Regarding draft-ietf-ipfix-structured-data, I see the risk that =
the=20
>>>>>> semantic of the exported structured data is not clear.=20
>>>>>>=20
>>>>>> How do you interpret the manifold occurrence of the same =
Information=20
>>>>>> Element (basicList) or the same group of Information Elements=20
>>>>>> (subTemplateList) in one record?=20
>>>>>>=20
>>>>>> What does it mean if basicList, subTemplateList, or =
subTemplateMultiList=20
>>>>>> is used for a Flow Key field? Or non-key field?=20
>>>>>>=20
>>>>>> Some Examples:=20
>>>>>>=20
>>>>>> - BasicList of egress interfaces in a Flow Record=20
>>>>>>  How should a Flow Record be interpreted which contains a list of=20=

>>>>>>  egress interfaces and a packet counter?=20
>>>>>>  Has every counted packet been sent on every egress interface?=20
>>>>>>    =3D> multicast case, AND semantic (see example in section 8.1)=20=

>>>>>>  Has every counted packet been sent on any one of the egress=20
>>>>>>  interfaces?=20
>>>>>>    =3D> load balancing case, OR semantic=20
>>>>>>  Can it be used as a Flow Key or not?=20
>>>>>>=20
>>>>>> - BasicList of destination ports in a Flow Record=20
>>>>>>  As every packet has only one destination port, the only =
reasonable=20
>>>>>>  interpretation is that the Flow contains packets having one of=20=

>>>>>>  the reported port numbers.=20
>>>>>>    =3D> OR semantic=20
>>>>>>  This would be a non-key field.=20
>>>>>>=20
>>>>>>=20
>>>>>> I think there are two solutions:=20
>>>>>>=20
>>>>>> 1. We decide that the semantic of list content is out of scope of=20=

>>>>>>   draft-ietf-ipfix-structured-data. We add a note to the draft =
that=20
>>>>>>   the semantic must be clear from the context or the definition =
of the=20
>>>>>>   Information Elements used within the lists.=20
>>>>>>=20
>>>>>> 2. We define semantic lists, such as=20
>>>>>>   - andBasicList, andSubTemplateList, andSubTemplateMultiList=20
>>>>>>   - orBasicList, orSubTemplateList, orSubTempalteMultiList=20
>>>>>>   describing AND and OR semantic of the contained IEs/Templates,=20=

>>>>>>   respectively.=20
>>>>>>=20
>>>>>>=20
>>>>>> As I wrote in an earlier mail, I see a good use case for =
orBasicList. It=20
>>>>>> could be used in the Selector Report Interpretation of Property =
Match=20
>>>>>> Filtering to report a filter like "port 80 or port 443".=20
>>>>>>=20
>>>>>> http://www.ietf.org/mail-archive/web/ipfix/current/msg04856.html
>>>>>> =20
>>>>>> At the moment, the Selector Report Interpretation is limited to =
AND.=20
>>>>>> However, if we also want to express a NOT, we still need another =
solution...=20
>>>>>>=20
>>>>>> Regards,=20
>>>>>> Gerhard=20
>>>>>>=20
>>>>>> _______________________________________________=20
>>>>>> IPFIX mailing list=20
>>>>>>=20
>>>>>> IPFIX@ietf.org
>>>>>> =20
>>>>>>=20
>>>>>> https://www.ietf.org/mailman/listinfo/ipfix
>>>>>> =20
>>>>>>=20
>>>>>>          =20
>>>>>>=20
>>>>> _______________________________________________=20
>>>>> IPFIX mailing list=20
>>>>>=20
>>>>> IPFIX@ietf.org
>>>>> =20
>>>>>=20
>>>>> https://www.ietf.org/mailman/listinfo/ipfix
>>>>> =20
>>>>>        =20
>>>>>=20
>>>    =20
>>  =20
>=20


From andrjohn@cisco.com  Tue Mar 16 14:57:55 2010
Return-Path: <andrjohn@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id D07F23A6824 for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 14:57:55 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -9.266
X-Spam-Level: 
X-Spam-Status: No, score=-9.266 tagged_above=-999 required=5 tests=[AWL=1.333,  BAYES_00=-2.599, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id uDMQOzUQWbsj for <ipfix@core3.amsl.com>; Tue, 16 Mar 2010 14:57:55 -0700 (PDT)
Received: from ams-iport-1.cisco.com (ams-iport-1.cisco.com [144.254.224.140]) by core3.amsl.com (Postfix) with ESMTP id 8B15F3A683F for <ipfix@ietf.org>; Tue, 16 Mar 2010 14:56:37 -0700 (PDT)
Authentication-Results: ams-iport-1.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AkoBAIqbn0uQ/uCWe2dsb2JhbACbBRUBAQsLJAYcoXyYdYR2BA
X-IronPort-AV: E=Sophos;i="4.49,652,1262563200"; d="scan'208";a="58146081"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 16 Mar 2010 21:56:45 +0000
Received: from cisco.com (mrwint.cisco.com [64.103.71.48]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2GLuji4028850; Tue, 16 Mar 2010 21:56:45 GMT
Received: from [10.55.163.37] (ams-andrjohn-8714.cisco.com [10.55.163.37]) by cisco.com (8.11.7p3+Sun/8.8.8) with ESMTP id o2GLuiX14156; Tue, 16 Mar 2010 21:56:44 GMT
Message-ID: <4B9FFE9C.3060908@cisco.com>
Date: Tue, 16 Mar 2010 21:56:44 +0000
From: Andrew Johnson <andrjohn@cisco.com>
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com> <4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com> <4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E06B4.6090203@net.in.tum.de>
In-Reply-To: <4B9E06B4.6090203@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 16 Mar 2010 21:57:55 -0000

Hi Gerhard

Gerhard Muenz wrote:
[SNIP]
> To make it clearer, I will add an explicit note that the counters in the 
> Selector class correspond to the counters in the IPFIX MIB.
> 
> Just like in the MIB specification, we do not specify how the counters 
> are related to the PSAMP statistics reports.

I think the current configuration model has a few problems:
   * Implementations seem forced to consider more than one cache part
     of the same Metering Process to be able to use the cascading
     selection configuration.
   * The "SelectionSequenceID" can not be a property of a
     "SelectionProcess Class" because that class does not map to a
     Selection Sequence (i.e. OP, selectors, Metering Process).
   * Selectors can not be re-used fully across Selection Sequences as
     they are done in the PSAMP model (so I don't think you should have
     a "SelectorID" property either).


The PSAMP model is to define selectors as simply configuration.  There 
are no statistics on a per Selector basis.  Instead, each use of a 
selector needs separate statistics.  From RFC 5476, section 6.5.1:

  Example of two Selection Sequences:

    Selection Sequence 7 (Filter->Sampling):
      ingressInterface    5
      selectorId          5 (Filter, match IPV4SourceAddress 192.0.2.1)
      selectorId         10 (Sampler, Random 1 out-of ten)

    Selection Sequence 9 (Sampling->Filtering):
      ingressInterface    5
      selectorId         10 (Sampler, Random 1 out-of ten)
      selectorId          5 (Filter, match IPV4SourceAddress 192.0.2.1)


If we're not going to map the PSAMP model in the configuration model 
then I think we should separate them completely.  In fact, I don't think 
any attempt should be made to directly configure a "Selection Process", 
instead the selection configuration is provided and the Exporting Device 
can sort it out.

This simply avoids the question about whether a Metering Device can (or 
should) have more than one cache.  It also allows greater flexibility in 
configuration because a Selector Configuration can have more than one 
Packet Stream as input.

I think this is a useful concept.  Imagine this relatively common 
configuration, where we want to monitor 10 interfaces, doing 1 in 100 
sampling for each.  Rather than define a new sampling selector for each 
interface, wouldn't it be better to simply do?:

   Selector MY_SAMPLER: sampling 1 in 100 -> cache: MY_CACHE
   OP A -> MY_SAMPLER
   OP B -> MY_SAMPLER
   ...

This is closer to the PSAMP model where you would do something like:

   Selector MY_SAMPLER: sampling 1 in 100

   Selection Sequence: OP A, MY_SAMPLER, MY_CACHE
   Selection Sequence: OP B, MY_SAMPLER, MY_CACHE
   ...


Cheers

Andrew


From muenz@net.in.tum.de  Wed Mar 17 02:58:04 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id A444E3A6A13 for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 02:58:04 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.588
X-Spam-Level: 
X-Spam-Status: No, score=-1.588 tagged_above=-999 required=5 tests=[AWL=-0.469, BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id yZc1066J+J0Y for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 02:58:03 -0700 (PDT)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 3242E3A695A for <ipfix@ietf.org>; Wed, 17 Mar 2010 02:57:58 -0700 (PDT)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 0EC6C2013E31; Wed, 17 Mar 2010 10:58:07 +0100 (CET)
Message-ID: <4BA0A7C8.9010205@net.in.tum.de>
Date: Wed, 17 Mar 2010 10:58:32 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Andrew Johnson <andrjohn@cisco.com>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com> <4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com> <4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E06B4.6090203@net.in.tum.de> <4B9FFE9C.3060908@cisco.com>
In-Reply-To: <4B9FFE9C.3060908@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms090603020304050605060601"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 17 Mar 2010 09:58:04 -0000

This is a cryptographically signed message in MIME format.

--------------ms090603020304050605060601
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable


Hi Andrew,

Andrew Johnson wrote:
> Hi Gerhard
>=20
> Gerhard Muenz wrote:
> [SNIP]
>> To make it clearer, I will add an explicit note that the counters in t=
he=20
>> Selector class correspond to the counters in the IPFIX MIB.
>>
>> Just like in the MIB specification, we do not specify how the counters=
=20
>> are related to the PSAMP statistics reports.
>=20
> I think the current configuration model has a few problems:
>    * Implementations seem forced to consider more than one cache part
>      of the same Metering Process to be able to use the cascading
>      selection configuration.

Not at all!

As I have written multiple times here on the list, no implementation is=20
expected to support the entire configuration data model. That is just=20
impossible. Every implementation will have restriction at one end or the =

other. This is not a specific problem to the Selection Process or Cache=20
at all.

YANG allows you to specify restrictions and extensions in a formal way.

If you do not think that this becomes clear from the text in the draft,=20
then please provide some better text.

>    * The "SelectionSequenceID" can not be a property of a
>      "SelectionProcess Class" because that class does not map to a
>      Selection Sequence (i.e. OP, selectors, Metering Process).

No. The Selection Process may be composed of multiple Selectors.=20
Therefore, the Selection Process represents a Selection Sequence and has =

its Selection Sequence ID as state parameter.

>    * Selectors can not be re-used fully across Selection Sequences as
>      they are done in the PSAMP model (so I don't think you should have=

>      a "SelectorID" property either).

You have always stressed that every Selection Sequence has its own=20
Selector instances with separate selection statistics and that only a=20
Selector configuration template can be shared.

This is reflected in the model!

The data model specifies instances, no configuration templates (again,=20
I'm repeating myself). So, if you need the same Selector config in two=20
Selection Sequences, you copy the data.

Just like the Selection Sequence ID, the Selector ID is a state=20
parameter assigned by the Monitoring Device. It gives the link to the=20
information exported via PSAMP protocol. Therefore it is useful to have=20
it in the configuration data model.

The Monitoring Device may assign the same Selector ID for two Selector=20
instances if the config is identical. There is no problem. Obviously,=20
the Monitoring Device needs to compare the configurations of the=20
different instances.

> The PSAMP model is to define selectors as simply configuration.  There =

> are no statistics on a per Selector basis.  Instead, each use of a=20
> selector needs separate statistics.  From RFC 5476, section 6.5.1:
>=20
>   Example of two Selection Sequences:
>=20
>     Selection Sequence 7 (Filter->Sampling):
>       ingressInterface    5
>       selectorId          5 (Filter, match IPV4SourceAddress 192.0.2.1)=

>       selectorId         10 (Sampler, Random 1 out-of ten)
>=20
>     Selection Sequence 9 (Sampling->Filtering):
>       ingressInterface    5
>       selectorId         10 (Sampler, Random 1 out-of ten)
>       selectorId          5 (Filter, match IPV4SourceAddress 192.0.2.1)=

>=20

What if these are two different Observation Domains?

    * Selector ID

       The Selector ID is the unique ID identifying a Primitive Selector.=

       The ID is unique within the Observation Domain.

It does not say: unique per PSAMP Device.

Configuration templates for Selectors would introduce another level of=20
indirection to the model. This will complicate things a bit.

If the Selector ID would be a unique state parameter of the=20
configuration template, then you force Selector IDs to be unique per=20
Monitoring Device. Or you must disallow reusing Selector configuration=20
templates across Observation Domains. Formally specifying such=20
restrictions in YANG results in complicated XPath expresions (or is not=20
possible at all).

Though, if you insist on configuring templates for Selectors instead of=20
instances, we can change the model accordingly.

However, this decision will have no effect on the question whether one=20
Selection Process can feed multiple Caches.

> If we're not going to map the PSAMP model in the configuration model=20
> then I think we should separate them completely.  In fact, I don't thin=
k=20
> any attempt should be made to directly configure a "Selection Process",=
=20
> instead the selection configuration is provided and the Exporting Devic=
e=20
> can sort it out.
>=20
> This simply avoids the question about whether a Metering Device can (or=
=20
> should) have more than one cache.  It also allows greater flexibility i=
n=20
> configuration because a Selector Configuration can have more than one=20
> Packet Stream as input.

The questions whether the Selector configuration is organized with=20
reusable configuration templates and whether the same Selection Process=20
may feed two Caches are completely orthogonal.

You must not mix up these two questions!

> I think this is a useful concept.  Imagine this relatively common=20
> configuration, where we want to monitor 10 interfaces, doing 1 in 100=20
> sampling for each.  Rather than define a new sampling selector for each=
=20
> interface, wouldn't it be better to simply do?:
>=20
>    Selector MY_SAMPLER: sampling 1 in 100 -> cache: MY_CACHE
>    OP A -> MY_SAMPLER
>    OP B -> MY_SAMPLER
>    ...

What do you mean with "simple"?
You think that you do not have to copy some lines of XML?
I assume that configuration management tools can do this work for you.

> This is closer to the PSAMP model where you would do something like:
>=20
>    Selector MY_SAMPLER: sampling 1 in 100
>=20
>    Selection Sequence: OP A, MY_SAMPLER, MY_CACHE
>    Selection Sequence: OP B, MY_SAMPLER, MY_CACHE
>    ...

Do you also request configuration templates for Caches? Or is this a=20
single instance?

Regards,
Gerhard

>=20
>=20
> Cheers
>=20
> Andrew
>=20

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms090603020304050605060601
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms090603020304050605060601--

From bclaise@cisco.com  Wed Mar 17 04:43:42 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id DEF1A3A6825 for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 04:43:42 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.732
X-Spam-Level: 
X-Spam-Status: No, score=-1.732 tagged_above=-999 required=5 tests=[AWL=-0.264, BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, HTML_MESSAGE=0.001]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id NVs7wZUzGdXb for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 04:43:41 -0700 (PDT)
Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id F0B633A6358 for <ipfix@ietf.org>; Wed, 17 Mar 2010 04:43:40 -0700 (PDT)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2HBXFhx001098 for <ipfix@ietf.org>; Wed, 17 Mar 2010 12:33:15 +0100 (CET)
Received: from [10.55.43.57] (ams-bclaise-8718.cisco.com [10.55.43.57]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2HBXExH026622 for <ipfix@ietf.org>; Wed, 17 Mar 2010 12:33:15 +0100 (CET)
Message-ID: <4BA0BDFA.80808@cisco.com>
Date: Wed, 17 Mar 2010 12:33:14 +0100
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: "ipfix@ietf.org" <ipfix@ietf.org>
Content-Type: multipart/alternative; boundary="------------040301020506030504020601"
Subject: [IPFIX] Fwd:  I-D ACTION:draft-ietf-ipfix-structured-data-01.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 17 Mar 2010 11:43:43 -0000

This is a multi-part message in MIME format.
--------------040301020506030504020601
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Dear all,

Next to some editorial changes, this new draft version contains an extra 
example illustrating how to encode a variable-length interface name with 
the BasicList.
The only open issue is the addition of the semantic in the structured 
data, currently discussed in a different email thread.
We would like to get closure on this remaining open issue by this IETF 
meeting. Please share your thoughts via email, specifically if you won't 
attend the IETF in person.

Regards, Benoit.

-------- Original Message --------
Subject: 	[IPFIX] I-D ACTION:draft-ietf-ipfix-structured-data-01.txt
Date: 	Mon, 8 Mar 2010 10:30:02 -0800 (PST)
From: 	Internet-Drafts@ietf.org
To: 	i-d-announce@ietf.org
CC: 	ipfix@ietf.org



A New Internet-Draft is available from the on-line Internet-Drafts
directories.
This draft is a work item of the IP Flow Information Export Working Group of the IETF.

	Title		: Export of Structured Data in IPFIX
	Author(s)	: B. Claise, G. Dhandapani, S. Yates, P. Aitken
	Filename	: draft-ietf-ipfix-structured-data-01.txt
	Pages		: 58
	Date		: 2010-3-8
	
This document specifies an extension to IP Flow Information
         eXport (IPFIX) protocol specification in [RFC5101] and the IPFIX
         information model specified in [RFC5102] to support hierarchical
         structured data and lists (sequences) of Information Elements in
         data records.  This extension allows definition of complex data
         structures such as variable-length lists and specification of
         hierarchical containment relationships between Templates.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-ipfix-structured-data-01.txt

Internet-Drafts are also available by anonymous FTP at:
ftp://ftp.ietf.org/internet-drafts/

Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.



--------------040301020506030504020601
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>

<meta http-equiv="content-type" content="text/html; charset=ISO-8859-1">
</head>
<body bgcolor="#ffffff" text="#000000">
Dear all,<br>
<br>
Next to some editorial changes, this new draft version contains an
extra example illustrating how to encode a variable-length interface
name with the BasicList.<br>
The only open issue is the addition of the semantic in the structured
data, currently discussed in a different email thread.<br>
We would like to get closure on this remaining open issue by this IETF
meeting. Please share your thoughts via email, specifically if you
won't attend the IETF in person.<br>
<br>
Regards, Benoit.<br>
<br>
-------- Original Message --------
<table class="moz-email-headers-table" border="0" cellpadding="0"
 cellspacing="0">
  <tbody>
    <tr>
      <th align="RIGHT" valign="BASELINE" nowrap="nowrap">Subject: </th>
      <td>[IPFIX] I-D ACTION:draft-ietf-ipfix-structured-data-01.txt</td>
    </tr>
    <tr>
      <th align="RIGHT" valign="BASELINE" nowrap="nowrap">Date: </th>
      <td>Mon, 8 Mar 2010 10:30:02 -0800 (PST)</td>
    </tr>
    <tr>
      <th align="RIGHT" valign="BASELINE" nowrap="nowrap">From: </th>
      <td><a class="moz-txt-link-abbreviated" href="mailto:Internet-Drafts@ietf.org">Internet-Drafts@ietf.org</a></td>
    </tr>
    <tr>
      <th align="RIGHT" valign="BASELINE" nowrap="nowrap">To: </th>
      <td><a class="moz-txt-link-abbreviated" href="mailto:i-d-announce@ietf.org">i-d-announce@ietf.org</a></td>
    </tr>
    <tr>
      <th align="RIGHT" valign="BASELINE" nowrap="nowrap">CC: </th>
      <td><a class="moz-txt-link-abbreviated" href="mailto:ipfix@ietf.org">ipfix@ietf.org</a></td>
    </tr>
  </tbody>
</table>
<br>
<br>
<pre>A New Internet-Draft is available from the on-line Internet-Drafts 
directories.
This draft is a work item of the IP Flow Information Export Working Group of the IETF.

	Title		: Export of Structured Data in IPFIX
	Author(s)	: B. Claise, G. Dhandapani, S. Yates, P. Aitken
	Filename	: draft-ietf-ipfix-structured-data-01.txt
	Pages		: 58
	Date		: 2010-3-8
	
This document specifies an extension to IP Flow Information 
        eXport (IPFIX) protocol specification in [RFC5101] and the IPFIX 
        information model specified in [RFC5102] to support hierarchical 
        structured data and lists (sequences) of Information Elements in 
        data records.  This extension allows definition of complex data 
        structures such as variable-length lists and specification of 
        hierarchical containment relationships between Templates.

A URL for this Internet-Draft is:
<a class="moz-txt-link-freetext" href="http://www.ietf.org/internet-drafts/draft-ietf-ipfix-structured-data-01.txt">http://www.ietf.org/internet-drafts/draft-ietf-ipfix-structured-data-01.txt</a>

Internet-Drafts are also available by anonymous FTP at:
<a class="moz-txt-link-freetext" href="ftp://ftp.ietf.org/internet-drafts/">ftp://ftp.ietf.org/internet-drafts/</a>

Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

</pre>
</body>
</html>

--------------040301020506030504020601--

From bclaise@cisco.com  Wed Mar 17 05:29:11 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 537643A6927 for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 05:29:11 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.712
X-Spam-Level: 
X-Spam-Status: No, score=-1.712 tagged_above=-999 required=5 tests=[AWL=-0.243, BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id krb3y4OmCHuz for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 05:29:09 -0700 (PDT)
Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id 4D9553A68FA for <ipfix@ietf.org>; Wed, 17 Mar 2010 05:29:09 -0700 (PDT)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2HBqHlp007471; Wed, 17 Mar 2010 12:52:17 +0100 (CET)
Received: from [10.55.43.57] (ams-bclaise-8718.cisco.com [10.55.43.57]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2HBqDUq018129; Wed, 17 Mar 2010 12:52:13 +0100 (CET)
Message-ID: <4BA0C26D.9070901@cisco.com>
Date: Wed, 17 Mar 2010 12:52:13 +0100
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: Brian Trammell <trammell@tik.ee.ethz.ch>
References: <4AF73525.8050009@net.in.tum.de> <4AF8F999.3000207@cisco.com>	<F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>	<4B9E33BE.7070907@cisco.com>	<7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch>	<4B9F520E.8060507@cisco.com> <DB6E59D9-B373-4919-BB58-00EB26014564@tik.ee.ethz.ch>
In-Reply-To: <DB6E59D9-B373-4919-BB58-00EB26014564@tik.ee.ethz.ch>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 17 Mar 2010 12:29:11 -0000

Hi Brian,
> Hi, Benoit,
>
> Replies inline...
>
> On Mar 16, 2010, at 2:40 AM, Benoit Claise wrote:
>
>   =20
>> Hi Brian,
>>
>> Thanks for your feedback.
>>     =20
>>> hi, Benoit, all...
>>>
>>> Agreed that we should do something about this (i.e., that solution 1 =
is no solution.); that said, a few comments in no particular order:
>>>
>>> 1. In considering adding explicit semantics to structured data, we as=
 a WG are taking on the task of defining semantics for IPFIX as a whole. =
Semantics, as I understand them, are in IPFIX largely contextual and temp=
late dependent, but in almost all cases I can think of it seems like thes=
e are implicitly "AND" semnantics (this flow has source IP A _AND_ destin=
ation IP B _AND_...).
>>>
>>>       =20
>> Agreed.
>>     =20
>>> We will need to make an explicit statement on this.
>>>       =20
>> Not sure why.
>>     =20
>>> We will need to determine whether these implicit semantics are a prop=
erty of the protocol (in which case Structured Data is really a protocol-=
level extension), or a property of each information element (in which cas=
e all 5103 IEs have implicit AND semantics; question 2: will we need to a=
dd semantics to the IANA registry in this case?). We will need to be quit=
e careful about this. It's not as simple as defining semantics within str=
uctured data then calling it done.
>>>
>>>
>>>       =20
>> I'm not sure why each individual IE should have a semantic ... in the =
IANA registry.
>>     =20
> I'm not either. On reflection it seems like overkill. I'm just saying t=
hat if we, as a WG, are moving from stating that semantics are explicitly=
 out of scope to defining them as in-scope, we need to have a consistent =
approach, and answers to all the questions that arise when we consider mo=
ving the protocol from a simple framing mechanism to a framing mechanism =
with some logic behind it, so we answer them once, so that all future eff=
orts having to do with semantics are consistent. This, I acknowledge, is =
an argument in favor of solution 3...)
>
> One very simple new question that arises here, to illustrate my point: =
Is it legal to export a record that has sourceIPAddress X AND NOT sourceI=
PAddress X?
>   =20
 From a protocol point of view, yes
 From a semantic point of view, I don't see a use case for that.
Now, this question is not different that: with RFC5101,  is it legal to=20
export a record that has two instances of sourceIPAddress?
>   =20
>> As you wrote, "in almost all cases I can think of it seems like these =
are implicitly "AND" semnantics", so the case we try to solve is when the=
re are multiple instances of a single IE. We know that RFC 5101 foresaw t=
hat case " The Collector MUST support the use of Templates containing mul=
tiple occurrences of the similar Information Elements", but the idea is n=
ot to change RFC5101.  If we put some semantic in the IPFIX structured da=
ta, that would solve the vast majority of our cases. Also, we could say: =
if you want some semantic when exporting multiple IEs, then you SHOULD us=
e IPFIX structured data.
>>
>> Also, the default value for the semantic field in the IPFIX structured=
 data SHOULD be "NONE", to express that flow record doesn't include any s=
emantic.... like in RFC5101. You might draw your own conclusion, maybe be=
cause you know your network, maybe because you have configured the export=
er, but then it's your decision.
>> The way I see the proposed solution is: in IPFIX structured data, you =
MAY use the semantic field as a way to express the relationship between I=
Es within the structure.
>>     =20
>>> 2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid argume=
nt against Solution 2, as it's trying to solve a somewhat different and m=
ore limited problem than structured data. Solution 2 _might_ cause a prob=
lem for this draft, but certainly not the other way around (unless we as =
a WG want to subsume that work into this draft, which would probably requ=
ire rechartering...). Also, I don't think we need semantics for all of th=
e list types, just basicList (Illustrative question: How do I meaningfull=
y interpret two "or" subTemplateMultiLists with disjoint IE sets? Two "an=
d" subTemplateMultiLists? Nestings thereof?)... So, we don't really have =
an explosion to deal with if we do Solution 2 correctly: andBasicList, or=
BasicList, xorBasicList, notBasicList; Four IEs, nestable, done. What can=
't we do with those four IEs? In this case, we could even step back and s=
ay that semantics outside these four are within the protocol explicitly _=
undefined_, and to be interpreted within the
>>>   context of each Template.
>>>
>>>
>>>       =20
>> If I translate some more what I wrote "While it's solved the router an=
d most mediation function needs today", this would be "Me, myself, and I =
don't need more that OR and AND now ;-)".
>> However, who am I to tell that others don't need it now... and that th=
e logical solution is to use the IPFIX structured data
>> Furthermore, if we think a little bit longer term, the next big step i=
n IPFIX is the mediation function. In my company, every features want to =
export his own data with NetFlow/IPFIX... up to the point where a CPE wou=
ld not have enough bandwidth across the WAN to export all the "management=
" information. So we'll need more and more of aggregated flow records (bo=
th in time and space) even in the router. Again, the logical solution wil=
l be to use the IPFIX structured data. At this point, we will most probab=
ly need something else than OR and AND, i.e. RANGE, ORDERRED, etc...
>>
>> An example of subTemplateMultiLists with disjoint IE sets, let's imagi=
ne that you have to export an aggregated observation point, composed of m=
ultiple template records
>>      template 1: exporterIPaddress
>>      template 2: exporterIPaddress, basicList of interfaces
>>      template 3: exporterIPaddress, LC
>>     =20
> and then you'd want to OR these... Okay, makes sense...
>
>   =20
>>     =20
>>> 3. If we really _do_ want ranges and so on (which, again, we'd need t=
o get WG consensus on; this is explicitly out of scope in my reading of t=
he present charter), then we could do them in the scope of Solution 3.
>>>       =20
>> Btw, acknowledging that one day we will have to solve this is good eno=
ugh for solution 3. I mean, we don't have to populate the semantic IANA n=
ow... even if that would be more efficient.
>>     =20
>>> However, this seems a little not-quite-fleshed-out-enough for me to s=
ay whether I like it or not. Could you present an example of how you woul=
d use the proposed Semantic field to model your example? "(eth1 OR eth2) =
AND (NOT (eth3 OR eth4)) OR linecard2"
>>>
>>> (FWIW, I would do this with my proposal to Solution 2 as follows:)
>>>
>>> (orBasicList (andBasicList (orBasicList ingressInterface eth1 eth2) (=
notBasicList (orBasicList ingressInterface eth3 eth4)) (andBasicList line=
CardID 2))
>>>
>>>
>>>       =20
>> (BasicList, OR, (basicList, AND, (basicList, OR, eth1, eth2), (basicLi=
st, NOR, eth3, eth4)), (basicList, NONE, linecard2)))
>>     =20
> Hm. Okay. This makes sense. A couple of questions then, about solution =
3:
>
> 1. Would we want to try and bitfield this, to define the semantics we _=
know_ we have, then leave the rest of it reserved? Something like:
>
> MSb                      LSb
> +---+------+---------------+
> | ! | multi| reserved      |
> +---+------+---------------+
>
> ! =3D negate sense of semantics if 1 (this is the NOT flag)
> multi (multiplicity) =3D 00: undefined, 01: or/oneOrMore, 10: xor/exact=
lyOne, 11: and/exactlyAll
> reserved =3D place for adding other bells and whistles like RANGE and s=
o on in the future.
>   =20
Maybe we want to start by asking the question: which semantic do we need =

now?
Is a need for NONE, OR, AND, ORDERED for now. Anything else?
> 2. Are we sure we want to do this in one byte? If we do bitfielding as =
above this gives us 32 possible extensions, which seems like _way_ more t=
han enough, but does stick another odd offset in there, which slows thing=
s down on machines that need aligned access. Probably one byte is okay an=
d we let the implementation use paddingOctets and set padding to fix this=
=2E..
>   =20
With structured data, is the alignment still important? When I look at=20
the examples throughout the draft... As you wrote, paddingOctets might=20
be the solution in this case.
> 3. Would it make sense maybe to have two sets of structured data elemen=
ts, one with the semantics byte, and one without (which is then explicitl=
y undefined)? Then exporters who don't need it don't have to bother stick=
ing an extra odd-aligned zero in the stream for every list.
>
> I'm sure I'l have more questions, but none come to mind now... But it s=
eems like we're converging on the least-unnecessarily-complex solution he=
re, which is good. :)
>   =20
Happy about that. ;-)
Note: I was envisioning even something more simpler: one byte,=20
containing all the semantic possibilities, administered by IANA.  So no=20
reserved, no !

Regards, Benoit.
> Best regards,
>
> Brian
>
>   =20
>>> On Mar 15, 2010, at 6:18 AM, Benoit Claise wrote:
>>>
>>>
>>>
>>>       =20
>>>> Dear all,
>>>>
>>>> We've been thinking about this one, and we see 3 solutions. We belie=
ve that the solution 3 is the way to go, as we show below.
>>>>
>>>> Solution 1.
>>>> Consider that the semantic is out of scope for this document.
>>>> This is the easiest solution.  However, we understand it's not right=
: we would not do a complete job wrt to IPFIX structured data
>>>> As Gerhard was expressing, the collector has not clue how to treat t=
he example of a BasicList of egress interfaces in a Flow Record.
>>>>      - Has every counted packet been sent on every egress interface?=

>>>>        =3D>  multicast case, AND semantic
>>>>      - Has every counted packet been sent on any one of the egress i=
nterfaces?
>>>>       =3D>  load balancing case, OR semantic
>>>>
>>>> Soltution 2
>>>> We could focus on the logical AND and OR semantics only , by definin=
g semantic lists, such as andBasicList, andSubTemplateList, andSubTemplat=
eMultiList, orBasicList, orSubTemplateList, and orSubTemplateMultiList
>>>> So 6 I.E.s in total, describing AND and OR semantics.
>>>> While it's solved the router and most mediation function needs today=
, we understand that this is not a complete solution.
>>>> Gerhard, in one of his old draft, draft-sommer-ipfix-mediator-ext-01=
, proposed ADTs for orderedList, orderedPair, and portRanges which allow =
the definition new IEs for port ranges etc.
>>>> Even if this solution 2 is extensible, in the long term, will lead t=
o an explosion of IEs: 3 list types * the semantic type, where the semant=
ic can be (and, or, orderedList, orderedPair, portRanges, random, etc...)=

>>>>
>>>> Soltution 3
>>>> We propose to add a semantic field in the 3 list types.
>>>> Something such as:
>>>>
>>>>      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+=
-+
>>>> |0|               Field ID    |       Element Length            |
>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>> | Semantic  |             BasicList Content ...                 |
>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>> |                           ...                                 |
>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>
>>>> This semantic field would be a new IANA registry, we could be popula=
ted initially with NONE, OR, AND, ORDERED, etc... Up to discussion.
>>>> The advantage of this solution is that it's extensible, and it doesn=
't need new IEs for each semantic.
>>>> If ever required (mainly in a mediation function), we can model  ver=
y complex semantic, eg, "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR linec=
ard2", which could be the new observation point from a aggregated Flow Re=
cord
>>>>
>>>>
>>>>
>>>> Conclusion:
>>>> After much debate internally, we believe that we should do the effor=
t to include this semantic now, and not postpone the problem, which will =
lead to an explosion of IEs in the future.
>>>> Personally, I was initially against the solution 3 ... mostly due to=
 the effort required to modify the complete specs.
>>>> We're now ready to modify the specifications in the IPFIX structured=
 data, but we would like to get your feedback and agreement in advance as=
 this is not a small piece of work.
>>>>
>>>> Please comment.
>>>>
>>>> Regards, Paul, Stan, Gowri, and Benoit.
>>>>
>>>>
>>>>         =20
>>>>> hi Benoit, Gerhard,
>>>>>
>>>>> In this case I'm strongly in favor of leaving semantics out of stru=
ctured data. Structured data defines containers. The semantics of the con=
tainers as a whole and the elements change based upon the information ele=
ments within the container and within the record containing it. Wedging s=
emantics into the structured data elements 1. risks further proliferation=
 of (potentially non-interoperable) ways to represent the same thing, 2. =
gives us more ways to represent nonsensical things (an OR basicList of MP=
LS stack entries...means...what?), and 3. risks defining an inadequate se=
mantic representation mechanism (what about semantics for records not usi=
ng structured data? what about ordered versus unordered sets? what about =
OR basicList vs AND basicList vs nested AND and OR basicLists vs just thr=
ee identical IEs?). If we really want an unambiguous semantic framework f=
or IPFIX (and here I'm not convinced either way) that's best done on its =
own, addressing things at the information elem
>>>>> ent and record level. Doing it here confuses the issue.
>>>>>
>>>>> Regards,
>>>>>
>>>>> Brian
>>>>>
>>>>> On Nov 10, 2009, at 6:26 AM, Benoit Claise wrote:
>>>>>
>>>>>
>>>>>
>>>>>           =20
>>>>>> Gerhard,
>>>>>>
>>>>>> Thanks for your email.
>>>>>> I have no strong feelings about the two solutions you proposed.
>>>>>>
>>>>>> > From a pure router point of view,  I don't see any use cases for=
 logical OR in exporting flow records.
>>>>>> However, from a IPFIX Mediator point of view, I see some use cases=
=2E
>>>>>> I mean that it requires an Intermediate Aggregation Process or Int=
ermediate Correlation Process to express: I've seen this flow record OR t=
hat flow record.
>>>>>> Now, it's true that even routers will have mediation functions...
>>>>>>
>>>>>> I'm inclined to add orBasicList, orSubTemplateList, orSubTempalteM=
ultiList to the draft (This is a small addition after all) and to express=
 that, by default, a logical AND is assumed... specifically if the struct=
ured data is used in the IPFIX Mediation Protocol.
>>>>>>
>>>>>> I'm not convinced by the NOT in the context of structured data, as=
 we don't even have a concept of NOT for a single information element!
>>>>>>
>>>>>> I would like to get some more feedback from others.
>>>>>>
>>>>>> Regards, Benoit.
>>>>>>
>>>>>>
>>>>>>
>>>>>>             =20
>>>>>>> Dear all,
>>>>>>>
>>>>>>> Regarding draft-ietf-ipfix-structured-data, I see the risk that t=
he
>>>>>>> semantic of the exported structured data is not clear.
>>>>>>>
>>>>>>> How do you interpret the manifold occurrence of the same Informat=
ion
>>>>>>> Element (basicList) or the same group of Information Elements
>>>>>>> (subTemplateList) in one record?
>>>>>>>
>>>>>>> What does it mean if basicList, subTemplateList, or subTemplateMu=
ltiList
>>>>>>> is used for a Flow Key field? Or non-key field?
>>>>>>>
>>>>>>> Some Examples:
>>>>>>>
>>>>>>> - BasicList of egress interfaces in a Flow Record
>>>>>>>   How should a Flow Record be interpreted which contains a list o=
f
>>>>>>>   egress interfaces and a packet counter?
>>>>>>>   Has every counted packet been sent on every egress interface?
>>>>>>>     =3D>  multicast case, AND semantic (see example in section 8.=
1)
>>>>>>>   Has every counted packet been sent on any one of the egress
>>>>>>>   interfaces?
>>>>>>>     =3D>  load balancing case, OR semantic
>>>>>>>   Can it be used as a Flow Key or not?
>>>>>>>
>>>>>>> - BasicList of destination ports in a Flow Record
>>>>>>>   As every packet has only one destination port, the only reasona=
ble
>>>>>>>   interpretation is that the Flow contains packets having one of
>>>>>>>   the reported port numbers.
>>>>>>>     =3D>  OR semantic
>>>>>>>   This would be a non-key field.
>>>>>>>
>>>>>>>
>>>>>>> I think there are two solutions:
>>>>>>>
>>>>>>> 1. We decide that the semantic of list content is out of scope of=

>>>>>>>    draft-ietf-ipfix-structured-data. We add a note to the draft t=
hat
>>>>>>>    the semantic must be clear from the context or the definition =
of the
>>>>>>>    Information Elements used within the lists.
>>>>>>>
>>>>>>> 2. We define semantic lists, such as
>>>>>>>    - andBasicList, andSubTemplateList, andSubTemplateMultiList
>>>>>>>    - orBasicList, orSubTemplateList, orSubTempalteMultiList
>>>>>>>    describing AND and OR semantic of the contained IEs/Templates,=

>>>>>>>    respectively.
>>>>>>>
>>>>>>>
>>>>>>> As I wrote in an earlier mail, I see a good use case for orBasicL=
ist. It
>>>>>>> could be used in the Selector Report Interpretation of Property M=
atch
>>>>>>> Filtering to report a filter like "port 80 or port 443".
>>>>>>>
>>>>>>> http://www.ietf.org/mail-archive/web/ipfix/current/msg04856.html
>>>>>>>
>>>>>>> At the moment, the Selector Report Interpretation is limited to A=
ND.
>>>>>>> However, if we also want to express a NOT, we still need another =
solution...
>>>>>>>
>>>>>>> Regards,
>>>>>>> Gerhard
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> IPFIX mailing list
>>>>>>>
>>>>>>> IPFIX@ietf.org
>>>>>>>
>>>>>>>
>>>>>>> https://www.ietf.org/mailman/listinfo/ipfix
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>               =20
>>>>>> _______________________________________________
>>>>>> IPFIX mailing list
>>>>>>
>>>>>> IPFIX@ietf.org
>>>>>>
>>>>>>
>>>>>> https://www.ietf.org/mailman/listinfo/ipfix
>>>>>>
>>>>>>
>>>>>>
>>>>>>             =20
>>>>
>>>>         =20
>>>
>>>       =20
>>     =20
>   =20



From muenz@net.in.tum.de  Wed Mar 17 05:55:51 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 5F71D3A6A3B for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 05:55:51 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.57
X-Spam-Level: 
X-Spam-Status: No, score=-1.57 tagged_above=-999 required=5 tests=[AWL=-0.451,  BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id siQc1v7oUiIB for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 05:55:50 -0700 (PDT)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id C15773A6A22 for <ipfix@ietf.org>; Wed, 17 Mar 2010 05:55:49 -0700 (PDT)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 07BE82010C77; Wed, 17 Mar 2010 13:55:58 +0100 (CET)
Message-ID: <4BA0D177.1070506@net.in.tum.de>
Date: Wed, 17 Mar 2010 13:56:23 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.23 (Windows/20090812)
MIME-Version: 1.0
To: Benoit Claise <bclaise@cisco.com>
References: <4AF73525.8050009@net.in.tum.de> <4AF8F999.3000207@cisco.com>	<F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>	<4B9E33BE.7070907@cisco.com>	<7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch>	<4B9F520E.8060507@cisco.com> <DB6E59D9-B373-4919-BB58-00EB26014564@tik.ee.ethz.ch> <4BA0C26D.9070901@cisco.com>
In-Reply-To: <4BA0C26D.9070901@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms050602000205040806020403"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 17 Mar 2010 12:55:51 -0000

This is a cryptographically signed message in MIME format.

--------------ms050602000205040806020403
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit


Hi all,

Some general thoughts from my side:

- I appreciate that you want to add a basic notion of semantic to the
   structured data.

- Up to now, semantic was not in the protocol but in the info model.

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|0|               Field ID    |       Element Length            |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Semantic  |             BasicList Content ...                 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                           ...                                 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   Can't we encode Semantic in an IE?
   Then, we could do without a new IANA registry.
   And: This IE could be used for other purposes (e.g., in Templates)

- I'm not sure whether the bitvector encoding has an advantage over an
   enumeration type.

Unfortunately, I do not have the time to participate in a deep 
discussion - I'm busy with IPFIX-CONFIG and other stuff.

Regards,
Gerhard

Benoit Claise wrote:
> Hi Brian,
>> Hi, Benoit,
>>
>> Replies inline...
>>
>> On Mar 16, 2010, at 2:40 AM, Benoit Claise wrote:
>>
>>    
>>> Hi Brian,
>>>
>>> Thanks for your feedback.
>>>      
>>>> hi, Benoit, all...
>>>>
>>>> Agreed that we should do something about this (i.e., that solution 1 is no solution.); that said, a few comments in no particular order:
>>>>
>>>> 1. In considering adding explicit semantics to structured data, we as a WG are taking on the task of defining semantics for IPFIX as a whole. Semantics, as I understand them, are in IPFIX largely contextual and template dependent, but in almost all cases I can think of it seems like these are implicitly "AND" semnantics (this flow has source IP A _AND_ destination IP B _AND_...).
>>>>
>>>>        
>>> Agreed.
>>>      
>>>> We will need to make an explicit statement on this.
>>>>        
>>> Not sure why.
>>>      
>>>> We will need to determine whether these implicit semantics are a property of the protocol (in which case Structured Data is really a protocol-level extension), or a property of each information element (in which case all 5103 IEs have implicit AND semantics; question 2: will we need to add semantics to the IANA registry in this case?). We will need to be quite careful about this. It's not as simple as defining semantics within structured data then calling it done.
>>>>
>>>>
>>>>        
>>> I'm not sure why each individual IE should have a semantic ... in the IANA registry.
>>>      
>> I'm not either. On reflection it seems like overkill. I'm just saying that if we, as a WG, are moving from stating that semantics are explicitly out of scope to defining them as in-scope, we need to have a consistent approach, and answers to all the questions that arise when we consider moving the protocol from a simple framing mechanism to a framing mechanism with some logic behind it, so we answer them once, so that all future efforts having to do with semantics are consistent. This, I acknowledge, is an argument in favor of solution 3...)
>>
>> One very simple new question that arises here, to illustrate my point: Is it legal to export a record that has sourceIPAddress X AND NOT sourceIPAddress X?
>>    
>  From a protocol point of view, yes
>  From a semantic point of view, I don't see a use case for that.
> Now, this question is not different that: with RFC5101,  is it legal to 
> export a record that has two instances of sourceIPAddress?
>>    
>>> As you wrote, "in almost all cases I can think of it seems like these are implicitly "AND" semnantics", so the case we try to solve is when there are multiple instances of a single IE. We know that RFC 5101 foresaw that case " The Collector MUST support the use of Templates containing multiple occurrences of the similar Information Elements", but the idea is not to change RFC5101.  If we put some semantic in the IPFIX structured data, that would solve the vast majority of our cases. Also, we could say: if you want some semantic when exporting multiple IEs, then you SHOULD use IPFIX structured data.
>>>
>>> Also, the default value for the semantic field in the IPFIX structured data SHOULD be "NONE", to express that flow record doesn't include any semantic.... like in RFC5101. You might draw your own conclusion, maybe because you know your network, maybe because you have configured the exporter, but then it's your decision.
>>> The way I see the proposed solution is: in IPFIX structured data, you MAY use the semantic field as a way to express the relationship between IEs within the structure.
>>>      
>>>> 2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid argument against Solution 2, as it's trying to solve a somewhat different and more limited problem than structured data. Solution 2 _might_ cause a problem for this draft, but certainly not the other way around (unless we as a WG want to subsume that work into this draft, which would probably require rechartering...). Also, I don't think we need semantics for all of the list types, just basicList (Illustrative question: How do I meaningfully interpret two "or" subTemplateMultiLists with disjoint IE sets? Two "and" subTemplateMultiLists? Nestings thereof?)... So, we don't really have an explosion to deal with if we do Solution 2 correctly: andBasicList, orBasicList, xorBasicList, notBasicList; Four IEs, nestable, done. What can't we do with those four IEs? In this case, we could even step back and say that semantics outside these four are within the protocol explicitly _undefined_, and to be interpreted withi
n the
>>>>   context of each Template.
>>>>
>>>>
>>>>        
>>> If I translate some more what I wrote "While it's solved the router and most mediation function needs today", this would be "Me, myself, and I don't need more that OR and AND now ;-)".
>>> However, who am I to tell that others don't need it now... and that the logical solution is to use the IPFIX structured data
>>> Furthermore, if we think a little bit longer term, the next big step in IPFIX is the mediation function. In my company, every features want to export his own data with NetFlow/IPFIX... up to the point where a CPE would not have enough bandwidth across the WAN to export all the "management" information. So we'll need more and more of aggregated flow records (both in time and space) even in the router. Again, the logical solution will be to use the IPFIX structured data. At this point, we will most probably need something else than OR and AND, i.e. RANGE, ORDERRED, etc...
>>>
>>> An example of subTemplateMultiLists with disjoint IE sets, let's imagine that you have to export an aggregated observation point, composed of multiple template records
>>>      template 1: exporterIPaddress
>>>      template 2: exporterIPaddress, basicList of interfaces
>>>      template 3: exporterIPaddress, LC
>>>      
>> and then you'd want to OR these... Okay, makes sense...
>>
>>    
>>>      
>>>> 3. If we really _do_ want ranges and so on (which, again, we'd need to get WG consensus on; this is explicitly out of scope in my reading of the present charter), then we could do them in the scope of Solution 3.
>>>>        
>>> Btw, acknowledging that one day we will have to solve this is good enough for solution 3. I mean, we don't have to populate the semantic IANA now... even if that would be more efficient.
>>>      
>>>> However, this seems a little not-quite-fleshed-out-enough for me to say whether I like it or not. Could you present an example of how you would use the proposed Semantic field to model your example? "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR linecard2"
>>>>
>>>> (FWIW, I would do this with my proposal to Solution 2 as follows:)
>>>>
>>>> (orBasicList (andBasicList (orBasicList ingressInterface eth1 eth2) (notBasicList (orBasicList ingressInterface eth3 eth4)) (andBasicList lineCardID 2))
>>>>
>>>>
>>>>        
>>> (BasicList, OR, (basicList, AND, (basicList, OR, eth1, eth2), (basicList, NOR, eth3, eth4)), (basicList, NONE, linecard2)))
>>>      
>> Hm. Okay. This makes sense. A couple of questions then, about solution 3:
>>
>> 1. Would we want to try and bitfield this, to define the semantics we _know_ we have, then leave the rest of it reserved? Something like:
>>
>> MSb                      LSb
>> +---+------+---------------+
>> | ! | multi| reserved      |
>> +---+------+---------------+
>>
>> ! = negate sense of semantics if 1 (this is the NOT flag)
>> multi (multiplicity) = 00: undefined, 01: or/oneOrMore, 10: xor/exactlyOne, 11: and/exactlyAll
>> reserved = place for adding other bells and whistles like RANGE and so on in the future.
>>    
> Maybe we want to start by asking the question: which semantic do we need 
> now?
> Is a need for NONE, OR, AND, ORDERED for now. Anything else?
>> 2. Are we sure we want to do this in one byte? If we do bitfielding as above this gives us 32 possible extensions, which seems like _way_ more than enough, but does stick another odd offset in there, which slows things down on machines that need aligned access. Probably one byte is okay and we let the implementation use paddingOctets and set padding to fix this...
>>    
> With structured data, is the alignment still important? When I look at 
> the examples throughout the draft... As you wrote, paddingOctets might 
> be the solution in this case.
>> 3. Would it make sense maybe to have two sets of structured data elements, one with the semantics byte, and one without (which is then explicitly undefined)? Then exporters who don't need it don't have to bother sticking an extra odd-aligned zero in the stream for every list.
>>
>> I'm sure I'l have more questions, but none come to mind now... But it seems like we're converging on the least-unnecessarily-complex solution here, which is good. :)
>>    
> Happy about that. ;-)
> Note: I was envisioning even something more simpler: one byte, 
> containing all the semantic possibilities, administered by IANA.  So no 
> reserved, no !
> 
> Regards, Benoit.


--------------ms050602000205040806020403
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms050602000205040806020403--

From bclaise@cisco.com  Wed Mar 17 06:14:01 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id CBD173A6A3F for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 06:14:01 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.695
X-Spam-Level: 
X-Spam-Status: No, score=-1.695 tagged_above=-999 required=5 tests=[AWL=-0.226, BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id pAsjX650dubm for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 06:14:00 -0700 (PDT)
Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id 555A93A6C3C for <ipfix@ietf.org>; Wed, 17 Mar 2010 06:10:28 -0700 (PDT)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2HDAaL4011419; Wed, 17 Mar 2010 14:10:36 +0100 (CET)
Received: from [10.55.43.57] (ams-bclaise-8718.cisco.com [10.55.43.57]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2HDAZQh001160; Wed, 17 Mar 2010 14:10:36 +0100 (CET)
Message-ID: <4BA0D4CB.6000308@cisco.com>
Date: Wed, 17 Mar 2010 14:10:35 +0100
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <4AF73525.8050009@net.in.tum.de> <4AF8F999.3000207@cisco.com>	<F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>	<4B9E33BE.7070907@cisco.com>	<7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch>	<4B9F520E.8060507@cisco.com>	<DB6E59D9-B373-4919-BB58-00EB26014564@tik.ee.ethz.ch>	<4BA0C26D.9070901@cisco.com> <4BA0D177.1070506@net.in.tum.de>
In-Reply-To: <4BA0D177.1070506@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 17 Mar 2010 13:14:01 -0000

Thanks Gerhard for your feedback.
See inline.
>
> Hi all,
>
> Some general thoughts from my side:
>
> - I appreciate that you want to add a basic notion of semantic to the
>   structured data.
Great.
>
> - Up to now, semantic was not in the protocol but in the info model.
>
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> |0|               Field ID    |       Element Length            |
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> | Semantic  |             BasicList Content ...                 |
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> |                           ...                                 |
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>
>   Can't we encode Semantic in an IE?
>   Then, we could do without a new IANA registry.
>   And: This IE could be used for other purposes (e.g., in Templates)
We thought about this one. However, one of the IPFIX principle is that 
the semantic of one IE can't depend on the value or position of some 
other IEs.
One example of this is the (MPLS label position, MPLS label). Way before 
structured data, our initial implementation contained: MPLS label 
position, MPLS label.
So the collector had to look first at the MPLS label position value in 
order to correctly understand the following  MPLS label.
Somehow, with these conventions, we were having an information model on 
the top of the IE. This was wrong.
Also it was not obvious for a collector, i.e. without hardcoding the 
information.
Finally, what if the IE order changed within the flow record?
Conclusion: we had to change back our implementation mplslabelposition1, 
mplslabelposition2, etc... Like we did in IPIFX ;-)


>
> - I'm not sure whether the bitvector encoding has an advantage over an
>   enumeration type.
>
> Unfortunately, I do not have the time to participate in a deep 
> discussion - I'm busy with IPFIX-CONFIG and other stuff.
Thanks and regards, Benoit.
>
> Regards,
> Gerhard
>
> Benoit Claise wrote:
>> Hi Brian,
>>> Hi, Benoit,
>>>
>>> Replies inline...
>>>
>>> On Mar 16, 2010, at 2:40 AM, Benoit Claise wrote:
>>>
>>>> Hi Brian,
>>>>
>>>> Thanks for your feedback.
>>>>> hi, Benoit, all...
>>>>>
>>>>> Agreed that we should do something about this (i.e., that solution 
>>>>> 1 is no solution.); that said, a few comments in no particular order:
>>>>>
>>>>> 1. In considering adding explicit semantics to structured data, we 
>>>>> as a WG are taking on the task of defining semantics for IPFIX as 
>>>>> a whole. Semantics, as I understand them, are in IPFIX largely 
>>>>> contextual and template dependent, but in almost all cases I can 
>>>>> think of it seems like these are implicitly "AND" semnantics (this 
>>>>> flow has source IP A _AND_ destination IP B _AND_...).
>>>>>
>>>> Agreed.
>>>>> We will need to make an explicit statement on this.
>>>> Not sure why.
>>>>> We will need to determine whether these implicit semantics are a 
>>>>> property of the protocol (in which case Structured Data is really 
>>>>> a protocol-level extension), or a property of each information 
>>>>> element (in which case all 5103 IEs have implicit AND semantics; 
>>>>> question 2: will we need to add semantics to the IANA registry in 
>>>>> this case?). We will need to be quite careful about this. It's not 
>>>>> as simple as defining semantics within structured data then 
>>>>> calling it done.
>>>>>
>>>>>
>>>> I'm not sure why each individual IE should have a semantic ... in 
>>>> the IANA registry.
>>> I'm not either. On reflection it seems like overkill. I'm just 
>>> saying that if we, as a WG, are moving from stating that semantics 
>>> are explicitly out of scope to defining them as in-scope, we need to 
>>> have a consistent approach, and answers to all the questions that 
>>> arise when we consider moving the protocol from a simple framing 
>>> mechanism to a framing mechanism with some logic behind it, so we 
>>> answer them once, so that all future efforts having to do with 
>>> semantics are consistent. This, I acknowledge, is an argument in 
>>> favor of solution 3...)
>>>
>>> One very simple new question that arises here, to illustrate my 
>>> point: Is it legal to export a record that has sourceIPAddress X AND 
>>> NOT sourceIPAddress X?
>>  From a protocol point of view, yes
>>  From a semantic point of view, I don't see a use case for that.
>> Now, this question is not different that: with RFC5101,  is it legal 
>> to export a record that has two instances of sourceIPAddress?
>>>> As you wrote, "in almost all cases I can think of it seems like 
>>>> these are implicitly "AND" semnantics", so the case we try to solve 
>>>> is when there are multiple instances of a single IE. We know that 
>>>> RFC 5101 foresaw that case " The Collector MUST support the use of 
>>>> Templates containing multiple occurrences of the similar 
>>>> Information Elements", but the idea is not to change RFC5101.  If 
>>>> we put some semantic in the IPFIX structured data, that would solve 
>>>> the vast majority of our cases. Also, we could say: if you want 
>>>> some semantic when exporting multiple IEs, then you SHOULD use 
>>>> IPFIX structured data.
>>>>
>>>> Also, the default value for the semantic field in the IPFIX 
>>>> structured data SHOULD be "NONE", to express that flow record 
>>>> doesn't include any semantic.... like in RFC5101. You might draw 
>>>> your own conclusion, maybe because you know your network, maybe 
>>>> because you have configured the exporter, but then it's your decision.
>>>> The way I see the proposed solution is: in IPFIX structured data, 
>>>> you MAY use the semantic field as a way to express the relationship 
>>>> between IEs within the structure.
>>>>> 2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid 
>>>>> argument against Solution 2, as it's trying to solve a somewhat 
>>>>> different and more limited problem than structured data. Solution 
>>>>> 2 _might_ cause a problem for this draft, but certainly not the 
>>>>> other way around (unless we as a WG want to subsume that work into 
>>>>> this draft, which would probably require rechartering...). Also, I 
>>>>> don't think we need semantics for all of the list types, just 
>>>>> basicList (Illustrative question: How do I meaningfully interpret 
>>>>> two "or" subTemplateMultiLists with disjoint IE sets? Two "and" 
>>>>> subTemplateMultiLists? Nestings thereof?)... So, we don't really 
>>>>> have an explosion to deal with if we do Solution 2 correctly: 
>>>>> andBasicList, orBasicList, xorBasicList, notBasicList; Four IEs, 
>>>>> nestable, done. What can't we do with those four IEs? In this 
>>>>> case, we could even step back and say that semantics outside these 
>>>>> four are within the protocol explicitly _undefined_, and to be 
>>>>> interpreted withi
> n the
>>>>>   context of each Template.
>>>>>
>>>>>
>>>> If I translate some more what I wrote "While it's solved the router 
>>>> and most mediation function needs today", this would be "Me, 
>>>> myself, and I don't need more that OR and AND now ;-)".
>>>> However, who am I to tell that others don't need it now... and that 
>>>> the logical solution is to use the IPFIX structured data
>>>> Furthermore, if we think a little bit longer term, the next big 
>>>> step in IPFIX is the mediation function. In my company, every 
>>>> features want to export his own data with NetFlow/IPFIX... up to 
>>>> the point where a CPE would not have enough bandwidth across the 
>>>> WAN to export all the "management" information. So we'll need more 
>>>> and more of aggregated flow records (both in time and space) even 
>>>> in the router. Again, the logical solution will be to use the IPFIX 
>>>> structured data. At this point, we will most probably need 
>>>> something else than OR and AND, i.e. RANGE, ORDERRED, etc...
>>>>
>>>> An example of subTemplateMultiLists with disjoint IE sets, let's 
>>>> imagine that you have to export an aggregated observation point, 
>>>> composed of multiple template records
>>>>      template 1: exporterIPaddress
>>>>      template 2: exporterIPaddress, basicList of interfaces
>>>>      template 3: exporterIPaddress, LC
>>> and then you'd want to OR these... Okay, makes sense...
>>>
>>>>> 3. If we really _do_ want ranges and so on (which, again, we'd 
>>>>> need to get WG consensus on; this is explicitly out of scope in my 
>>>>> reading of the present charter), then we could do them in the 
>>>>> scope of Solution 3.
>>>> Btw, acknowledging that one day we will have to solve this is good 
>>>> enough for solution 3. I mean, we don't have to populate the 
>>>> semantic IANA now... even if that would be more efficient.
>>>>> However, this seems a little not-quite-fleshed-out-enough for me 
>>>>> to say whether I like it or not. Could you present an example of 
>>>>> how you would use the proposed Semantic field to model your 
>>>>> example? "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR linecard2"
>>>>>
>>>>> (FWIW, I would do this with my proposal to Solution 2 as follows:)
>>>>>
>>>>> (orBasicList (andBasicList (orBasicList ingressInterface eth1 
>>>>> eth2) (notBasicList (orBasicList ingressInterface eth3 eth4)) 
>>>>> (andBasicList lineCardID 2))
>>>>>
>>>>>
>>>> (BasicList, OR, (basicList, AND, (basicList, OR, eth1, eth2), 
>>>> (basicList, NOR, eth3, eth4)), (basicList, NONE, linecard2)))
>>> Hm. Okay. This makes sense. A couple of questions then, about 
>>> solution 3:
>>>
>>> 1. Would we want to try and bitfield this, to define the semantics 
>>> we _know_ we have, then leave the rest of it reserved? Something like:
>>>
>>> MSb                      LSb
>>> +---+------+---------------+
>>> | ! | multi| reserved      |
>>> +---+------+---------------+
>>>
>>> ! = negate sense of semantics if 1 (this is the NOT flag)
>>> multi (multiplicity) = 00: undefined, 01: or/oneOrMore, 10: 
>>> xor/exactlyOne, 11: and/exactlyAll
>>> reserved = place for adding other bells and whistles like RANGE and 
>>> so on in the future.
>> Maybe we want to start by asking the question: which semantic do we 
>> need now?
>> Is a need for NONE, OR, AND, ORDERED for now. Anything else?
>>> 2. Are we sure we want to do this in one byte? If we do bitfielding 
>>> as above this gives us 32 possible extensions, which seems like 
>>> _way_ more than enough, but does stick another odd offset in there, 
>>> which slows things down on machines that need aligned access. 
>>> Probably one byte is okay and we let the implementation use 
>>> paddingOctets and set padding to fix this...
>> With structured data, is the alignment still important? When I look 
>> at the examples throughout the draft... As you wrote, paddingOctets 
>> might be the solution in this case.
>>> 3. Would it make sense maybe to have two sets of structured data 
>>> elements, one with the semantics byte, and one without (which is 
>>> then explicitly undefined)? Then exporters who don't need it don't 
>>> have to bother sticking an extra odd-aligned zero in the stream for 
>>> every list.
>>>
>>> I'm sure I'l have more questions, but none come to mind now... But 
>>> it seems like we're converging on the least-unnecessarily-complex 
>>> solution here, which is good. :)
>> Happy about that. ;-)
>> Note: I was envisioning even something more simpler: one byte, 
>> containing all the semantic possibilities, administered by IANA.  So 
>> no reserved, no !
>>
>> Regards, Benoit.
>


From paitken@cisco.com  Wed Mar 17 06:22:20 2010
Return-Path: <paitken@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 131743A6A3B for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 06:22:17 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -7.497
X-Spam-Level: 
X-Spam-Status: No, score=-7.497 tagged_above=-999 required=5 tests=[AWL=1.972,  BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id obZdHDyEVA4i for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 06:22:09 -0700 (PDT)
Received: from ams-iport-1.cisco.com (ams-iport-1.cisco.com [144.254.224.140]) by core3.amsl.com (Postfix) with ESMTP id 5B3C43A6C3C for <ipfix@ietf.org>; Wed, 17 Mar 2010 06:19:25 -0700 (PDT)
Authentication-Results: ams-iport-1.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-AV: E=Sophos;i="4.49,657,1262563200"; d="scan'208";a="58179086"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 17 Mar 2010 13:19:33 +0000
Received: from [144.254.153.37] (dhcp-144-254-153-37.cisco.com [144.254.153.37]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2HDJWrw002268; Wed, 17 Mar 2010 13:19:32 GMT
Message-ID: <4BA0D6E4.6030509@cisco.com>
Date: Wed, 17 Mar 2010 13:19:32 +0000
From: Paul Aitken <paitken@cisco.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-GB; rv:1.8.1.22) Gecko/20090605 SeaMonkey/1.1.17 (Ubuntu-1.1.17+nobinonly-0ubuntu0.9.04.1)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <4AF73525.8050009@net.in.tum.de>	<4AF8F999.3000207@cisco.com>	<F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>	<4B9E33BE.7070907@cisco.com>	<7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch>	<4B9F520E.8060507@cisco.com>	<DB6E59D9-B373-4919-BB58-00EB26014564@tik.ee.ethz.ch>	<4BA0C26D.9070901@cisco.com> <4BA0D177.1070506@net.in.tum.de>
In-Reply-To: <4BA0D177.1070506@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] [Sender: ipfix-bounces@ietf.org] Re: Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 17 Mar 2010 13:22:20 -0000

Gerhard,

> Some general thoughts from my side:
> 
> - I appreciate that you want to add a basic notion of semantic to the
>   structured data.
> 
> - Up to now, semantic was not in the protocol but in the info model.
> 
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> |0|               Field ID    |       Element Length            |
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> | Semantic  |             BasicList Content ...                 |
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> |                           ...                                 |
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> 
>   Can't we encode Semantic in an IE?
>   Then, we could do without a new IANA registry.

We'd still need a (sub)registry to allow the IE to be extensible.
eg, see the "MPLS label type" registry.


>   And: This IE could be used for other purposes (e.g., in Templates)

How do you see that being used?


> - I'm not sure whether the bitvector encoding has an advantage over an
>   enumeration type.

A bitvector would allow multiple semantics where most combinations would 
not be useful. eg, "AND OR". Only "NOT" + "second semantic" seems useful.


> Unfortunately, I do not have the time to participate in a deep 
> discussion - I'm busy with IPFIX-CONFIG and other stuff.

Thanks for taking the time.

P.


> Benoit Claise wrote:
>> Hi Brian,
>>> Hi, Benoit,
>>>
>>> Replies inline...
>>>
>>> On Mar 16, 2010, at 2:40 AM, Benoit Claise wrote:
>>>
>>>   
>>>> Hi Brian,
>>>>
>>>> Thanks for your feedback.
>>>>     
>>>>> hi, Benoit, all...
>>>>>
>>>>> Agreed that we should do something about this (i.e., that solution 
>>>>> 1 is no solution.); that said, a few comments in no particular order:
>>>>>
>>>>> 1. In considering adding explicit semantics to structured data, we 
>>>>> as a WG are taking on the task of defining semantics for IPFIX as a 
>>>>> whole. Semantics, as I understand them, are in IPFIX largely 
>>>>> contextual and template dependent, but in almost all cases I can 
>>>>> think of it seems like these are implicitly "AND" semnantics (this 
>>>>> flow has source IP A _AND_ destination IP B _AND_...).
>>>>>
>>>>>        
>>>> Agreed.
>>>>     
>>>>> We will need to make an explicit statement on this.
>>>>>        
>>>> Not sure why.
>>>>     
>>>>> We will need to determine whether these implicit semantics are a 
>>>>> property of the protocol (in which case Structured Data is really a 
>>>>> protocol-level extension), or a property of each information 
>>>>> element (in which case all 5103 IEs have implicit AND semantics; 
>>>>> question 2: will we need to add semantics to the IANA registry in 
>>>>> this case?). We will need to be quite careful about this. It's not 
>>>>> as simple as defining semantics within structured data then calling 
>>>>> it done.
>>>>>
>>>>>
>>>>>        
>>>> I'm not sure why each individual IE should have a semantic ... in 
>>>> the IANA registry.
>>>>      
>>> I'm not either. On reflection it seems like overkill. I'm just saying 
>>> that if we, as a WG, are moving from stating that semantics are 
>>> explicitly out of scope to defining them as in-scope, we need to have 
>>> a consistent approach, and answers to all the questions that arise 
>>> when we consider moving the protocol from a simple framing mechanism 
>>> to a framing mechanism with some logic behind it, so we answer them 
>>> once, so that all future efforts having to do with semantics are 
>>> consistent. This, I acknowledge, is an argument in favor of solution 
>>> 3...)
>>>
>>> One very simple new question that arises here, to illustrate my 
>>> point: Is it legal to export a record that has sourceIPAddress X AND 
>>> NOT sourceIPAddress X?
>>>    
>>  From a protocol point of view, yes
>>  From a semantic point of view, I don't see a use case for that.
>> Now, this question is not different that: with RFC5101,  is it legal 
>> to export a record that has two instances of sourceIPAddress?
>>>   
>>>> As you wrote, "in almost all cases I can think of it seems like 
>>>> these are implicitly "AND" semnantics", so the case we try to solve 
>>>> is when there are multiple instances of a single IE. We know that 
>>>> RFC 5101 foresaw that case " The Collector MUST support the use of 
>>>> Templates containing multiple occurrences of the similar Information 
>>>> Elements", but the idea is not to change RFC5101.  If we put some 
>>>> semantic in the IPFIX structured data, that would solve the vast 
>>>> majority of our cases. Also, we could say: if you want some semantic 
>>>> when exporting multiple IEs, then you SHOULD use IPFIX structured data.
>>>>
>>>> Also, the default value for the semantic field in the IPFIX 
>>>> structured data SHOULD be "NONE", to express that flow record 
>>>> doesn't include any semantic.... like in RFC5101. You might draw 
>>>> your own conclusion, maybe because you know your network, maybe 
>>>> because you have configured the exporter, but then it's your decision.
>>>> The way I see the proposed solution is: in IPFIX structured data, 
>>>> you MAY use the semantic field as a way to express the relationship 
>>>> between IEs within the structure.
>>>>     
>>>>> 2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid 
>>>>> argument against Solution 2, as it's trying to solve a somewhat 
>>>>> different and more limited problem than structured data. Solution 2 
>>>>> _might_ cause a problem for this draft, but certainly not the other 
>>>>> way around (unless we as a WG want to subsume that work into this 
>>>>> draft, which would probably require rechartering...). Also, I don't 
>>>>> think we need semantics for all of the list types, just basicList 
>>>>> (Illustrative question: How do I meaningfully interpret two "or" 
>>>>> subTemplateMultiLists with disjoint IE sets? Two "and" 
>>>>> subTemplateMultiLists? Nestings thereof?)... So, we don't really 
>>>>> have an explosion to deal with if we do Solution 2 correctly: 
>>>>> andBasicList, orBasicList, xorBasicList, notBasicList; Four IEs, 
>>>>> nestable, done. What can't we do with those four IEs? In this case, 
>>>>> we could even step back and say that semantics outside these four 
>>>>> are within the protocol explicitly _undefined_, and to be 
>>>>> interpreted withi
> n the
>>>>>   context of each Template.
>>>>>
>>>>>
>>>>>        
>>>> If I translate some more what I wrote "While it's solved the router 
>>>> and most mediation function needs today", this would be "Me, myself, 
>>>> and I don't need more that OR and AND now ;-)".
>>>> However, who am I to tell that others don't need it now... and that 
>>>> the logical solution is to use the IPFIX structured data
>>>> Furthermore, if we think a little bit longer term, the next big step 
>>>> in IPFIX is the mediation function. In my company, every features 
>>>> want to export his own data with NetFlow/IPFIX... up to the point 
>>>> where a CPE would not have enough bandwidth across the WAN to export 
>>>> all the "management" information. So we'll need more and more of 
>>>> aggregated flow records (both in time and space) even in the router. 
>>>> Again, the logical solution will be to use the IPFIX structured 
>>>> data. At this point, we will most probably need something else than 
>>>> OR and AND, i.e. RANGE, ORDERRED, etc...
>>>>
>>>> An example of subTemplateMultiLists with disjoint IE sets, let's 
>>>> imagine that you have to export an aggregated observation point, 
>>>> composed of multiple template records
>>>>      template 1: exporterIPaddress
>>>>      template 2: exporterIPaddress, basicList of interfaces
>>>>      template 3: exporterIPaddress, LC
>>>>      
>>> and then you'd want to OR these... Okay, makes sense...
>>>
>>>   
>>>>     
>>>>> 3. If we really _do_ want ranges and so on (which, again, we'd need 
>>>>> to get WG consensus on; this is explicitly out of scope in my 
>>>>> reading of the present charter), then we could do them in the scope 
>>>>> of Solution 3.
>>>>>        
>>>> Btw, acknowledging that one day we will have to solve this is good 
>>>> enough for solution 3. I mean, we don't have to populate the 
>>>> semantic IANA now... even if that would be more efficient.
>>>>     
>>>>> However, this seems a little not-quite-fleshed-out-enough for me to 
>>>>> say whether I like it or not. Could you present an example of how 
>>>>> you would use the proposed Semantic field to model your example? 
>>>>> "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR linecard2"
>>>>>
>>>>> (FWIW, I would do this with my proposal to Solution 2 as follows:)
>>>>>
>>>>> (orBasicList (andBasicList (orBasicList ingressInterface eth1 eth2) 
>>>>> (notBasicList (orBasicList ingressInterface eth3 eth4)) 
>>>>> (andBasicList lineCardID 2))
>>>>>
>>>>>
>>>>>        
>>>> (BasicList, OR, (basicList, AND, (basicList, OR, eth1, eth2), 
>>>> (basicList, NOR, eth3, eth4)), (basicList, NONE, linecard2)))
>>>>      
>>> Hm. Okay. This makes sense. A couple of questions then, about 
>>> solution 3:
>>>
>>> 1. Would we want to try and bitfield this, to define the semantics we 
>>> _know_ we have, then leave the rest of it reserved? Something like:
>>>
>>> MSb                      LSb
>>> +---+------+---------------+
>>> | ! | multi| reserved      |
>>> +---+------+---------------+
>>>
>>> ! = negate sense of semantics if 1 (this is the NOT flag)
>>> multi (multiplicity) = 00: undefined, 01: or/oneOrMore, 10: 
>>> xor/exactlyOne, 11: and/exactlyAll
>>> reserved = place for adding other bells and whistles like RANGE and 
>>> so on in the future.
>>>    
>> Maybe we want to start by asking the question: which semantic do we 
>> need now?
>> Is a need for NONE, OR, AND, ORDERED for now. Anything else?
>>> 2. Are we sure we want to do this in one byte? If we do bitfielding 
>>> as above this gives us 32 possible extensions, which seems like _way_ 
>>> more than enough, but does stick another odd offset in there, which 
>>> slows things down on machines that need aligned access. Probably one 
>>> byte is okay and we let the implementation use paddingOctets and set 
>>> padding to fix this...
>>>    
>> With structured data, is the alignment still important? When I look at 
>> the examples throughout the draft... As you wrote, paddingOctets might 
>> be the solution in this case.
>>> 3. Would it make sense maybe to have two sets of structured data 
>>> elements, one with the semantics byte, and one without (which is then 
>>> explicitly undefined)? Then exporters who don't need it don't have to 
>>> bother sticking an extra odd-aligned zero in the stream for every list.
>>>
>>> I'm sure I'l have more questions, but none come to mind now... But it 
>>> seems like we're converging on the least-unnecessarily-complex 
>>> solution here, which is good. :)
>>>    
>> Happy about that. ;-)
>> Note: I was envisioning even something more simpler: one byte, 
>> containing all the semantic possibilities, administered by IANA.  So 
>> no reserved, no !
>>
>> Regards, Benoit.
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> IPFIX mailing list
> IPFIX@ietf.org
> https://www.ietf.org/mailman/listinfo/ipfix


-- 
Paul Aitken
Cisco Systems Ltd, Edinburgh, Scotland.
http://www.cisco.com/web/about/doing_business/legal/cri/index.html

From andrjohn@cisco.com  Wed Mar 17 07:38:02 2010
Return-Path: <andrjohn@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id CEC1E3A6CF6 for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 07:38:02 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -10.034
X-Spam-Level: 
X-Spam-Status: No, score=-10.034 tagged_above=-999 required=5 tests=[AWL=-0.565, BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Z9XmkASsYnu3 for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 07:38:01 -0700 (PDT)
Received: from sj-iport-5.cisco.com (sj-iport-5.cisco.com [171.68.10.87]) by core3.amsl.com (Postfix) with ESMTP id ACE423A699A for <ipfix@ietf.org>; Wed, 17 Mar 2010 07:29:24 -0700 (PDT)
Authentication-Results: sj-iport-5.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AvsEAIKEoEurRN+J/2dsb2JhbACbBHOhSphhhHYE
X-IronPort-AV: E=Sophos;i="4.49,657,1262563200"; d="scan'208";a="167688100"
Received: from sj-core-3.cisco.com ([171.68.223.137]) by sj-iport-5.cisco.com with ESMTP; 17 Mar 2010 14:29:34 +0000
Received: from cisco.com (mrwint.cisco.com [64.103.71.48]) by sj-core-3.cisco.com (8.13.8/8.14.3) with ESMTP id o2HETXRg026940; Wed, 17 Mar 2010 14:29:34 GMT
Received: from [144.254.153.59] (dhcp-144-254-153-59.cisco.com [144.254.153.59]) by cisco.com (8.11.7p3+Sun/8.8.8) with ESMTP id o2HETWX02756; Wed, 17 Mar 2010 14:29:32 GMT
Message-ID: <4BA0E75E.7060507@cisco.com>
Date: Wed, 17 Mar 2010 14:29:50 +0000
From: Andrew Johnson <andrjohn@cisco.com>
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com> <4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com> <4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E06B4.6090203@net.in.tum.de> <4B9FFE9C.3060908@cisco.com> <4BA0A7C8.9010205@net.in.tum.de>
In-Reply-To: <4BA0A7C8.9010205@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 17 Mar 2010 14:38:02 -0000

Hi Gerhard

One question before I comment inline.  How do you directly attach a
cache to an observation point?  i.e. no sampling or filtering.


Gerhard Muenz wrote:
> 
> Hi Andrew,
> 
> Andrew Johnson wrote:
>> Hi Gerhard
>>
>> Gerhard Muenz wrote:
>> [SNIP]
>>> To make it clearer, I will add an explicit note that the counters in 
>>> the Selector class correspond to the counters in the IPFIX MIB.
>>>
>>> Just like in the MIB specification, we do not specify how the 
>>> counters are related to the PSAMP statistics reports.
>>
>> I think the current configuration model has a few problems:
>>    * Implementations seem forced to consider more than one cache part
>>      of the same Metering Process to be able to use the cascading
>>      selection configuration.
> 
> Not at all!
> 
> As I have written multiple times here on the list, no implementation is 
> expected to support the entire configuration data model. That is just 
> impossible. Every implementation will have restriction at one end or the 
> other. This is not a specific problem to the Selection Process or Cache 
> at all.
> 
> YANG allows you to specify restrictions and extensions in a formal way.
> 
> If you do not think that this becomes clear from the text in the draft, 
> then please provide some better text.

You seem to have misunderstood my point.  I will try to clarify it: An
implementation that treats each cache as a separate Metering Process
will be forced to use a limited subset of the configuration model.

To put this another way, the cascading selector model offers a reasonably
concise way to state the configuration and more implementations could use
this model no attempt was made to map the PSAMP model onto the
configuration model.


>>    * The "SelectionSequenceID" can not be a property of a
>>      "SelectionProcess Class" because that class does not map to a
>>      Selection Sequence (i.e. OP, selectors, Metering Process).
> 
> No. The Selection Process may be composed of multiple Selectors. 
> Therefore, the Selection Process represents a Selection Sequence and has 
> its Selection Sequence ID as state parameter.

Since a "Selection Process" can pass a packet stream on to a second
"Selection Process", it does not necessarily represent the full
Selection Sequence and therefore won't necessarily have a Selection
Sequence ID.

In the case where you have:

               +-> SP2 -> cache A
  OP1 -> SP1 --+
               +-> SP3 -> cache B

SP1 is part of two selection sequences.


>>    * Selectors can not be re-used fully across Selection Sequences as
>>      they are done in the PSAMP model (so I don't think you should have
>>      a "SelectorID" property either).
> 
> You have always stressed that every Selection Sequence has its own 
> Selector instances with separate selection statistics and that only a 
> Selector configuration template can be shared.
> 
> This is reflected in the model!
> 
> The data model specifies instances, no configuration templates (again, 
> I'm repeating myself). So, if you need the same Selector config in two 
> Selection Sequences, you copy the data.

Exactly!  Doesn't it strike you as odd that the configuration model
needs every use of a configuration to be repeated?


> Just like the Selection Sequence ID, the Selector ID is a state 
> parameter assigned by the Monitoring Device. It gives the link to the 
> information exported via PSAMP protocol. Therefore it is useful to have 
> it in the configuration data model.

I'm not convinced any run-time state really belongs in the configuration
data model.  Worse, the attempt to make that state available has made it
really difficult to model simple configurations with configuration model!


> The Monitoring Device may assign the same Selector ID for two Selector 
> instances if the config is identical. There is no problem. Obviously, 
> the Monitoring Device needs to compare the configurations of the 
> different instances.

And you don't think this is a problem?!  Here is one of my assumptions
about a good configuration model.  It is relatively simple for a Metering
Device to generate it's configuration in the model and then read that
configuration back and produce an internal representation of that
configuration that is isomorphic to the original.


>> The PSAMP model is to define selectors as simply configuration.  There 
>> are no statistics on a per Selector basis.  Instead, each use of a 
>> selector needs separate statistics.  From RFC 5476, section 6.5.1:
>>
>>   Example of two Selection Sequences:
>>
>>     Selection Sequence 7 (Filter->Sampling):
>>       ingressInterface    5
>>       selectorId          5 (Filter, match IPV4SourceAddress 192.0.2.1)
>>       selectorId         10 (Sampler, Random 1 out-of ten)
>>
>>     Selection Sequence 9 (Sampling->Filtering):
>>       ingressInterface    5
>>       selectorId         10 (Sampler, Random 1 out-of ten)
>>       selectorId          5 (Filter, match IPV4SourceAddress 192.0.2.1)
>>
> 
> What if these are two different Observation Domains?

The example is very specifically within the same observation domain.


>    * Selector ID
> 
>       The Selector ID is the unique ID identifying a Primitive Selector.
>       The ID is unique within the Observation Domain.
> 
> It does not say: unique per PSAMP Device.

That doesn't matter if you don't put the selector ID in the configuration
model.


> Configuration templates for Selectors would introduce another level of 
> indirection to the model. This will complicate things a bit.
> 
> gIf the Selector ID would be a unique state parameter of the 
> configuration template, then you force Selector IDs to be unique per 
> Monitoring Device. Or you must disallow reusing Selector configuration 
> templates across Observation Domains. Formally specifying such 
> restrictions in YANG results in complicated XPath expresions (or is not 
> possible at all).
> 
> Though, if you insist on configuring templates for Selectors instead of 
> instances, we can change the model accordingly.
> 
> However, this decision will have no effect on the question whether one 
> Selection Process can feed multiple Caches.

For me it will have an impact.  Or rather, I have no objection to a
"Selection Process" feeding multiple caches, what I object to is the
forcing of the sharing of selection state between caches, just because
they share configuration.


>> If we're not going to map the PSAMP model in the configuration model 
>> then I think we should separate them completely.  In fact, I don't 
>> think any attempt should be made to directly configure a "Selection 
>> Process", instead the selection configuration is provided and the 
>> Exporting Device can sort it out.
>>
>> This simply avoids the question about whether a Metering Device can 
>> (or should) have more than one cache.  It also allows greater 
>> flexibility in configuration because a Selector Configuration can have 
>> more than one Packet Stream as input.
> 
> The questions whether the Selector configuration is organized with 
> reusable configuration templates and whether the same Selection Process 
> may feed two Caches are completely orthogonal.
> 
> You must not mix up these two questions!

They aren't completely orthogonal.  An implementation that considers each
cache a separate Metering Process should not be forced to build an
excessively complicated configuration model.


>> I think this is a useful concept.  Imagine this relatively common 
>> configuration, where we want to monitor 10 interfaces, doing 1 in 100 
>> sampling for each.  Rather than define a new sampling selector for 
>> each interface, wouldn't it be better to simply do?:
>>
>>    Selector MY_SAMPLER: sampling 1 in 100 -> cache: MY_CACHE
>>    OP A -> MY_SAMPLER
>>    OP B -> MY_SAMPLER
>>    ...
> 
> What do you mean with "simple"?
> You think that you do not have to copy some lines of XML?
> I assume that configuration management tools can do this work for you.
> 
>> This is closer to the PSAMP model where you would do something like:
>>
>>    Selector MY_SAMPLER: sampling 1 in 100
>>
>>    Selection Sequence: OP A, MY_SAMPLER, MY_CACHE
>>    Selection Sequence: OP B, MY_SAMPLER, MY_CACHE
>>    ...
> 
> Do you also request configuration templates for Caches? Or is this a 
> single instance?

It's the same cache.


Andrew

From trammell@tik.ee.ethz.ch  Wed Mar 17 07:54:37 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 823653A6C3C for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 07:54:37 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.023
X-Spam-Level: 
X-Spam-Status: No, score=-6.023 tagged_above=-999 required=5 tests=[AWL=-0.554, BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 7+jGeIgBYKQ7 for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 07:54:35 -0700 (PDT)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id 5D6F028B56A for <ipfix@ietf.org>; Wed, 17 Mar 2010 07:45:51 -0700 (PDT)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id B0DCFD9480; Wed, 17 Mar 2010 15:45:59 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id FwBNal2M7le2; Wed, 17 Mar 2010 15:45:59 +0100 (MET)
Received: from [10.0.1.5] (unknown [75.85.164.240]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id EFBCBD947A; Wed, 17 Mar 2010 15:45:57 +0100 (MET)
Mime-Version: 1.0 (Apple Message framework v1077)
Content-Type: text/plain; charset=us-ascii
From: Brian Trammell <trammell@tik.ee.ethz.ch>
In-Reply-To: <4BA0C26D.9070901@cisco.com>
Date: Wed, 17 Mar 2010 07:44:03 -0700
Content-Transfer-Encoding: quoted-printable
Message-Id: <6CD84D0C-333F-4C65-8E41-B1DF3774BA15@tik.ee.ethz.ch>
References: <4AF73525.8050009@net.in.tum.de> <4AF8F999.3000207@cisco.com>	<F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>	<4B9E33BE.7070907@cisco.com>	<7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch>	<4B9F520E.8060507@cisco.com> <DB6E59D9-B373-4919-BB58-00EB26014564@tik.ee.ethz.ch> <4BA0C26D.9070901@cisco.com>
To: Benoit Claise <bclaise@cisco.com>
X-Mailer: Apple Mail (2.1077)
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 17 Mar 2010 14:54:37 -0000

Hi, Benoit,

As always... inline.

On Mar 17, 2010, at 4:52 AM, Benoit Claise wrote:

> Hi Brian,
>> Hi, Benoit,
>>=20
>> Replies inline...
>>=20
>> On Mar 16, 2010, at 2:40 AM, Benoit Claise wrote:
>>=20
>>  =20
>>> Hi Brian,
>>>=20
>>> Thanks for your feedback.
>>>    =20
>>>> hi, Benoit, all...
>>>>=20
>>>> Agreed that we should do something about this (i.e., that solution =
1 is no solution.); that said, a few comments in no particular order:
>>>>=20
>>>> 1. In considering adding explicit semantics to structured data, we =
as a WG are taking on the task of defining semantics for IPFIX as a =
whole. Semantics, as I understand them, are in IPFIX largely contextual =
and template dependent, but in almost all cases I can think of it seems =
like these are implicitly "AND" semnantics (this flow has source IP A =
_AND_ destination IP B _AND_...).
>>>>=20
>>>>      =20
>>> Agreed.
>>>    =20
>>>> We will need to make an explicit statement on this.
>>>>      =20
>>> Not sure why.

Because if we're adding semantics to the protocol in any way (which we =
have so far left out of scope), anything we say in structured data =
(standards track) may be taken to apply to 5101 without structured data. =
So a statement that says "semantic interpretations of recieved Data =
Records at or beyond the CP remain explicitly undefined, unless that =
data is transmitted using this extension with explicit semantic =
information" (or something like it) reduces the scope for confusion.=20

>>>   =20
>>>> We will need to determine whether these implicit semantics are a =
property of the protocol (in which case Structured Data is really a =
protocol-level extension), or a property of each information element (in =
which case all 5103 IEs have implicit AND semantics; question 2: will we =
need to add semantics to the IANA registry in this case?). We will need =
to be quite careful about this. It's not as simple as defining semantics =
within structured data then calling it done.
>>>>=20
>>>>=20
>>>>      =20
>>> I'm not sure why each individual IE should have a semantic ... in =
the IANA registry.
>>>    =20
>> I'm not either. On reflection it seems like overkill. I'm just saying =
that if we, as a WG, are moving from stating that semantics are =
explicitly out of scope to defining them as in-scope, we need to have a =
consistent approach, and answers to all the questions that arise when we =
consider moving the protocol from a simple framing mechanism to a =
framing mechanism with some logic behind it, so we answer them once, so =
that all future efforts having to do with semantics are consistent. =
This, I acknowledge, is an argument in favor of solution 3...)
>>=20
>> One very simple new question that arises here, to illustrate my =
point: Is it legal to export a record that has sourceIPAddress X AND NOT =
sourceIPAddress X?
>>  =20
> =46rom a protocol point of view, yes
> =46rom a semantic point of view, I don't see a use case for that.
> Now, this question is not different that: with RFC5101,  is it legal =
to export a record that has two instances of sourceIPAddress?

Point. Okay, yes, it is legal. So in Structured Data we should have a =
statement as well that semantically meaningless combinations are legal, =
and up to the CP to interpret.

>>> As you wrote, "in almost all cases I can think of it seems like =
these are implicitly "AND" semnantics", so the case we try to solve is =
when there are multiple instances of a single IE. We know that RFC 5101 =
foresaw that case " The Collector MUST support the use of Templates =
containing multiple occurrences of the similar Information Elements", =
but the idea is not to change RFC5101.  If we put some semantic in the =
IPFIX structured data, that would solve the vast majority of our cases. =
Also, we could say: if you want some semantic when exporting multiple =
IEs, then you SHOULD use IPFIX structured data.
>>>=20
>>> Also, the default value for the semantic field in the IPFIX =
structured data SHOULD be "NONE", to express that flow record doesn't =
include any semantic.... like in RFC5101. You might draw your own =
conclusion, maybe because you know your network, maybe because you have =
configured the exporter, but then it's your decision.
>>> The way I see the proposed solution is: in IPFIX structured data, =
you MAY use the semantic field as a way to express the relationship =
between IEs within the structure.
>>>    =20
>>>> 2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid =
argument against Solution 2, as it's trying to solve a somewhat =
different and more limited problem than structured data. Solution 2 =
_might_ cause a problem for this draft, but certainly not the other way =
around (unless we as a WG want to subsume that work into this draft, =
which would probably require rechartering...). Also, I don't think we =
need semantics for all of the list types, just basicList (Illustrative =
question: How do I meaningfully interpret two "or" subTemplateMultiLists =
with disjoint IE sets? Two "and" subTemplateMultiLists? Nestings =
thereof?)... So, we don't really have an explosion to deal with if we do =
Solution 2 correctly: andBasicList, orBasicList, xorBasicList, =
notBasicList; Four IEs, nestable, done. What can't we do with those four =
IEs? In this case, we could even step back and say that semantics =
outside these four are within the protocol explicitly _undefined_, and =
to be interpreted within the
>>>>  context of each Template.
>>>>=20
>>>>=20
>>>>      =20
>>> If I translate some more what I wrote "While it's solved the router =
and most mediation function needs today", this would be "Me, myself, and =
I don't need more that OR and AND now ;-)".
>>> However, who am I to tell that others don't need it now... and that =
the logical solution is to use the IPFIX structured data
>>> Furthermore, if we think a little bit longer term, the next big step =
in IPFIX is the mediation function. In my company, every features want =
to export his own data with NetFlow/IPFIX... up to the point where a CPE =
would not have enough bandwidth across the WAN to export all the =
"management" information. So we'll need more and more of aggregated flow =
records (both in time and space) even in the router. Again, the logical =
solution will be to use the IPFIX structured data. At this point, we =
will most probably need something else than OR and AND, i.e. RANGE, =
ORDERRED, etc...
>>>=20
>>> An example of subTemplateMultiLists with disjoint IE sets, let's =
imagine that you have to export an aggregated observation point, =
composed of multiple template records
>>>     template 1: exporterIPaddress
>>>     template 2: exporterIPaddress, basicList of interfaces
>>>     template 3: exporterIPaddress, LC
>>>    =20
>> and then you'd want to OR these... Okay, makes sense...
>>=20
>>  =20
>>>    =20
>>>> 3. If we really _do_ want ranges and so on (which, again, we'd need =
to get WG consensus on; this is explicitly out of scope in my reading of =
the present charter), then we could do them in the scope of Solution 3.
>>>>      =20
>>> Btw, acknowledging that one day we will have to solve this is good =
enough for solution 3. I mean, we don't have to populate the semantic =
IANA now... even if that would be more efficient.
>>>    =20
>>>> However, this seems a little not-quite-fleshed-out-enough for me to =
say whether I like it or not. Could you present an example of how you =
would use the proposed Semantic field to model your example? "(eth1 OR =
eth2) AND (NOT (eth3 OR eth4)) OR linecard2"
>>>>=20
>>>> (FWIW, I would do this with my proposal to Solution 2 as follows:)
>>>>=20
>>>> (orBasicList (andBasicList (orBasicList ingressInterface eth1 eth2) =
(notBasicList (orBasicList ingressInterface eth3 eth4)) (andBasicList =
lineCardID 2))
>>>>=20
>>>>=20
>>>>      =20
>>> (BasicList, OR, (basicList, AND, (basicList, OR, eth1, eth2), =
(basicList, NOR, eth3, eth4)), (basicList, NONE, linecard2)))
>>>    =20
>> Hm. Okay. This makes sense. A couple of questions then, about =
solution 3:
>>=20
>> 1. Would we want to try and bitfield this, to define the semantics we =
_know_ we have, then leave the rest of it reserved? Something like:
>>=20
>> MSb                      LSb
>> +---+------+---------------+
>> | ! | multi| reserved      |
>> +---+------+---------------+
>>=20
>> ! =3D negate sense of semantics if 1 (this is the NOT flag)
>> multi (multiplicity) =3D 00: undefined, 01: or/oneOrMore, 10: =
xor/exactlyOne, 11: and/exactlyAll
>> reserved =3D place for adding other bells and whistles like RANGE and =
so on in the future.
>>  =20
> Maybe we want to start by asking the question: which semantic do we =
need now?
> Is a need for NONE, OR, AND, ORDERED for now. Anything else?

What's ORDERED for? It seems like we're getting multiplicities confused =
with other things here... Is this an AND/allOf with additional =
stipulation that the ordering is important?

>> 2. Are we sure we want to do this in one byte? If we do bitfielding =
as above this gives us 32 possible extensions, which seems like _way_ =
more than enough, but does stick another odd offset in there, which =
slows things down on machines that need aligned access. Probably one =
byte is okay and we let the implementation use paddingOctets and set =
padding to fix this...
>>  =20
> With structured data, is the alignment still important? When I look at =
the examples throughout the draft... As you wrote, paddingOctets might =
be the solution in this case.
>> 3. Would it make sense maybe to have two sets of structured data =
elements, one with the semantics byte, and one without (which is then =
explicitly undefined)? Then exporters who don't need it don't have to =
bother sticking an extra odd-aligned zero in the stream for every list.
>>=20
>> I'm sure I'l have more questions, but none come to mind now... But it =
seems like we're converging on the least-unnecessarily-complex solution =
here, which is good. :)
>>  =20
> Happy about that. ;-)
> Note: I was envisioning even something more simpler: one byte, =
containing all the semantic possibilities, administered by IANA.  So no =
reserved, no !

Okay, then... by "reserved" I meant, "future stuff (IANA) goes here"... =
but yes, that's more complicated than the rest of the subregistries for =
the protocol. So, agreed.

On reflection, I would suggest defining these in terms of multiplicities =
(as we're dealing with N-element lists) as opposed to logical =
operations. So I'd propose the following:

0x00: undefined (default) (CP may assume allOf)
0x01: allOf
0x02: oneOrMoreOf
0x03: exactlyOneOf
0x04: noneOf

If I understand what you mean by ORDERED, add:
0x05: allInOrder

If we really want to do ranges:
0x06: inclusiveRanges (MUST have even number of elements, and elements =
MUST be numeric; this is then equivalent to inclusive allOf [e_0,e_1], =
[e_2,e_3], ... , [e_n-1]..[e_n]);=20

Still unanswered is whether we want to force all =
basicList/subTemplateList/subTemplateMultiList instances to state =
semantics, or whether we want "semanticless" versions as well...

Best regards,

Brian


>>>> On Mar 15, 2010, at 6:18 AM, Benoit Claise wrote:
>>>>=20
>>>>=20
>>>>=20
>>>>      =20
>>>>> Dear all,
>>>>>=20
>>>>> We've been thinking about this one, and we see 3 solutions. We =
believe that the solution 3 is the way to go, as we show below.
>>>>>=20
>>>>> Solution 1.
>>>>> Consider that the semantic is out of scope for this document.
>>>>> This is the easiest solution.  However, we understand it's not =
right: we would not do a complete job wrt to IPFIX structured data
>>>>> As Gerhard was expressing, the collector has not clue how to treat =
the example of a BasicList of egress interfaces in a Flow Record.
>>>>>     - Has every counted packet been sent on every egress =
interface?
>>>>>       =3D>  multicast case, AND semantic
>>>>>     - Has every counted packet been sent on any one of the egress =
interfaces?
>>>>>      =3D>  load balancing case, OR semantic
>>>>>=20
>>>>> Soltution 2
>>>>> We could focus on the logical AND and OR semantics only , by =
defining semantic lists, such as andBasicList, andSubTemplateList, =
andSubTemplateMultiList, orBasicList, orSubTemplateList, and =
orSubTemplateMultiList
>>>>> So 6 I.E.s in total, describing AND and OR semantics.
>>>>> While it's solved the router and most mediation function needs =
today, we understand that this is not a complete solution.
>>>>> Gerhard, in one of his old draft, =
draft-sommer-ipfix-mediator-ext-01, proposed ADTs for orderedList, =
orderedPair, and portRanges which allow the definition new IEs for port =
ranges etc.
>>>>> Even if this solution 2 is extensible, in the long term, will lead =
to an explosion of IEs: 3 list types * the semantic type, where the =
semantic can be (and, or, orderedList, orderedPair, portRanges, random, =
etc...)
>>>>>=20
>>>>> Soltution 3
>>>>> We propose to add a semantic field in the 3 list types.
>>>>> Something such as:
>>>>>=20
>>>>>     =
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>> |0|               Field ID    |       Element Length            |
>>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>> | Semantic  |             BasicList Content ...                 |
>>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>> |                           ...                                 |
>>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>>=20
>>>>> This semantic field would be a new IANA registry, we could be =
populated initially with NONE, OR, AND, ORDERED, etc... Up to =
discussion.
>>>>> The advantage of this solution is that it's extensible, and it =
doesn't need new IEs for each semantic.
>>>>> If ever required (mainly in a mediation function), we can model  =
very complex semantic, eg, "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR =
linecard2", which could be the new observation point from a aggregated =
Flow Record
>>>>>=20
>>>>>=20
>>>>>=20
>>>>> Conclusion:
>>>>> After much debate internally, we believe that we should do the =
effort to include this semantic now, and not postpone the problem, which =
will lead to an explosion of IEs in the future.
>>>>> Personally, I was initially against the solution 3 ... mostly due =
to the effort required to modify the complete specs.
>>>>> We're now ready to modify the specifications in the IPFIX =
structured data, but we would like to get your feedback and agreement in =
advance as this is not a small piece of work.
>>>>>=20
>>>>> Please comment.
>>>>>=20
>>>>> Regards, Paul, Stan, Gowri, and Benoit.
>>>>>=20
>>>>>=20
>>>>>        =20
>>>>>> hi Benoit, Gerhard,
>>>>>>=20
>>>>>> In this case I'm strongly in favor of leaving semantics out of =
structured data. Structured data defines containers. The semantics of =
the containers as a whole and the elements change based upon the =
information elements within the container and within the record =
containing it. Wedging semantics into the structured data elements 1. =
risks further proliferation of (potentially non-interoperable) ways to =
represent the same thing, 2. gives us more ways to represent nonsensical =
things (an OR basicList of MPLS stack entries...means...what?), and 3. =
risks defining an inadequate semantic representation mechanism (what =
about semantics for records not using structured data? what about =
ordered versus unordered sets? what about OR basicList vs AND basicList =
vs nested AND and OR basicLists vs just three identical IEs?). If we =
really want an unambiguous semantic framework for IPFIX (and here I'm =
not convinced either way) that's best done on its own, addressing things =
at the information elem
>>>>>> ent and record level. Doing it here confuses the issue.
>>>>>>=20
>>>>>> Regards,
>>>>>>=20
>>>>>> Brian
>>>>>>=20
>>>>>> On Nov 10, 2009, at 6:26 AM, Benoit Claise wrote:
>>>>>>=20
>>>>>>=20
>>>>>>=20
>>>>>>          =20
>>>>>>> Gerhard,
>>>>>>>=20
>>>>>>> Thanks for your email.
>>>>>>> I have no strong feelings about the two solutions you proposed.
>>>>>>>=20
>>>>>>> > =46rom a pure router point of view,  I don't see any use cases =
for logical OR in exporting flow records.
>>>>>>> However, from a IPFIX Mediator point of view, I see some use =
cases.
>>>>>>> I mean that it requires an Intermediate Aggregation Process or =
Intermediate Correlation Process to express: I've seen this flow record =
OR that flow record.
>>>>>>> Now, it's true that even routers will have mediation =
functions...
>>>>>>>=20
>>>>>>> I'm inclined to add orBasicList, orSubTemplateList, =
orSubTempalteMultiList to the draft (This is a small addition after all) =
and to express that, by default, a logical AND is assumed... =
specifically if the structured data is used in the IPFIX Mediation =
Protocol.
>>>>>>>=20
>>>>>>> I'm not convinced by the NOT in the context of structured data, =
as we don't even have a concept of NOT for a single information element!
>>>>>>>=20
>>>>>>> I would like to get some more feedback from others.
>>>>>>>=20
>>>>>>> Regards, Benoit.
>>>>>>>=20
>>>>>>>=20
>>>>>>>=20
>>>>>>>            =20
>>>>>>>> Dear all,
>>>>>>>>=20
>>>>>>>> Regarding draft-ietf-ipfix-structured-data, I see the risk that =
the
>>>>>>>> semantic of the exported structured data is not clear.
>>>>>>>>=20
>>>>>>>> How do you interpret the manifold occurrence of the same =
Information
>>>>>>>> Element (basicList) or the same group of Information Elements
>>>>>>>> (subTemplateList) in one record?
>>>>>>>>=20
>>>>>>>> What does it mean if basicList, subTemplateList, or =
subTemplateMultiList
>>>>>>>> is used for a Flow Key field? Or non-key field?
>>>>>>>>=20
>>>>>>>> Some Examples:
>>>>>>>>=20
>>>>>>>> - BasicList of egress interfaces in a Flow Record
>>>>>>>>  How should a Flow Record be interpreted which contains a list =
of
>>>>>>>>  egress interfaces and a packet counter?
>>>>>>>>  Has every counted packet been sent on every egress interface?
>>>>>>>>    =3D>  multicast case, AND semantic (see example in section =
8.1)
>>>>>>>>  Has every counted packet been sent on any one of the egress
>>>>>>>>  interfaces?
>>>>>>>>    =3D>  load balancing case, OR semantic
>>>>>>>>  Can it be used as a Flow Key or not?
>>>>>>>>=20
>>>>>>>> - BasicList of destination ports in a Flow Record
>>>>>>>>  As every packet has only one destination port, the only =
reasonable
>>>>>>>>  interpretation is that the Flow contains packets having one of
>>>>>>>>  the reported port numbers.
>>>>>>>>    =3D>  OR semantic
>>>>>>>>  This would be a non-key field.
>>>>>>>>=20
>>>>>>>>=20
>>>>>>>> I think there are two solutions:
>>>>>>>>=20
>>>>>>>> 1. We decide that the semantic of list content is out of scope =
of
>>>>>>>>   draft-ietf-ipfix-structured-data. We add a note to the draft =
that
>>>>>>>>   the semantic must be clear from the context or the definition =
of the
>>>>>>>>   Information Elements used within the lists.
>>>>>>>>=20
>>>>>>>> 2. We define semantic lists, such as
>>>>>>>>   - andBasicList, andSubTemplateList, andSubTemplateMultiList
>>>>>>>>   - orBasicList, orSubTemplateList, orSubTempalteMultiList
>>>>>>>>   describing AND and OR semantic of the contained =
IEs/Templates,
>>>>>>>>   respectively.
>>>>>>>>=20
>>>>>>>>=20
>>>>>>>> As I wrote in an earlier mail, I see a good use case for =
orBasicList. It
>>>>>>>> could be used in the Selector Report Interpretation of Property =
Match
>>>>>>>> Filtering to report a filter like "port 80 or port 443".
>>>>>>>>=20
>>>>>>>> =
http://www.ietf.org/mail-archive/web/ipfix/current/msg04856.html
>>>>>>>>=20
>>>>>>>> At the moment, the Selector Report Interpretation is limited to =
AND.
>>>>>>>> However, if we also want to express a NOT, we still need =
another solution...
>>>>>>>>=20
>>>>>>>> Regards,
>>>>>>>> Gerhard

From andrjohn@cisco.com  Wed Mar 17 08:15:36 2010
Return-Path: <andrjohn@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id A0A123A6C52 for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 08:15:36 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -9.751
X-Spam-Level: 
X-Spam-Status: No, score=-9.751 tagged_above=-999 required=5 tests=[AWL=-0.282, BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id U67MlvDzrWHj for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 08:15:35 -0700 (PDT)
Received: from sj-iport-6.cisco.com (sj-iport-6.cisco.com [171.71.176.117]) by core3.amsl.com (Postfix) with ESMTP id 31F843A6AC3 for <ipfix@ietf.org>; Wed, 17 Mar 2010 08:13:48 -0700 (PDT)
Authentication-Results: sj-iport-6.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AvsEAFuOoEurR7Ht/2dsb2JhbACbBHOhOZhohHYE
X-IronPort-AV: E=Sophos;i="4.49,657,1262563200"; d="scan'208";a="498197461"
Received: from sj-core-1.cisco.com ([171.71.177.237]) by sj-iport-6.cisco.com with ESMTP; 17 Mar 2010 15:13:56 +0000
Received: from cisco.com (mrwint.cisco.com [64.103.71.48]) by sj-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2HFDsOb013520; Wed, 17 Mar 2010 15:13:55 GMT
Received: from [144.254.153.59] (dhcp-144-254-153-59.cisco.com [144.254.153.59]) by cisco.com (8.11.7p3+Sun/8.8.8) with ESMTP id o2HFDsX05931; Wed, 17 Mar 2010 15:13:54 GMT
Message-ID: <4BA0F1C3.5070808@cisco.com>
Date: Wed, 17 Mar 2010 15:14:11 +0000
From: Andrew Johnson <andrjohn@cisco.com>
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com> <4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com> <4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E06B4.6090203@net.in.tum.de> <4B9FFE9C.3060908@cisco.com> <4BA0A7C8.9010205@net.in.tum.de>
In-Reply-To: <4BA0A7C8.9010205@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: [IPFIX] draft-ietf-ipfix-configuration-model by example: Examples 1 - 2
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 17 Mar 2010 15:15:36 -0000

Hi Gerhard

Perhaps we can come to a better understanding if you help me
see how my implementation would use the configuration data
model.  I was originally very optimistic about the configuration
draft because many of the concepts seemed to map very neatly to
how my implementation models its config via the command line.

Would you be willing to map out some real-world configurations
into the configuration data model?  It doesn't have to be the
full YANG output, but I think some text showing the basic
structure would be very useful.


Example 1:

  flow record MY_RECORD
    match ipv4 source address
    match ipv4 destination address
    collect counter bytes

  flow monitor MY_MON
    record MY_RECORD

  interface ethernet 0/0
    ip flow monitor MY_MON input

  interface ethernet 0/1
    ip flow monitor MY_MON input


Typically a customer would configure a monitor on many more
interfaces, perhaps even using a virtual-template or interface
group to add the monitor to hundreds of interfaces, but for now
lets look at just two.

The above configuration directly applies a single monitor
to two interfaces.  In our feature each monitor would be
considered a separate Metering Process, but we don't have to
worry about that here.


Example 2:

  sampler MY_SAMPLER
    mode random 1 out-of 100

  flow record MY_RECORD
    match ipv4 source address
    match ipv4 destination address
    collect counter bytes

  flow monitor MY_MON
    record MY_RECORD

  interface ethernet 0/0
    ip flow monitor MY_MON sampler MY_SAMPLER input

  interface ethernet 0/1
    ip flow monitor MY_MON sampler MY_SAMPLER input


Very similar, except that a we have added a sampling selector.


As I think about these common configurations, what strikes
me as very typical is a reasonably simple configuration that
is then re-used over many interfaces.  It seems to me that
in the current configuration data model that leads to a lot
of repetition.

Cheers

Andrew

From akoba@nttv6.net  Wed Mar 17 14:45:56 2010
Return-Path: <akoba@nttv6.net>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 26E443A69E1 for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 14:45:56 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 1.408
X-Spam-Level: *
X-Spam-Status: No, score=1.408 tagged_above=-999 required=5 tests=[AWL=-0.333,  BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, HOST_MISMATCH_COM=0.311, RDNS_DYNAMIC=0.1, SUBJ_RE_NUM=2.799]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id XhlCcks1GLOY for <ipfix@core3.amsl.com>; Wed, 17 Mar 2010 14:45:54 -0700 (PDT)
Received: from mail.nttv6.net (mail.nttv6.net [IPv6:2001:fa8::25]) by core3.amsl.com (Postfix) with ESMTP id 8FA763A69CD for <ipfix@ietf.org>; Wed, 17 Mar 2010 14:45:53 -0700 (PDT)
Received: from [127.0.0.1] (dhcp-3-152.nttv6.com [192.47.163.152]) by mail.nttv6.net (8.14.3/8.14.3) with ESMTP id o2HLjw7q039751; Thu, 18 Mar 2010 06:46:00 +0900 (JST) (envelope-from akoba@nttv6.net)
Date: Thu, 18 Mar 2010 06:40:40 +0900
From: Atsushi Kobayashi <akoba@nttv6.net>
To: Benoit Claise <bclaise@cisco.com>
In-Reply-To: <4BA0D4CB.6000308@cisco.com>
References: <4BA0D177.1070506@net.in.tum.de> <4BA0D4CB.6000308@cisco.com>
Message-Id: <20100318051549.AFC9.17391CF2@nttv6.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
X-Mailer: Becky! ver. 2.50.05 [ja] (Unregistered)
X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.2.3 (mail.nttv6.net [192.16.178.5]); Thu, 18 Mar 2010 06:46:00 +0900 (JST)
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 17 Mar 2010 21:45:56 -0000

Hi Benoit, and all,

I agree solution3 rather than new semantic IE. 
As Benoit mentioned, semantic IE seems to increase the difficulty to
interpret them.

As Brian mentioned, I think it should avoid to have unnecessary
complexity. But, I am not sure what type, e.g., NONE, OR, AND, NOR,
RANGE, and ORDERRED, is needed. I would like to avoid that one data
structure is represented in multiple ways. To examine it, we needs more
practical examples in addition to: Selector Report
Interpretation and interface lists on aggregated Flow Records.

How about BGP AS Path, Community? Anything else?

When I have the following BGP AS Path mixing as-sequence and as-set, how
to present it?

10 20 30 40 {50,60}

(basicList, ORDERED, (basicList, ORDERED, AS10,AS20,AS30,AS40),
(basicList, OR, AS50, AS60))

Is it correct?

Regards,
Atsushi

On Wed, 17 Mar 2010 14:10:35 +0100
Benoit Claise <bclaise@cisco.com> wrote:

> Thanks Gerhard for your feedback.
> See inline.
> >
> > Hi all,
> >
> > Some general thoughts from my side:
> >
> > - I appreciate that you want to add a basic notion of semantic to the
> >   structured data.
> Great.
> >
> > - Up to now, semantic was not in the protocol but in the info model.
> >
> > +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> > |0|               Field ID    |       Element Length            |
> > +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> > | Semantic  |             BasicList Content ...                 |
> > +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> > |                           ...                                 |
> > +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >
> >   Can't we encode Semantic in an IE?
> >   Then, we could do without a new IANA registry.
> >   And: This IE could be used for other purposes (e.g., in Templates)
> We thought about this one. However, one of the IPFIX principle is that 
> the semantic of one IE can't depend on the value or position of some 
> other IEs.
> One example of this is the (MPLS label position, MPLS label). Way before 
> structured data, our initial implementation contained: MPLS label 
> position, MPLS label.
> So the collector had to look first at the MPLS label position value in 
> order to correctly understand the following  MPLS label.
> Somehow, with these conventions, we were having an information model on 
> the top of the IE. This was wrong.
> Also it was not obvious for a collector, i.e. without hardcoding the 
> information.
> Finally, what if the IE order changed within the flow record?
> Conclusion: we had to change back our implementation mplslabelposition1, 
> mplslabelposition2, etc... Like we did in IPIFX ;-)
> 
> 
> >
> > - I'm not sure whether the bitvector encoding has an advantage over an
> >   enumeration type.
> >
> > Unfortunately, I do not have the time to participate in a deep 
> > discussion - I'm busy with IPFIX-CONFIG and other stuff.
> Thanks and regards, Benoit.
> >
> > Regards,
> > Gerhard
> >
> > Benoit Claise wrote:
> >> Hi Brian,
> >>> Hi, Benoit,
> >>>
> >>> Replies inline...
> >>>
> >>> On Mar 16, 2010, at 2:40 AM, Benoit Claise wrote:
> >>>
> >>>> Hi Brian,
> >>>>
> >>>> Thanks for your feedback.
> >>>>> hi, Benoit, all...
> >>>>>
> >>>>> Agreed that we should do something about this (i.e., that solution 
> >>>>> 1 is no solution.); that said, a few comments in no particular order:
> >>>>>
> >>>>> 1. In considering adding explicit semantics to structured data, we 
> >>>>> as a WG are taking on the task of defining semantics for IPFIX as 
> >>>>> a whole. Semantics, as I understand them, are in IPFIX largely 
> >>>>> contextual and template dependent, but in almost all cases I can 
> >>>>> think of it seems like these are implicitly "AND" semnantics (this 
> >>>>> flow has source IP A _AND_ destination IP B _AND_...).
> >>>>>
> >>>> Agreed.
> >>>>> We will need to make an explicit statement on this.
> >>>> Not sure why.
> >>>>> We will need to determine whether these implicit semantics are a 
> >>>>> property of the protocol (in which case Structured Data is really 
> >>>>> a protocol-level extension), or a property of each information 
> >>>>> element (in which case all 5103 IEs have implicit AND semantics; 
> >>>>> question 2: will we need to add semantics to the IANA registry in 
> >>>>> this case?). We will need to be quite careful about this. It's not 
> >>>>> as simple as defining semantics within structured data then 
> >>>>> calling it done.
> >>>>>
> >>>>>
> >>>> I'm not sure why each individual IE should have a semantic ... in 
> >>>> the IANA registry.
> >>> I'm not either. On reflection it seems like overkill. I'm just 
> >>> saying that if we, as a WG, are moving from stating that semantics 
> >>> are explicitly out of scope to defining them as in-scope, we need to 
> >>> have a consistent approach, and answers to all the questions that 
> >>> arise when we consider moving the protocol from a simple framing 
> >>> mechanism to a framing mechanism with some logic behind it, so we 
> >>> answer them once, so that all future efforts having to do with 
> >>> semantics are consistent. This, I acknowledge, is an argument in 
> >>> favor of solution 3...)
> >>>
> >>> One very simple new question that arises here, to illustrate my 
> >>> point: Is it legal to export a record that has sourceIPAddress X AND 
> >>> NOT sourceIPAddress X?
> >>  From a protocol point of view, yes
> >>  From a semantic point of view, I don't see a use case for that.
> >> Now, this question is not different that: with RFC5101,  is it legal 
> >> to export a record that has two instances of sourceIPAddress?
> >>>> As you wrote, "in almost all cases I can think of it seems like 
> >>>> these are implicitly "AND" semnantics", so the case we try to solve 
> >>>> is when there are multiple instances of a single IE. We know that 
> >>>> RFC 5101 foresaw that case " The Collector MUST support the use of 
> >>>> Templates containing multiple occurrences of the similar 
> >>>> Information Elements", but the idea is not to change RFC5101.  If 
> >>>> we put some semantic in the IPFIX structured data, that would solve 
> >>>> the vast majority of our cases. Also, we could say: if you want 
> >>>> some semantic when exporting multiple IEs, then you SHOULD use 
> >>>> IPFIX structured data.
> >>>>
> >>>> Also, the default value for the semantic field in the IPFIX 
> >>>> structured data SHOULD be "NONE", to express that flow record 
> >>>> doesn't include any semantic.... like in RFC5101. You might draw 
> >>>> your own conclusion, maybe because you know your network, maybe 
> >>>> because you have configured the exporter, but then it's your decision.
> >>>> The way I see the proposed solution is: in IPFIX structured data, 
> >>>> you MAY use the semantic field as a way to express the relationship 
> >>>> between IEs within the structure.
> >>>>> 2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid 
> >>>>> argument against Solution 2, as it's trying to solve a somewhat 
> >>>>> different and more limited problem than structured data. Solution 
> >>>>> 2 _might_ cause a problem for this draft, but certainly not the 
> >>>>> other way around (unless we as a WG want to subsume that work into 
> >>>>> this draft, which would probably require rechartering...). Also, I 
> >>>>> don't think we need semantics for all of the list types, just 
> >>>>> basicList (Illustrative question: How do I meaningfully interpret 
> >>>>> two "or" subTemplateMultiLists with disjoint IE sets? Two "and" 
> >>>>> subTemplateMultiLists? Nestings thereof?)... So, we don't really 
> >>>>> have an explosion to deal with if we do Solution 2 correctly: 
> >>>>> andBasicList, orBasicList, xorBasicList, notBasicList; Four IEs, 
> >>>>> nestable, done. What can't we do with those four IEs? In this 
> >>>>> case, we could even step back and say that semantics outside these 
> >>>>> four are within the protocol explicitly _undefined_, and to be 
> >>>>> interpreted withi
> > n the
> >>>>>   context of each Template.
> >>>>>
> >>>>>
> >>>> If I translate some more what I wrote "While it's solved the router 
> >>>> and most mediation function needs today", this would be "Me, 
> >>>> myself, and I don't need more that OR and AND now ;-)".
> >>>> However, who am I to tell that others don't need it now... and that 
> >>>> the logical solution is to use the IPFIX structured data
> >>>> Furthermore, if we think a little bit longer term, the next big 
> >>>> step in IPFIX is the mediation function. In my company, every 
> >>>> features want to export his own data with NetFlow/IPFIX... up to 
> >>>> the point where a CPE would not have enough bandwidth across the 
> >>>> WAN to export all the "management" information. So we'll need more 
> >>>> and more of aggregated flow records (both in time and space) even 
> >>>> in the router. Again, the logical solution will be to use the IPFIX 
> >>>> structured data. At this point, we will most probably need 
> >>>> something else than OR and AND, i.e. RANGE, ORDERRED, etc...
> >>>>
> >>>> An example of subTemplateMultiLists with disjoint IE sets, let's 
> >>>> imagine that you have to export an aggregated observation point, 
> >>>> composed of multiple template records
> >>>>      template 1: exporterIPaddress
> >>>>      template 2: exporterIPaddress, basicList of interfaces
> >>>>      template 3: exporterIPaddress, LC
> >>> and then you'd want to OR these... Okay, makes sense...
> >>>
> >>>>> 3. If we really _do_ want ranges and so on (which, again, we'd 
> >>>>> need to get WG consensus on; this is explicitly out of scope in my 
> >>>>> reading of the present charter), then we could do them in the 
> >>>>> scope of Solution 3.
> >>>> Btw, acknowledging that one day we will have to solve this is good 
> >>>> enough for solution 3. I mean, we don't have to populate the 
> >>>> semantic IANA now... even if that would be more efficient.
> >>>>> However, this seems a little not-quite-fleshed-out-enough for me 
> >>>>> to say whether I like it or not. Could you present an example of 
> >>>>> how you would use the proposed Semantic field to model your 
> >>>>> example? "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR linecard2"
> >>>>>
> >>>>> (FWIW, I would do this with my proposal to Solution 2 as follows:)
> >>>>>
> >>>>> (orBasicList (andBasicList (orBasicList ingressInterface eth1 
> >>>>> eth2) (notBasicList (orBasicList ingressInterface eth3 eth4)) 
> >>>>> (andBasicList lineCardID 2))
> >>>>>
> >>>>>
> >>>> (BasicList, OR, (basicList, AND, (basicList, OR, eth1, eth2), 
> >>>> (basicList, NOR, eth3, eth4)), (basicList, NONE, linecard2)))
> >>> Hm. Okay. This makes sense. A couple of questions then, about 
> >>> solution 3:
> >>>
> >>> 1. Would we want to try and bitfield this, to define the semantics 
> >>> we _know_ we have, then leave the rest of it reserved? Something like:
> >>>
> >>> MSb                      LSb
> >>> +---+------+---------------+
> >>> | ! | multi| reserved      |
> >>> +---+------+---------------+
> >>>
> >>> ! = negate sense of semantics if 1 (this is the NOT flag)
> >>> multi (multiplicity) = 00: undefined, 01: or/oneOrMore, 10: 
> >>> xor/exactlyOne, 11: and/exactlyAll
> >>> reserved = place for adding other bells and whistles like RANGE and 
> >>> so on in the future.
> >> Maybe we want to start by asking the question: which semantic do we 
> >> need now?
> >> Is a need for NONE, OR, AND, ORDERED for now. Anything else?
> >>> 2. Are we sure we want to do this in one byte? If we do bitfielding 
> >>> as above this gives us 32 possible extensions, which seems like 
> >>> _way_ more than enough, but does stick another odd offset in there, 
> >>> which slows things down on machines that need aligned access. 
> >>> Probably one byte is okay and we let the implementation use 
> >>> paddingOctets and set padding to fix this...
> >> With structured data, is the alignment still important? When I look 
> >> at the examples throughout the draft... As you wrote, paddingOctets 
> >> might be the solution in this case.
> >>> 3. Would it make sense maybe to have two sets of structured data 
> >>> elements, one with the semantics byte, and one without (which is 
> >>> then explicitly undefined)? Then exporters who don't need it don't 
> >>> have to bother sticking an extra odd-aligned zero in the stream for 
> >>> every list.
> >>>
> >>> I'm sure I'l have more questions, but none come to mind now... But 
> >>> it seems like we're converging on the least-unnecessarily-complex 
> >>> solution here, which is good. :)
> >> Happy about that. ;-)
> >> Note: I was envisioning even something more simpler: one byte, 
> >> containing all the semantic possibilities, administered by IANA.  So 
> >> no reserved, no !
> >>
> >> Regards, Benoit.
> >
> 
> _______________________________________________
> IPFIX mailing list
> IPFIX@ietf.org
> https://www.ietf.org/mailman/listinfo/ipfix
> 

--- 
Atsushi KOBAYASHI  <akoba@nttv6.net>
NTT Information Sharing Platform Lab.
tel:+81-(0)422-59-3978 fax:+81-(0)422-59-5637



From bclaise@cisco.com  Thu Mar 18 01:31:39 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 222F03A6AD3 for <ipfix@core3.amsl.com>; Thu, 18 Mar 2010 01:31:39 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.68
X-Spam-Level: 
X-Spam-Status: No, score=-1.68 tagged_above=-999 required=5 tests=[AWL=-0.211,  BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id ESAhegR6W-kS for <ipfix@core3.amsl.com>; Thu, 18 Mar 2010 01:31:37 -0700 (PDT)
Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id 7CFAE3A6AC4 for <ipfix@ietf.org>; Thu, 18 Mar 2010 01:30:51 -0700 (PDT)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2I8V1wx008765; Thu, 18 Mar 2010 09:31:01 +0100 (CET)
Received: from [10.55.43.57] (ams-bclaise-8718.cisco.com [10.55.43.57]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2I8V0Gt009720; Thu, 18 Mar 2010 09:31:00 +0100 (CET)
Message-ID: <4BA1E4C4.3000204@cisco.com>
Date: Thu, 18 Mar 2010 09:31:00 +0100
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: Brian Trammell <trammell@tik.ee.ethz.ch>
References: <4AF73525.8050009@net.in.tum.de> <4AF8F999.3000207@cisco.com>	<F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>	<4B9E33BE.7070907@cisco.com>	<7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch>	<4B9F520E.8060507@cisco.com>	<DB6E59D9-B373-4919-BB58-00EB26014564@tik.ee.ethz.ch>	<4BA0C26D.9070901@cisco.com> <6CD84D0C-333F-4C65-8E41-B1DF3774BA15@tik.ee.ethz.ch>
In-Reply-To: <6CD84D0C-333F-4C65-8E41-B1DF3774BA15@tik.ee.ethz.ch>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 18 Mar 2010 08:31:39 -0000

On 17/03/2010 15:44, Brian Trammell wrote:
> Hi, Benoit,
>
> As always... inline.
>
> On Mar 17, 2010, at 4:52 AM, Benoit Claise wrote:
>
>   =20
>> Hi Brian,
>>     =20
>>> Hi, Benoit,
>>>
>>> Replies inline...
>>>
>>> On Mar 16, 2010, at 2:40 AM, Benoit Claise wrote:
>>>
>>>
>>>       =20
>>>> Hi Brian,
>>>>
>>>> Thanks for your feedback.
>>>>
>>>>         =20
>>>>> hi, Benoit, all...
>>>>>
>>>>> Agreed that we should do something about this (i.e., that solution =
1 is no solution.); that said, a few comments in no particular order:
>>>>>
>>>>> 1. In considering adding explicit semantics to structured data, we =
as a WG are taking on the task of defining semantics for IPFIX as a whole=
=2E Semantics, as I understand them, are in IPFIX largely contextual and =
template dependent, but in almost all cases I can think of it seems like =
these are implicitly "AND" semnantics (this flow has source IP A _AND_ de=
stination IP B _AND_...).
>>>>>
>>>>>
>>>>>           =20
>>>> Agreed.
>>>>
>>>>         =20
>>>>> We will need to make an explicit statement on this.
>>>>>
>>>>>           =20
>>>> Not sure why.
>>>>         =20
> Because if we're adding semantics to the protocol in any way (which we =
have so far left out of scope), anything we say in structured data (stand=
ards track) may be taken to apply to 5101 without structured data. So a s=
tatement that says "semantic interpretations of recieved Data Records at =
or beyond the CP remain explicitly undefined, unless that data is transmi=
tted using this extension with explicit semantic information" (or somethi=
ng like it) reduces the scope for confusion.
>   =20
Ok with the sentence.
I thought you wanted to add a sentence such as "RFC 5101 =3D implicit=20
AND", which I would not agree with.
>   =20
>>>>
>>>>         =20
>>>>> We will need to determine whether these implicit semantics are a pr=
operty of the protocol (in which case Structured Data is really a protoco=
l-level extension), or a property of each information element (in which c=
ase all 5103 IEs have implicit AND semantics; question 2: will we need to=
 add semantics to the IANA registry in this case?). We will need to be qu=
ite careful about this. It's not as simple as defining semantics within s=
tructured data then calling it done.
>>>>>
>>>>>
>>>>>
>>>>>           =20
>>>> I'm not sure why each individual IE should have a semantic ... in th=
e IANA registry.
>>>>
>>>>         =20
>>> I'm not either. On reflection it seems like overkill. I'm just saying=
 that if we, as a WG, are moving from stating that semantics are explicit=
ly out of scope to defining them as in-scope, we need to have a consisten=
t approach, and answers to all the questions that arise when we consider =
moving the protocol from a simple framing mechanism to a framing mechanis=
m with some logic behind it, so we answer them once, so that all future e=
fforts having to do with semantics are consistent. This, I acknowledge, i=
s an argument in favor of solution 3...)
>>>
>>> One very simple new question that arises here, to illustrate my point=
: Is it legal to export a record that has sourceIPAddress X AND NOT sourc=
eIPAddress X?
>>>
>>>       =20
>>  From a protocol point of view, yes
>>  From a semantic point of view, I don't see a use case for that.
>> Now, this question is not different that: with RFC5101,  is it legal t=
o export a record that has two instances of sourceIPAddress?
>>     =20
> Point. Okay, yes, it is legal. So in Structured Data we should have a s=
tatement as well that semantically meaningless combinations are legal, an=
d up to the CP to interpret.
>   =20
Right.
However, we should try to minimize this likelihood.
>   =20
>>>> As you wrote, "in almost all cases I can think of it seems like thes=
e are implicitly "AND" semnantics", so the case we try to solve is when t=
here are multiple instances of a single IE. We know that RFC 5101 foresaw=
 that case " The Collector MUST support the use of Templates containing m=
ultiple occurrences of the similar Information Elements", but the idea is=
 not to change RFC5101.  If we put some semantic in the IPFIX structured =
data, that would solve the vast majority of our cases. Also, we could say=
: if you want some semantic when exporting multiple IEs, then you SHOULD =
use IPFIX structured data.
>>>>
>>>> Also, the default value for the semantic field in the IPFIX structur=
ed data SHOULD be "NONE", to express that flow record doesn't include any=
 semantic.... like in RFC5101. You might draw your own conclusion, maybe =
because you know your network, maybe because you have configured the expo=
rter, but then it's your decision.
>>>> The way I see the proposed solution is: in IPFIX structured data, yo=
u MAY use the semantic field as a way to express the relationship between=
 IEs within the structure.
>>>>
>>>>         =20
>>>>> 2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid argu=
ment against Solution 2, as it's trying to solve a somewhat different and=
 more limited problem than structured data. Solution 2 _might_ cause a pr=
oblem for this draft, but certainly not the other way around (unless we a=
s a WG want to subsume that work into this draft, which would probably re=
quire rechartering...). Also, I don't think we need semantics for all of =
the list types, just basicList (Illustrative question: How do I meaningfu=
lly interpret two "or" subTemplateMultiLists with disjoint IE sets? Two "=
and" subTemplateMultiLists? Nestings thereof?)... So, we don't really hav=
e an explosion to deal with if we do Solution 2 correctly: andBasicList, =
orBasicList, xorBasicList, notBasicList; Four IEs, nestable, done. What c=
an't we do with those four IEs? In this case, we could even step back and=
 say that semantics outside these four are within the protocol explicitly=
 _undefined_, and to be interpreted within the
>>>>>   context of each Template.
>>>>>
>>>>>
>>>>>
>>>>>           =20
>>>> If I translate some more what I wrote "While it's solved the router =
and most mediation function needs today", this would be "Me, myself, and =
I don't need more that OR and AND now ;-)".
>>>> However, who am I to tell that others don't need it now... and that =
the logical solution is to use the IPFIX structured data
>>>> Furthermore, if we think a little bit longer term, the next big step=
 in IPFIX is the mediation function. In my company, every features want t=
o export his own data with NetFlow/IPFIX... up to the point where a CPE w=
ould not have enough bandwidth across the WAN to export all the "manageme=
nt" information. So we'll need more and more of aggregated flow records (=
both in time and space) even in the router. Again, the logical solution w=
ill be to use the IPFIX structured data. At this point, we will most prob=
ably need something else than OR and AND, i.e. RANGE, ORDERRED, etc...
>>>>
>>>> An example of subTemplateMultiLists with disjoint IE sets, let's ima=
gine that you have to export an aggregated observation point, composed of=
 multiple template records
>>>>      template 1: exporterIPaddress
>>>>      template 2: exporterIPaddress, basicList of interfaces
>>>>      template 3: exporterIPaddress, LC
>>>>
>>>>         =20
>>> and then you'd want to OR these... Okay, makes sense...
>>>
>>>
>>>       =20
>>>>
>>>>         =20
>>>>> 3. If we really _do_ want ranges and so on (which, again, we'd need=
 to get WG consensus on; this is explicitly out of scope in my reading of=
 the present charter), then we could do them in the scope of Solution 3.
>>>>>
>>>>>           =20
>>>> Btw, acknowledging that one day we will have to solve this is good e=
nough for solution 3. I mean, we don't have to populate the semantic IANA=
 now... even if that would be more efficient.
>>>>
>>>>         =20
>>>>> However, this seems a little not-quite-fleshed-out-enough for me to=
 say whether I like it or not. Could you present an example of how you wo=
uld use the proposed Semantic field to model your example? "(eth1 OR eth2=
) AND (NOT (eth3 OR eth4)) OR linecard2"
>>>>>
>>>>> (FWIW, I would do this with my proposal to Solution 2 as follows:)
>>>>>
>>>>> (orBasicList (andBasicList (orBasicList ingressInterface eth1 eth2)=
 (notBasicList (orBasicList ingressInterface eth3 eth4)) (andBasicList li=
neCardID 2))
>>>>>
>>>>>
>>>>>
>>>>>           =20
>>>> (BasicList, OR, (basicList, AND, (basicList, OR, eth1, eth2), (basic=
List, NOR, eth3, eth4)), (basicList, NONE, linecard2)))
>>>>
>>>>         =20
>>> Hm. Okay. This makes sense. A couple of questions then, about solutio=
n 3:
>>>
>>> 1. Would we want to try and bitfield this, to define the semantics we=
 _know_ we have, then leave the rest of it reserved? Something like:
>>>
>>> MSb                      LSb
>>> +---+------+---------------+
>>> | ! | multi| reserved      |
>>> +---+------+---------------+
>>>
>>> ! =3D negate sense of semantics if 1 (this is the NOT flag)
>>> multi (multiplicity) =3D 00: undefined, 01: or/oneOrMore, 10: xor/exa=
ctlyOne, 11: and/exactlyAll
>>> reserved =3D place for adding other bells and whistles like RANGE and=
 so on in the future.
>>>
>>>       =20
>> Maybe we want to start by asking the question: which semantic do we ne=
ed now?
>> Is a need for NONE, OR, AND, ORDERED for now. Anything else?
>>     =20
> What's ORDERED for? It seems like we're getting multiplicities confused=
 with other things here... Is this an AND/allOf with additional stipulati=
on that the ordering is important?
>   =20
An example, as discussed by Kobayashi-san is the AS-PATH.
>   =20
>>> 2. Are we sure we want to do this in one byte? If we do bitfielding a=
s above this gives us 32 possible extensions, which seems like _way_ more=
 than enough, but does stick another odd offset in there, which slows thi=
ngs down on machines that need aligned access. Probably one byte is okay =
and we let the implementation use paddingOctets and set padding to fix th=
is...
>>>
>>>       =20
>> With structured data, is the alignment still important? When I look at=
 the examples throughout the draft... As you wrote, paddingOctets might b=
e the solution in this case.
>>     =20
>>> 3. Would it make sense maybe to have two sets of structured data elem=
ents, one with the semantics byte, and one without (which is then explici=
tly undefined)? Then exporters who don't need it don't have to bother sti=
cking an extra odd-aligned zero in the stream for every list.
>>>
>>> I'm sure I'l have more questions, but none come to mind now... But it=
 seems like we're converging on the least-unnecessarily-complex solution =
here, which is good. :)
>>>
>>>       =20
>> Happy about that. ;-)
>> Note: I was envisioning even something more simpler: one byte, contain=
ing all the semantic possibilities, administered by IANA.  So no reserved=
, no !
>>     =20
> Okay, then... by "reserved" I meant, "future stuff (IANA) goes here"...=
 but yes, that's more complicated than the rest of the subregistries for =
the protocol. So, agreed.
>
> On reflection, I would suggest defining these in terms of multiplicitie=
s (as we're dealing with N-element lists) as opposed to logical operation=
s. So I'd propose the following:
>
> 0x00: undefined (default) (CP may assume allOf)
>   =20
This corresponds to NONE
> 0x01: allOf
>   =20
This corresponds to AND
> 0x02: oneOrMoreOf
>   =20
This corresponds to OR
> 0x03: exactlyOneOf
>   =20
Not sure what it corresponds to. At first glance, I don't see a use case =

for it.
> 0x04: noneOf
>   =20
This corresponds to NOT
> If I understand what you mean by ORDERED, add:
> 0x05: allInOrder
>   =20
This corresponds to ORDERED
> If we really want to do ranges:
> 0x06: inclusiveRanges (MUST have even number of elements, and elements =
MUST be numeric; this is then equivalent to inclusive allOf [e_0,e_1], [e=
_2,e_3], ... , [e_n-1]..[e_n]);
>   =20
This corresponds to RANGE

What is required? NONE, AND, OR, potentially NOT, ORDERED

> Still unanswered is whether we want to force all basicList/subTemplateL=
ist/subTemplateMultiList instances to state semantics, or whether we want=
 "semanticless" versions as well...
>   =20
No strong feeling about this one.
Maybe it would be better to have additional semanticBasicList,=20
semanticSubTemplateList, semanticSubTemplateMultiList

Regards, Benoit.
> Best regards,
>
> Brian
>
>
>   =20
>>>>> On Mar 15, 2010, at 6:18 AM, Benoit Claise wrote:
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>           =20
>>>>>> Dear all,
>>>>>>
>>>>>> We've been thinking about this one, and we see 3 solutions. We bel=
ieve that the solution 3 is the way to go, as we show below.
>>>>>>
>>>>>> Solution 1.
>>>>>> Consider that the semantic is out of scope for this document.
>>>>>> This is the easiest solution.  However, we understand it's not rig=
ht: we would not do a complete job wrt to IPFIX structured data
>>>>>> As Gerhard was expressing, the collector has not clue how to treat=
 the example of a BasicList of egress interfaces in a Flow Record.
>>>>>>      - Has every counted packet been sent on every egress interfac=
e?
>>>>>>        =3D>   multicast case, AND semantic
>>>>>>      - Has every counted packet been sent on any one of the egress=
 interfaces?
>>>>>>       =3D>   load balancing case, OR semantic
>>>>>>
>>>>>> Soltution 2
>>>>>> We could focus on the logical AND and OR semantics only , by defin=
ing semantic lists, such as andBasicList, andSubTemplateList, andSubTempl=
ateMultiList, orBasicList, orSubTemplateList, and orSubTemplateMultiList
>>>>>> So 6 I.E.s in total, describing AND and OR semantics.
>>>>>> While it's solved the router and most mediation function needs tod=
ay, we understand that this is not a complete solution.
>>>>>> Gerhard, in one of his old draft, draft-sommer-ipfix-mediator-ext-=
01, proposed ADTs for orderedList, orderedPair, and portRanges which allo=
w the definition new IEs for port ranges etc.
>>>>>> Even if this solution 2 is extensible, in the long term, will lead=
 to an explosion of IEs: 3 list types * the semantic type, where the sema=
ntic can be (and, or, orderedList, orderedPair, portRanges, random, etc..=
=2E)
>>>>>>
>>>>>> Soltution 3
>>>>>> We propose to add a semantic field in the 3 list types.
>>>>>> Something such as:
>>>>>>
>>>>>>      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+=
-+-+
>>>>>> |0|               Field ID    |       Element Length            |
>>>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>>> | Semantic  |             BasicList Content ...                 |
>>>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>>> |                           ...                                 |
>>>>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>>>>
>>>>>> This semantic field would be a new IANA registry, we could be popu=
lated initially with NONE, OR, AND, ORDERED, etc... Up to discussion.
>>>>>> The advantage of this solution is that it's extensible, and it doe=
sn't need new IEs for each semantic.
>>>>>> If ever required (mainly in a mediation function), we can model  v=
ery complex semantic, eg, "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR lin=
ecard2", which could be the new observation point from a aggregated Flow =
Record
>>>>>>
>>>>>>
>>>>>>
>>>>>> Conclusion:
>>>>>> After much debate internally, we believe that we should do the eff=
ort to include this semantic now, and not postpone the problem, which wil=
l lead to an explosion of IEs in the future.
>>>>>> Personally, I was initially against the solution 3 ... mostly due =
to the effort required to modify the complete specs.
>>>>>> We're now ready to modify the specifications in the IPFIX structur=
ed data, but we would like to get your feedback and agreement in advance =
as this is not a small piece of work.
>>>>>>
>>>>>> Please comment.
>>>>>>
>>>>>> Regards, Paul, Stan, Gowri, and Benoit.
>>>>>>
>>>>>>
>>>>>>
>>>>>>             =20
>>>>>>> hi Benoit, Gerhard,
>>>>>>>
>>>>>>> In this case I'm strongly in favor of leaving semantics out of st=
ructured data. Structured data defines containers. The semantics of the c=
ontainers as a whole and the elements change based upon the information e=
lements within the container and within the record containing it. Wedging=
 semantics into the structured data elements 1. risks further proliferati=
on of (potentially non-interoperable) ways to represent the same thing, 2=
=2E gives us more ways to represent nonsensical things (an OR basicList o=
f MPLS stack entries...means...what?), and 3. risks defining an inadequat=
e semantic representation mechanism (what about semantics for records not=
 using structured data? what about ordered versus unordered sets? what ab=
out OR basicList vs AND basicList vs nested AND and OR basicLists vs just=
 three identical IEs?). If we really want an unambiguous semantic framewo=
rk for IPFIX (and here I'm not convinced either way) that's best done on =
its own, addressing things at the information elem
>>>>>>> ent and record level. Doing it here confuses the issue.
>>>>>>>
>>>>>>> Regards,
>>>>>>>
>>>>>>> Brian
>>>>>>>
>>>>>>> On Nov 10, 2009, at 6:26 AM, Benoit Claise wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>               =20
>>>>>>>> Gerhard,
>>>>>>>>
>>>>>>>> Thanks for your email.
>>>>>>>> I have no strong feelings about the two solutions you proposed.
>>>>>>>>
>>>>>>>>                 =20
>>>>>>>>>  From a pure router point of view,  I don't see any use cases f=
or logical OR in exporting flow records.
>>>>>>>>>                   =20
>>>>>>>> However, from a IPFIX Mediator point of view, I see some use cas=
es.
>>>>>>>> I mean that it requires an Intermediate Aggregation Process or I=
ntermediate Correlation Process to express: I've seen this flow record OR=
 that flow record.
>>>>>>>> Now, it's true that even routers will have mediation functions..=
=2E
>>>>>>>>
>>>>>>>> I'm inclined to add orBasicList, orSubTemplateList, orSubTempalt=
eMultiList to the draft (This is a small addition after all) and to expre=
ss that, by default, a logical AND is assumed... specifically if the stru=
ctured data is used in the IPFIX Mediation Protocol.
>>>>>>>>
>>>>>>>> I'm not convinced by the NOT in the context of structured data, =
as we don't even have a concept of NOT for a single information element!
>>>>>>>>
>>>>>>>> I would like to get some more feedback from others.
>>>>>>>>
>>>>>>>> Regards, Benoit.
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>                 =20
>>>>>>>>> Dear all,
>>>>>>>>>
>>>>>>>>> Regarding draft-ietf-ipfix-structured-data, I see the risk that=
 the
>>>>>>>>> semantic of the exported structured data is not clear.
>>>>>>>>>
>>>>>>>>> How do you interpret the manifold occurrence of the same Inform=
ation
>>>>>>>>> Element (basicList) or the same group of Information Elements
>>>>>>>>> (subTemplateList) in one record?
>>>>>>>>>
>>>>>>>>> What does it mean if basicList, subTemplateList, or subTemplate=
MultiList
>>>>>>>>> is used for a Flow Key field? Or non-key field?
>>>>>>>>>
>>>>>>>>> Some Examples:
>>>>>>>>>
>>>>>>>>> - BasicList of egress interfaces in a Flow Record
>>>>>>>>>   How should a Flow Record be interpreted which contains a list=
 of
>>>>>>>>>   egress interfaces and a packet counter?
>>>>>>>>>   Has every counted packet been sent on every egress interface?=

>>>>>>>>>     =3D>   multicast case, AND semantic (see example in section=
 8.1)
>>>>>>>>>   Has every counted packet been sent on any one of the egress
>>>>>>>>>   interfaces?
>>>>>>>>>     =3D>   load balancing case, OR semantic
>>>>>>>>>   Can it be used as a Flow Key or not?
>>>>>>>>>
>>>>>>>>> - BasicList of destination ports in a Flow Record
>>>>>>>>>   As every packet has only one destination port, the only reaso=
nable
>>>>>>>>>   interpretation is that the Flow contains packets having one o=
f
>>>>>>>>>   the reported port numbers.
>>>>>>>>>     =3D>   OR semantic
>>>>>>>>>   This would be a non-key field.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> I think there are two solutions:
>>>>>>>>>
>>>>>>>>> 1. We decide that the semantic of list content is out of scope =
of
>>>>>>>>>    draft-ietf-ipfix-structured-data. We add a note to the draft=
 that
>>>>>>>>>    the semantic must be clear from the context or the definitio=
n of the
>>>>>>>>>    Information Elements used within the lists.
>>>>>>>>>
>>>>>>>>> 2. We define semantic lists, such as
>>>>>>>>>    - andBasicList, andSubTemplateList, andSubTemplateMultiList
>>>>>>>>>    - orBasicList, orSubTemplateList, orSubTempalteMultiList
>>>>>>>>>    describing AND and OR semantic of the contained IEs/Template=
s,
>>>>>>>>>    respectively.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> As I wrote in an earlier mail, I see a good use case for orBasi=
cList. It
>>>>>>>>> could be used in the Selector Report Interpretation of Property=
 Match
>>>>>>>>> Filtering to report a filter like "port 80 or port 443".
>>>>>>>>>
>>>>>>>>> http://www.ietf.org/mail-archive/web/ipfix/current/msg04856.htm=
l
>>>>>>>>>
>>>>>>>>> At the moment, the Selector Report Interpretation is limited to=
 AND.
>>>>>>>>> However, if we also want to express a NOT, we still need anothe=
r solution...
>>>>>>>>>
>>>>>>>>> Regards,
>>>>>>>>> Gerhard
>>>>>>>>>                   =20



From bclaise@cisco.com  Thu Mar 18 01:56:06 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id D14D93A6867 for <ipfix@core3.amsl.com>; Thu, 18 Mar 2010 01:56:06 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.666
X-Spam-Level: 
X-Spam-Status: No, score=-1.666 tagged_above=-999 required=5 tests=[AWL=-0.198, BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, HTML_MESSAGE=0.001]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id aSlsINGhdZAy for <ipfix@core3.amsl.com>; Thu, 18 Mar 2010 01:56:04 -0700 (PDT)
Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id 33A7D3A6831 for <ipfix@ietf.org>; Thu, 18 Mar 2010 01:56:04 -0700 (PDT)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2I8qSxH016468; Thu, 18 Mar 2010 09:52:28 +0100 (CET)
Received: from [10.55.43.57] (ams-bclaise-8718.cisco.com [10.55.43.57]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2I8qLDY004088; Thu, 18 Mar 2010 09:52:23 +0100 (CET)
Message-ID: <4BA1E9C4.9030204@cisco.com>
Date: Thu, 18 Mar 2010 09:52:20 +0100
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: Atsushi Kobayashi <akoba@nttv6.net>
References: <4BA0D177.1070506@net.in.tum.de> <4BA0D4CB.6000308@cisco.com> <20100318051549.AFC9.17391CF2@nttv6.net>
In-Reply-To: <20100318051549.AFC9.17391CF2@nttv6.net>
Content-Type: multipart/alternative; boundary="------------040601060409040805080109"
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 18 Mar 2010 08:56:06 -0000

This is a multi-part message in MIME format.
--------------040601060409040805080109
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Kobayashi-san,
> Hi Benoit, and all,
>
> I agree solution3 rather than new semantic IE.
> As Benoit mentioned, semantic IE seems to increase the difficulty to
> interpret them.
>
> As Brian mentioned, I think it should avoid to have unnecessary
> complexity. But, I am not sure what type, e.g., NONE, OR, AND, NOR,
> RANGE, and ORDERRED, is needed. I would like to avoid that one data
> structure is represented in multiple ways.
So do I.
 From this, I conclude that a simple enumeration type is better compared 
to the bitvector encoding.
If we include a bitvector with one bit for NOT, we end up in this situation.

    NOT OR = NOR
    NOT AND = NAND
    NOT ORDERED = UNORDERED (*)
    NOT UNORDERED = ORDERED (*)
    NOT NONE => doesn't make sense  (**)
    NOT RANGE => doesn't make sense  (**)
    NOT NOT => doesn't make sense  (**)


(*) implies that we can represent the same semantic different ways. Now, 
we can argue that we don't need UNORDERED
(**) add non possible and confusing semantics
> To examine it, we needs more
> practical examples in addition to: Selector Report
> Interpretation and interface lists on aggregated Flow Records.
>
> How about BGP AS Path, Community? Anything else?
>
> When I have the following BGP AS Path mixing as-sequence and as-set, how
> to present it?
>
> 10 20 30 40 {50,60}
>
> (basicList, ORDERED, (basicList, ORDERED, AS10,AS20,AS30,AS40),
> (basicList, OR, AS50, AS60))
>
> Is it correct?
>    
Exactly, which implies that the ORDER semantic is required for this use 
case.

Regards, Benoit.
> Regards,
> Atsushi
>
> On Wed, 17 Mar 2010 14:10:35 +0100
> Benoit Claise<bclaise@cisco.com>  wrote:
>
>    
>> Thanks Gerhard for your feedback.
>> See inline.
>>      
>>> Hi all,
>>>
>>> Some general thoughts from my side:
>>>
>>> - I appreciate that you want to add a basic notion of semantic to the
>>>    structured data.
>>>        
>> Great.
>>      
>>> - Up to now, semantic was not in the protocol but in the info model.
>>>
>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>> |0|               Field ID    |       Element Length            |
>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>> | Semantic  |             BasicList Content ...                 |
>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>> |                           ...                                 |
>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>
>>>    Can't we encode Semantic in an IE?
>>>    Then, we could do without a new IANA registry.
>>>    And: This IE could be used for other purposes (e.g., in Templates)
>>>        
>> We thought about this one. However, one of the IPFIX principle is that
>> the semantic of one IE can't depend on the value or position of some
>> other IEs.
>> One example of this is the (MPLS label position, MPLS label). Way before
>> structured data, our initial implementation contained: MPLS label
>> position, MPLS label.
>> So the collector had to look first at the MPLS label position value in
>> order to correctly understand the following  MPLS label.
>> Somehow, with these conventions, we were having an information model on
>> the top of the IE. This was wrong.
>> Also it was not obvious for a collector, i.e. without hardcoding the
>> information.
>> Finally, what if the IE order changed within the flow record?
>> Conclusion: we had to change back our implementation mplslabelposition1,
>> mplslabelposition2, etc... Like we did in IPIFX ;-)
>>
>>
>>      
>>> - I'm not sure whether the bitvector encoding has an advantage over an
>>>    enumeration type.
>>>
>>> Unfortunately, I do not have the time to participate in a deep
>>> discussion - I'm busy with IPFIX-CONFIG and other stuff.
>>>        
>> Thanks and regards, Benoit.
>>      
>>> Regards,
>>> Gerhard
>>>
>>> Benoit Claise wrote:
>>>        
>>>> Hi Brian,
>>>>          
>>>>> Hi, Benoit,
>>>>>
>>>>> Replies inline...
>>>>>
>>>>> On Mar 16, 2010, at 2:40 AM, Benoit Claise wrote:
>>>>>
>>>>>            
>>>>>> Hi Brian,
>>>>>>
>>>>>> Thanks for your feedback.
>>>>>>              
>>>>>>> hi, Benoit, all...
>>>>>>>
>>>>>>> Agreed that we should do something about this (i.e., that solution
>>>>>>> 1 is no solution.); that said, a few comments in no particular order:
>>>>>>>
>>>>>>> 1. In considering adding explicit semantics to structured data, we
>>>>>>> as a WG are taking on the task of defining semantics for IPFIX as
>>>>>>> a whole. Semantics, as I understand them, are in IPFIX largely
>>>>>>> contextual and template dependent, but in almost all cases I can
>>>>>>> think of it seems like these are implicitly "AND" semnantics (this
>>>>>>> flow has source IP A _AND_ destination IP B _AND_...).
>>>>>>>
>>>>>>>                
>>>>>> Agreed.
>>>>>>              
>>>>>>> We will need to make an explicit statement on this.
>>>>>>>                
>>>>>> Not sure why.
>>>>>>              
>>>>>>> We will need to determine whether these implicit semantics are a
>>>>>>> property of the protocol (in which case Structured Data is really
>>>>>>> a protocol-level extension), or a property of each information
>>>>>>> element (in which case all 5103 IEs have implicit AND semantics;
>>>>>>> question 2: will we need to add semantics to the IANA registry in
>>>>>>> this case?). We will need to be quite careful about this. It's not
>>>>>>> as simple as defining semantics within structured data then
>>>>>>> calling it done.
>>>>>>>
>>>>>>>
>>>>>>>                
>>>>>> I'm not sure why each individual IE should have a semantic ... in
>>>>>> the IANA registry.
>>>>>>              
>>>>> I'm not either. On reflection it seems like overkill. I'm just
>>>>> saying that if we, as a WG, are moving from stating that semantics
>>>>> are explicitly out of scope to defining them as in-scope, we need to
>>>>> have a consistent approach, and answers to all the questions that
>>>>> arise when we consider moving the protocol from a simple framing
>>>>> mechanism to a framing mechanism with some logic behind it, so we
>>>>> answer them once, so that all future efforts having to do with
>>>>> semantics are consistent. This, I acknowledge, is an argument in
>>>>> favor of solution 3...)
>>>>>
>>>>> One very simple new question that arises here, to illustrate my
>>>>> point: Is it legal to export a record that has sourceIPAddress X AND
>>>>> NOT sourceIPAddress X?
>>>>>            
>>>>    From a protocol point of view, yes
>>>>   From a semantic point of view, I don't see a use case for that.
>>>> Now, this question is not different that: with RFC5101,  is it legal
>>>> to export a record that has two instances of sourceIPAddress?
>>>>          
>>>>>> As you wrote, "in almost all cases I can think of it seems like
>>>>>> these are implicitly "AND" semnantics", so the case we try to solve
>>>>>> is when there are multiple instances of a single IE. We know that
>>>>>> RFC 5101 foresaw that case " The Collector MUST support the use of
>>>>>> Templates containing multiple occurrences of the similar
>>>>>> Information Elements", but the idea is not to change RFC5101.  If
>>>>>> we put some semantic in the IPFIX structured data, that would solve
>>>>>> the vast majority of our cases. Also, we could say: if you want
>>>>>> some semantic when exporting multiple IEs, then you SHOULD use
>>>>>> IPFIX structured data.
>>>>>>
>>>>>> Also, the default value for the semantic field in the IPFIX
>>>>>> structured data SHOULD be "NONE", to express that flow record
>>>>>> doesn't include any semantic.... like in RFC5101. You might draw
>>>>>> your own conclusion, maybe because you know your network, maybe
>>>>>> because you have configured the exporter, but then it's your decision.
>>>>>> The way I see the proposed solution is: in IPFIX structured data,
>>>>>> you MAY use the semantic field as a way to express the relationship
>>>>>> between IEs within the structure.
>>>>>>              
>>>>>>> 2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid
>>>>>>> argument against Solution 2, as it's trying to solve a somewhat
>>>>>>> different and more limited problem than structured data. Solution
>>>>>>> 2 _might_ cause a problem for this draft, but certainly not the
>>>>>>> other way around (unless we as a WG want to subsume that work into
>>>>>>> this draft, which would probably require rechartering...). Also, I
>>>>>>> don't think we need semantics for all of the list types, just
>>>>>>> basicList (Illustrative question: How do I meaningfully interpret
>>>>>>> two "or" subTemplateMultiLists with disjoint IE sets? Two "and"
>>>>>>> subTemplateMultiLists? Nestings thereof?)... So, we don't really
>>>>>>> have an explosion to deal with if we do Solution 2 correctly:
>>>>>>> andBasicList, orBasicList, xorBasicList, notBasicList; Four IEs,
>>>>>>> nestable, done. What can't we do with those four IEs? In this
>>>>>>> case, we could even step back and say that semantics outside these
>>>>>>> four are within the protocol explicitly _undefined_, and to be
>>>>>>> interpreted withi
>>>>>>>                
>>> n the
>>>        
>>>>>>>    context of each Template.
>>>>>>>
>>>>>>>
>>>>>>>                
>>>>>> If I translate some more what I wrote "While it's solved the router
>>>>>> and most mediation function needs today", this would be "Me,
>>>>>> myself, and I don't need more that OR and AND now ;-)".
>>>>>> However, who am I to tell that others don't need it now... and that
>>>>>> the logical solution is to use the IPFIX structured data
>>>>>> Furthermore, if we think a little bit longer term, the next big
>>>>>> step in IPFIX is the mediation function. In my company, every
>>>>>> features want to export his own data with NetFlow/IPFIX... up to
>>>>>> the point where a CPE would not have enough bandwidth across the
>>>>>> WAN to export all the "management" information. So we'll need more
>>>>>> and more of aggregated flow records (both in time and space) even
>>>>>> in the router. Again, the logical solution will be to use the IPFIX
>>>>>> structured data. At this point, we will most probably need
>>>>>> something else than OR and AND, i.e. RANGE, ORDERRED, etc...
>>>>>>
>>>>>> An example of subTemplateMultiLists with disjoint IE sets, let's
>>>>>> imagine that you have to export an aggregated observation point,
>>>>>> composed of multiple template records
>>>>>>       template 1: exporterIPaddress
>>>>>>       template 2: exporterIPaddress, basicList of interfaces
>>>>>>       template 3: exporterIPaddress, LC
>>>>>>              
>>>>> and then you'd want to OR these... Okay, makes sense...
>>>>>
>>>>>            
>>>>>>> 3. If we really _do_ want ranges and so on (which, again, we'd
>>>>>>> need to get WG consensus on; this is explicitly out of scope in my
>>>>>>> reading of the present charter), then we could do them in the
>>>>>>> scope of Solution 3.
>>>>>>>                
>>>>>> Btw, acknowledging that one day we will have to solve this is good
>>>>>> enough for solution 3. I mean, we don't have to populate the
>>>>>> semantic IANA now... even if that would be more efficient.
>>>>>>              
>>>>>>> However, this seems a little not-quite-fleshed-out-enough for me
>>>>>>> to say whether I like it or not. Could you present an example of
>>>>>>> how you would use the proposed Semantic field to model your
>>>>>>> example? "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR linecard2"
>>>>>>>
>>>>>>> (FWIW, I would do this with my proposal to Solution 2 as follows:)
>>>>>>>
>>>>>>> (orBasicList (andBasicList (orBasicList ingressInterface eth1
>>>>>>> eth2) (notBasicList (orBasicList ingressInterface eth3 eth4))
>>>>>>> (andBasicList lineCardID 2))
>>>>>>>
>>>>>>>
>>>>>>>                
>>>>>> (BasicList, OR, (basicList, AND, (basicList, OR, eth1, eth2),
>>>>>> (basicList, NOR, eth3, eth4)), (basicList, NONE, linecard2)))
>>>>>>              
>>>>> Hm. Okay. This makes sense. A couple of questions then, about
>>>>> solution 3:
>>>>>
>>>>> 1. Would we want to try and bitfield this, to define the semantics
>>>>> we _know_ we have, then leave the rest of it reserved? Something like:
>>>>>
>>>>> MSb                      LSb
>>>>> +---+------+---------------+
>>>>> | ! | multi| reserved      |
>>>>> +---+------+---------------+
>>>>>
>>>>> ! = negate sense of semantics if 1 (this is the NOT flag)
>>>>> multi (multiplicity) = 00: undefined, 01: or/oneOrMore, 10:
>>>>> xor/exactlyOne, 11: and/exactlyAll
>>>>> reserved = place for adding other bells and whistles like RANGE and
>>>>> so on in the future.
>>>>>            
>>>> Maybe we want to start by asking the question: which semantic do we
>>>> need now?
>>>> Is a need for NONE, OR, AND, ORDERED for now. Anything else?
>>>>          
>>>>> 2. Are we sure we want to do this in one byte? If we do bitfielding
>>>>> as above this gives us 32 possible extensions, which seems like
>>>>> _way_ more than enough, but does stick another odd offset in there,
>>>>> which slows things down on machines that need aligned access.
>>>>> Probably one byte is okay and we let the implementation use
>>>>> paddingOctets and set padding to fix this...
>>>>>            
>>>> With structured data, is the alignment still important? When I look
>>>> at the examples throughout the draft... As you wrote, paddingOctets
>>>> might be the solution in this case.
>>>>          
>>>>> 3. Would it make sense maybe to have two sets of structured data
>>>>> elements, one with the semantics byte, and one without (which is
>>>>> then explicitly undefined)? Then exporters who don't need it don't
>>>>> have to bother sticking an extra odd-aligned zero in the stream for
>>>>> every list.
>>>>>
>>>>> I'm sure I'l have more questions, but none come to mind now... But
>>>>> it seems like we're converging on the least-unnecessarily-complex
>>>>> solution here, which is good. :)
>>>>>            
>>>> Happy about that. ;-)
>>>> Note: I was envisioning even something more simpler: one byte,
>>>> containing all the semantic possibilities, administered by IANA.  So
>>>> no reserved, no !
>>>>
>>>> Regards, Benoit.
>>>>          
>>>        
>> _______________________________________________
>> IPFIX mailing list
>> IPFIX@ietf.org
>> https://www.ietf.org/mailman/listinfo/ipfix
>>
>>      
> ---
> Atsushi KOBAYASHI<akoba@nttv6.net>
> NTT Information Sharing Platform Lab.
> tel:+81-(0)422-59-3978 fax:+81-(0)422-59-5637
>
>    


--------------040601060409040805080109
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <meta content="text/html; charset=ISO-8859-1"
 http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Kobayashi-san,<br>
<blockquote cite="mid:20100318051549.AFC9.17391CF2@nttv6.net"
 type="cite">
  <pre wrap="">
Hi Benoit, and all,

I agree solution3 rather than new semantic IE. 
As Benoit mentioned, semantic IE seems to increase the difficulty to
interpret them.

As Brian mentioned, I think it should avoid to have unnecessary
complexity. But, I am not sure what type, e.g., NONE, OR, AND, NOR,
RANGE, and ORDERRED, is needed. I would like to avoid that one data
structure is represented in multiple ways. </pre>
</blockquote>
So do I.<br>
>From this, I conclude that a simple enumeration type is better compared
to the bitvector encoding.<br>
If we include a bitvector with one bit for NOT, we end up in this
situation.<br>
<blockquote>NOT OR = NOR
  <br>
NOT AND = NAND
  <br>
NOT ORDERED = UNORDERED (*)
  <br>
NOT UNORDERED = ORDERED (*)
  <br>
NOT NONE =&gt; doesn't make sense&nbsp; (**)
  <br>
NOT RANGE =&gt; doesn't make sense&nbsp; (**)
  <br>
NOT NOT =&gt; doesn't make sense&nbsp; (**)
  <br>
</blockquote>
<br>
(*) implies that we can represent the same semantic different ways.
Now, we can argue that we don't need UNORDERED<br>
(**) add non possible and confusing semantics
<br>
<blockquote cite="mid:20100318051549.AFC9.17391CF2@nttv6.net"
 type="cite">
  <pre wrap="">To examine it, we needs more
practical examples in addition to: Selector Report
Interpretation and interface lists on aggregated Flow Records.

How about BGP AS Path, Community? Anything else?

When I have the following BGP AS Path mixing as-sequence and as-set, how
to present it?

10 20 30 40 {50,60}

(basicList, ORDERED, (basicList, ORDERED, AS10,AS20,AS30,AS40),
(basicList, OR, AS50, AS60))

Is it correct?
  </pre>
</blockquote>
Exactly, which implies that the ORDER semantic is required for this use
case.<br>
<br>
Regards, Benoit.<br>
<blockquote cite="mid:20100318051549.AFC9.17391CF2@nttv6.net"
 type="cite">
  <pre wrap="">
Regards,
Atsushi

On Wed, 17 Mar 2010 14:10:35 +0100
Benoit Claise <a class="moz-txt-link-rfc2396E" href="mailto:bclaise@cisco.com">&lt;bclaise@cisco.com&gt;</a> wrote:

  </pre>
  <blockquote type="cite">
    <pre wrap="">Thanks Gerhard for your feedback.
See inline.
    </pre>
    <blockquote type="cite">
      <pre wrap="">
Hi all,

Some general thoughts from my side:

- I appreciate that you want to add a basic notion of semantic to the
  structured data.
      </pre>
    </blockquote>
    <pre wrap="">Great.
    </pre>
    <blockquote type="cite">
      <pre wrap="">
- Up to now, semantic was not in the protocol but in the info model.

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|0|               Field ID    |       Element Length            |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Semantic  |             BasicList Content ...                 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                           ...                                 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

  Can't we encode Semantic in an IE?
  Then, we could do without a new IANA registry.
  And: This IE could be used for other purposes (e.g., in Templates)
      </pre>
    </blockquote>
    <pre wrap="">We thought about this one. However, one of the IPFIX principle is that 
the semantic of one IE can't depend on the value or position of some 
other IEs.
One example of this is the (MPLS label position, MPLS label). Way before 
structured data, our initial implementation contained: MPLS label 
position, MPLS label.
So the collector had to look first at the MPLS label position value in 
order to correctly understand the following  MPLS label.
Somehow, with these conventions, we were having an information model on 
the top of the IE. This was wrong.
Also it was not obvious for a collector, i.e. without hardcoding the 
information.
Finally, what if the IE order changed within the flow record?
Conclusion: we had to change back our implementation mplslabelposition1, 
mplslabelposition2, etc... Like we did in IPIFX ;-)


    </pre>
    <blockquote type="cite">
      <pre wrap="">
- I'm not sure whether the bitvector encoding has an advantage over an
  enumeration type.

Unfortunately, I do not have the time to participate in a deep 
discussion - I'm busy with IPFIX-CONFIG and other stuff.
      </pre>
    </blockquote>
    <pre wrap="">Thanks and regards, Benoit.
    </pre>
    <blockquote type="cite">
      <pre wrap="">
Regards,
Gerhard

Benoit Claise wrote:
      </pre>
      <blockquote type="cite">
        <pre wrap="">Hi Brian,
        </pre>
        <blockquote type="cite">
          <pre wrap="">Hi, Benoit,

Replies inline...

On Mar 16, 2010, at 2:40 AM, Benoit Claise wrote:

          </pre>
          <blockquote type="cite">
            <pre wrap="">Hi Brian,

Thanks for your feedback.
            </pre>
            <blockquote type="cite">
              <pre wrap="">hi, Benoit, all...

Agreed that we should do something about this (i.e., that solution 
1 is no solution.); that said, a few comments in no particular order:

1. In considering adding explicit semantics to structured data, we 
as a WG are taking on the task of defining semantics for IPFIX as 
a whole. Semantics, as I understand them, are in IPFIX largely 
contextual and template dependent, but in almost all cases I can 
think of it seems like these are implicitly "AND" semnantics (this 
flow has source IP A _AND_ destination IP B _AND_...).

              </pre>
            </blockquote>
            <pre wrap="">Agreed.
            </pre>
            <blockquote type="cite">
              <pre wrap="">We will need to make an explicit statement on this.
              </pre>
            </blockquote>
            <pre wrap="">Not sure why.
            </pre>
            <blockquote type="cite">
              <pre wrap="">We will need to determine whether these implicit semantics are a 
property of the protocol (in which case Structured Data is really 
a protocol-level extension), or a property of each information 
element (in which case all 5103 IEs have implicit AND semantics; 
question 2: will we need to add semantics to the IANA registry in 
this case?). We will need to be quite careful about this. It's not 
as simple as defining semantics within structured data then 
calling it done.


              </pre>
            </blockquote>
            <pre wrap="">I'm not sure why each individual IE should have a semantic ... in 
the IANA registry.
            </pre>
          </blockquote>
          <pre wrap="">I'm not either. On reflection it seems like overkill. I'm just 
saying that if we, as a WG, are moving from stating that semantics 
are explicitly out of scope to defining them as in-scope, we need to 
have a consistent approach, and answers to all the questions that 
arise when we consider moving the protocol from a simple framing 
mechanism to a framing mechanism with some logic behind it, so we 
answer them once, so that all future efforts having to do with 
semantics are consistent. This, I acknowledge, is an argument in 
favor of solution 3...)

One very simple new question that arises here, to illustrate my 
point: Is it legal to export a record that has sourceIPAddress X AND 
NOT sourceIPAddress X?
          </pre>
        </blockquote>
        <pre wrap=""> From a protocol point of view, yes
 From a semantic point of view, I don't see a use case for that.
Now, this question is not different that: with RFC5101,  is it legal 
to export a record that has two instances of sourceIPAddress?
        </pre>
        <blockquote type="cite">
          <blockquote type="cite">
            <pre wrap="">As you wrote, "in almost all cases I can think of it seems like 
these are implicitly "AND" semnantics", so the case we try to solve 
is when there are multiple instances of a single IE. We know that 
RFC 5101 foresaw that case " The Collector MUST support the use of 
Templates containing multiple occurrences of the similar 
Information Elements", but the idea is not to change RFC5101.  If 
we put some semantic in the IPFIX structured data, that would solve 
the vast majority of our cases. Also, we could say: if you want 
some semantic when exporting multiple IEs, then you SHOULD use 
IPFIX structured data.

Also, the default value for the semantic field in the IPFIX 
structured data SHOULD be "NONE", to express that flow record 
doesn't include any semantic.... like in RFC5101. You might draw 
your own conclusion, maybe because you know your network, maybe 
because you have configured the exporter, but then it's your decision.
The way I see the proposed solution is: in IPFIX structured data, 
you MAY use the semantic field as a way to express the relationship 
between IEs within the structure.
            </pre>
            <blockquote type="cite">
              <pre wrap="">2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid 
argument against Solution 2, as it's trying to solve a somewhat 
different and more limited problem than structured data. Solution 
2 _might_ cause a problem for this draft, but certainly not the 
other way around (unless we as a WG want to subsume that work into 
this draft, which would probably require rechartering...). Also, I 
don't think we need semantics for all of the list types, just 
basicList (Illustrative question: How do I meaningfully interpret 
two "or" subTemplateMultiLists with disjoint IE sets? Two "and" 
subTemplateMultiLists? Nestings thereof?)... So, we don't really 
have an explosion to deal with if we do Solution 2 correctly: 
andBasicList, orBasicList, xorBasicList, notBasicList; Four IEs, 
nestable, done. What can't we do with those four IEs? In this 
case, we could even step back and say that semantics outside these 
four are within the protocol explicitly _undefined_, and to be 
interpreted withi
              </pre>
            </blockquote>
          </blockquote>
        </blockquote>
      </blockquote>
      <pre wrap="">n the
      </pre>
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <blockquote type="cite">
              <pre wrap="">  context of each Template.


              </pre>
            </blockquote>
            <pre wrap="">If I translate some more what I wrote "While it's solved the router 
and most mediation function needs today", this would be "Me, 
myself, and I don't need more that OR and AND now ;-)".
However, who am I to tell that others don't need it now... and that 
the logical solution is to use the IPFIX structured data
Furthermore, if we think a little bit longer term, the next big 
step in IPFIX is the mediation function. In my company, every 
features want to export his own data with NetFlow/IPFIX... up to 
the point where a CPE would not have enough bandwidth across the 
WAN to export all the "management" information. So we'll need more 
and more of aggregated flow records (both in time and space) even 
in the router. Again, the logical solution will be to use the IPFIX 
structured data. At this point, we will most probably need 
something else than OR and AND, i.e. RANGE, ORDERRED, etc...

An example of subTemplateMultiLists with disjoint IE sets, let's 
imagine that you have to export an aggregated observation point, 
composed of multiple template records
     template 1: exporterIPaddress
     template 2: exporterIPaddress, basicList of interfaces
     template 3: exporterIPaddress, LC
            </pre>
          </blockquote>
          <pre wrap="">and then you'd want to OR these... Okay, makes sense...

          </pre>
          <blockquote type="cite">
            <blockquote type="cite">
              <pre wrap="">3. If we really _do_ want ranges and so on (which, again, we'd 
need to get WG consensus on; this is explicitly out of scope in my 
reading of the present charter), then we could do them in the 
scope of Solution 3.
              </pre>
            </blockquote>
            <pre wrap="">Btw, acknowledging that one day we will have to solve this is good 
enough for solution 3. I mean, we don't have to populate the 
semantic IANA now... even if that would be more efficient.
            </pre>
            <blockquote type="cite">
              <pre wrap="">However, this seems a little not-quite-fleshed-out-enough for me 
to say whether I like it or not. Could you present an example of 
how you would use the proposed Semantic field to model your 
example? "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR linecard2"

(FWIW, I would do this with my proposal to Solution 2 as follows:)

(orBasicList (andBasicList (orBasicList ingressInterface eth1 
eth2) (notBasicList (orBasicList ingressInterface eth3 eth4)) 
(andBasicList lineCardID 2))


              </pre>
            </blockquote>
            <pre wrap="">(BasicList, OR, (basicList, AND, (basicList, OR, eth1, eth2), 
(basicList, NOR, eth3, eth4)), (basicList, NONE, linecard2)))
            </pre>
          </blockquote>
          <pre wrap="">Hm. Okay. This makes sense. A couple of questions then, about 
solution 3:

1. Would we want to try and bitfield this, to define the semantics 
we _know_ we have, then leave the rest of it reserved? Something like:

MSb                      LSb
+---+------+---------------+
| ! | multi| reserved      |
+---+------+---------------+

! = negate sense of semantics if 1 (this is the NOT flag)
multi (multiplicity) = 00: undefined, 01: or/oneOrMore, 10: 
xor/exactlyOne, 11: and/exactlyAll
reserved = place for adding other bells and whistles like RANGE and 
so on in the future.
          </pre>
        </blockquote>
        <pre wrap="">Maybe we want to start by asking the question: which semantic do we 
need now?
Is a need for NONE, OR, AND, ORDERED for now. Anything else?
        </pre>
        <blockquote type="cite">
          <pre wrap="">2. Are we sure we want to do this in one byte? If we do bitfielding 
as above this gives us 32 possible extensions, which seems like 
_way_ more than enough, but does stick another odd offset in there, 
which slows things down on machines that need aligned access. 
Probably one byte is okay and we let the implementation use 
paddingOctets and set padding to fix this...
          </pre>
        </blockquote>
        <pre wrap="">With structured data, is the alignment still important? When I look 
at the examples throughout the draft... As you wrote, paddingOctets 
might be the solution in this case.
        </pre>
        <blockquote type="cite">
          <pre wrap="">3. Would it make sense maybe to have two sets of structured data 
elements, one with the semantics byte, and one without (which is 
then explicitly undefined)? Then exporters who don't need it don't 
have to bother sticking an extra odd-aligned zero in the stream for 
every list.

I'm sure I'l have more questions, but none come to mind now... But 
it seems like we're converging on the least-unnecessarily-complex 
solution here, which is good. :)
          </pre>
        </blockquote>
        <pre wrap="">Happy about that. ;-)
Note: I was envisioning even something more simpler: one byte, 
containing all the semantic possibilities, administered by IANA.  So 
no reserved, no !

Regards, Benoit.
        </pre>
      </blockquote>
      <pre wrap="">
      </pre>
    </blockquote>
    <pre wrap="">
_______________________________________________
IPFIX mailing list
<a class="moz-txt-link-abbreviated" href="mailto:IPFIX@ietf.org">IPFIX@ietf.org</a>
<a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/ipfix">https://www.ietf.org/mailman/listinfo/ipfix</a>

    </pre>
  </blockquote>
  <pre wrap="">
--- 
Atsushi KOBAYASHI  <a class="moz-txt-link-rfc2396E" href="mailto:akoba@nttv6.net">&lt;akoba@nttv6.net&gt;</a>
NTT Information Sharing Platform Lab.
tel:+81-(0)422-59-3978 fax:+81-(0)422-59-5637

  </pre>
</blockquote>
<br>
</body>
</html>

--------------040601060409040805080109--

From paitken@cisco.com  Thu Mar 18 04:21:50 2010
Return-Path: <paitken@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id BF5FF3A690B for <ipfix@core3.amsl.com>; Thu, 18 Mar 2010 04:21:50 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -3.716
X-Spam-Level: 
X-Spam-Status: No, score=-3.716 tagged_above=-999 required=5 tests=[AWL=-2.247, BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id G3a75usxMbtk for <ipfix@core3.amsl.com>; Thu, 18 Mar 2010 04:21:48 -0700 (PDT)
Received: from ams-iport-2.cisco.com (ams-iport-2.cisco.com [144.254.224.141]) by core3.amsl.com (Postfix) with ESMTP id D8B693A68D9 for <ipfix@ietf.org>; Thu, 18 Mar 2010 04:21:47 -0700 (PDT)
Authentication-Results: ams-iport-2.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-AV: E=Sophos;i="4.51,265,1267401600";  d="scan'208";a="4496907"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-2.cisco.com with ESMTP; 18 Mar 2010 10:48:10 +0000
Received: from [10.61.98.100] (dhcp-10-61-98-100.cisco.com [10.61.98.100]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2IBLvaI012753; Thu, 18 Mar 2010 11:21:57 GMT
Message-ID: <4BA20CD5.6010800@cisco.com>
Date: Thu, 18 Mar 2010 11:21:57 +0000
From: Paul Aitken <paitken@cisco.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-GB; rv:1.8.1.22) Gecko/20090605 SeaMonkey/1.1.17 (Ubuntu-1.1.17+nobinonly-0ubuntu0.9.04.1)
MIME-Version: 1.0
To: Benoit Claise <bclaise@cisco.com>
References: <4AF73525.8050009@net.in.tum.de>	<4AF8F999.3000207@cisco.com>	<F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>	<4B9E33BE.7070907@cisco.com>	<7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch>	<4B9F520E.8060507@cisco.com>	<DB6E59D9-B373-4919-BB58-00EB26014564@tik.ee.ethz.ch>	<4BA0C26D.9070901@cisco.com>	<6CD84D0C-333F-4C65-8E41-B1DF3774BA15@tik.ee.ethz.ch> <4BA1E4C4.3000204@cisco.com>
In-Reply-To: <4BA1E4C4.3000204@cisco.com>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] [Sender: ipfix-bounces@ietf.org] Re: Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 18 Mar 2010 11:21:50 -0000

Benoit, Brian,

>> On reflection, I would suggest defining these in terms of 
>> multiplicities (as we're dealing with N-element lists) as opposed to 
>> logical operations. So I'd propose the following:


>> 0x00: undefined (default) (CP may assume allOf)
>    
> This corresponds to NONE


>> 0x01: allOf
>    
> This corresponds to AND


>> 0x02: oneOrMoreOf
>    
> This corresponds to OR


>> 0x03: exactlyOneOf
>    
> Not sure what it corresponds to. At first glance, I don't see a use case for it.

0x02 and 0x03 are two different kinds of OR: "at least one of" as 
opposed to "just one of".


>> 0x04: noneOf
>
> This corresponds to NOT

- which introduces the idea of NOT as a semantic in its own right, 
rather than simply as a modifier for other semantics.

However, "noneOf (x, y, z)" could be "NOT-AND (x, y, z)". Or possibly 
"NOT-OR (x, y, z)", though the distinction is subtle.

So NOT may still be considered as a modifier, rather than a distinct and 
separate semantic.



>> If I understand what you mean by ORDERED, add:
>> 0x05: allInOrder
>
> This corresponds to ORDERED


>> If we really want to do ranges:
>> 0x06: inclusiveRanges (MUST have even number of elements, and elements 
>> MUST be numeric; this is then equivalent to inclusive allOf [e_0,e_1], 
>> [e_2,e_3], ... , [e_n-1]..[e_n]);
>    
> This corresponds to RANGE
> 
> What is required? NONE, AND, OR, potentially NOT, ORDERED

And RANGE.


>> Still unanswered is whether we want to force all 
>> basicList/subTemplateList/subTemplateMultiList instances to state 
>> semantics, or whether we want "semanticless" versions as well...
>>    
> No strong feeling about this one.
> Maybe it would be better to have additional semanticBasicList, 
> semanticSubTemplateList, semanticSubTemplateMultiList

Then there's no need for semantics = NONE.

I think it's more straightforward not to have "semanticless" versions.

Thanks,
P.

-- 
Paul Aitken
Cisco Systems Ltd, Edinburgh, Scotland.
http://www.cisco.com/web/about/doing_business/legal/cri/index.html

From trammell@tik.ee.ethz.ch  Thu Mar 18 07:57:19 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 22A283A6C37 for <ipfix@core3.amsl.com>; Thu, 18 Mar 2010 07:57:19 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.838
X-Spam-Level: 
X-Spam-Status: No, score=-5.838 tagged_above=-999 required=5 tests=[AWL=-0.369, BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id D6p1rxNxnsCk for <ipfix@core3.amsl.com>; Thu, 18 Mar 2010 07:57:18 -0700 (PDT)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id EA2493A6C25 for <ipfix@ietf.org>; Thu, 18 Mar 2010 07:57:17 -0700 (PDT)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id DA73BD94AC; Thu, 18 Mar 2010 15:57:27 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id NZeATXdnkqD0; Thu, 18 Mar 2010 15:57:27 +0100 (MET)
Received: from [10.0.1.5] (unknown [75.85.164.240]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id 7F9BAD9303; Thu, 18 Mar 2010 15:57:26 +0100 (MET)
Mime-Version: 1.0 (Apple Message framework v1077)
Content-Type: text/plain; charset=us-ascii
From: Brian Trammell <trammell@tik.ee.ethz.ch>
In-Reply-To: <4BA20CD5.6010800@cisco.com>
Date: Thu, 18 Mar 2010 07:56:56 -0700
Content-Transfer-Encoding: quoted-printable
Message-Id: <EA4C7E42-1F67-4898-8CA3-D06A4FCD4C2E@tik.ee.ethz.ch>
References: <4AF73525.8050009@net.in.tum.de>	<4AF8F999.3000207@cisco.com>	<F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>	<4B9E33BE.7070907@cisco.com>	<7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch>	<4B9F520E.8060507@cisco.com>	<DB6E59D9-B373-4919-BB58-00EB26014564@tik.ee.ethz.ch>	<4BA0C26D.9070901@cisco.com>	<6CD84D0C-333F-4C65-8E41-B1DF3774BA15@tik.ee.ethz.ch> <4BA1E4C4.3000204@cisco.com> <4BA20CD5.6010800@cisco.com>
To: Paul Aitken <paitken@cisco.com>
X-Mailer: Apple Mail (2.1077)
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] [Sender: ipfix-bounces@ietf.org] Re: Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 18 Mar 2010 14:57:19 -0000

On Mar 18, 2010, at 4:21 AM, Paul Aitken wrote:

> Benoit, Brian,
>=20
>>> On reflection, I would suggest defining these in terms of =
multiplicities (as we're dealing with N-element lists) as opposed to =
logical operations. So I'd propose the following:
>=20
>=20
>>> 0x00: undefined (default) (CP may assume allOf)
>>   This corresponds to NONE

Ahhh..... okay. This is confusing, though. I'd _definitely_ call this =
"undefined", because "none" could be interpreted as "not".

>=20
>>> 0x01: allOf
>>   This corresponds to AND
>=20
>=20
>>> 0x02: oneOrMoreOf
>>   This corresponds to OR
>=20
>=20
>>> 0x03: exactlyOneOf
>>   Not sure what it corresponds to. At first glance, I don't see a use =
case for it.
>=20
> 0x02 and 0x03 are two different kinds of OR: "at least one of" as =
opposed to "just one of".

With two elements in the list, it corresponds to XOR; exactlyOneOf is =
the multiplicity extension of this (which lends itself nicely to clarity =
for lists of more than two).

The use case is when you want to express uncertainty within a certain =
set: for example, say a mediator knows that a flow was observed on one =
of a set of interfaces at a MP on an original exporter (by virtue of =
having come from that exporter), but that the flow ONLY traversed one of =
them (say, because the exporter is only looking at ingress interfaces at =
the network border, and the routing configuration precludes balancing =
across those interfaces). With oneOrMoreOf ("OR"), the mediator cannot =
exclude the possibility it was seen on both. It's also one of the basic =
multiplicities {0, 1, 1..n, n}, so should be included for completeness =
in any case.

>=20
>>> 0x04: noneOf
>>=20
>> This corresponds to NOT
>=20
> - which introduces the idea of NOT as a semantic in its own right, =
rather than simply as a modifier for other semantics.
>=20
> However, "noneOf (x, y, z)" could be "NOT-AND (x, y, z)". Or possibly =
"NOT-OR (x, y, z)", though the distinction is subtle.
>=20
> So NOT may still be considered as a modifier, rather than a distinct =
and separate semantic.

I think the argument for the flag goes away if you treat these as =
multiplicities. I'd treat the flag as a set negation: NOT AND =3D allOf =
(set) (negated) =3D noneOf: the represented value set can contain zero =
or more elements in the universe that do not appear in the list.=20

Here, NOT OR =3D oneOrMoreOf (set) (negated) =3D notAllOf, meaning, the =
represented value set can contain zero or more elements in the universe =
but not in the list, as well as up to n-1 elements in the list, but not =
all n elements of the list; at least one (unspecified) value in the list =
must be absent from the set of represented values. It is not clear to me =
that this is useful...

NOT XOR =3D exactlyOneOf (set) (negated) seems even less useful.


>=20
>=20
>>> If I understand what you mean by ORDERED, add:
>>> 0x05: allInOrder
>>=20
>> This corresponds to ORDERED
>=20
>=20
>>> If we really want to do ranges:
>>> 0x06: inclusiveRanges (MUST have even number of elements, and =
elements MUST be numeric; this is then equivalent to inclusive allOf =
[e_0,e_1], [e_2,e_3], ... , [e_n-1]..[e_n]);
>>   This corresponds to RANGE
>> What is required? NONE, AND, OR, potentially NOT, ORDERED
>=20
> And RANGE.

The case for this one's not clear to me. I see the use case, but I'm =
undecided on whether it's in scope for this draft...

>=20
>>> Still unanswered is whether we want to force all =
basicList/subTemplateList/subTemplateMultiList instances to state =
semantics, or whether we want "semanticless" versions as well...
>>>  =20
>> No strong feeling about this one.
>> Maybe it would be better to have additional semanticBasicList, =
semanticSubTemplateList, semanticSubTemplateMultiList
>=20
> Then there's no need for semantics =3D NONE.
>=20
> I think it's more straightforward not to have "semanticless" versions.

arguments for "semanticless" versions: clean representation and no waste =
of a byte per list (okay, minor considering how big lists are likely to =
be, but maybe annoying for alignment purposes) for the (majority of?) =
applications that don't care about semantics. Arguments against: we need =
an undefined (again, "NONE" might lead to confusion) semantic flag =
anyway so now we have two legal ways to represent the same thing, which =
makes the collector's life harder... plus it doesn't look as clean in =
the IE registry.=20

No super strong opinion here either way...

Best regards,

Brian=

From Carsten.Schmoll9ab33xy531fokus.fraunhofer.de@bounce.antispameurope.com  Thu Mar 18 11:57:15 2010
Return-Path: <Carsten.Schmoll9ab33xy531fokus.fraunhofer.de@bounce.antispameurope.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 6BFD43A6AEF for <ipfix@core3.amsl.com>; Thu, 18 Mar 2010 11:57:15 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.468
X-Spam-Level: 
X-Spam-Status: No, score=-2.468 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-1]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id gWQgBcLWvNOU for <ipfix@core3.amsl.com>; Thu, 18 Mar 2010 11:57:10 -0700 (PDT)
Received: from relay04-haj2.antispameurope.com (relay04-haj2.antispameurope.com [83.246.65.54]) by core3.amsl.com (Postfix) with ESMTP id D421C3A6896 for <ipfix@ietf.org>; Thu, 18 Mar 2010 11:57:09 -0700 (PDT)
Received: by relay04-haj2.antispameurope.com (ASE-Secure-MTA, from userid 1000) id 9E60A5EC1D4; Thu, 18 Mar 2010 19:57:18 +0100 (CET)
Received: from pluto.fokus.fraunhofer.de (pluto.fokus.fraunhofer.de [195.37.77.164]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by relay04-haj2.antispameurope.com (ASE-Secure-MTA) with ESMTP id 8CF445EC1DD; Thu, 18 Mar 2010 19:57:16 +0100 (CET)
Received: from EXCHSRV.fokus.fraunhofer.de (bohr.fokus.fraunhofer.de [10.147.9.231]) by pluto.fokus.fraunhofer.de (8.14.2/8.14.2) with SMTP id o2IIvG8g020511; Thu, 18 Mar 2010 19:57:16 +0100 (MET)
x-mimeole: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Date: Thu, 18 Mar 2010 19:57:16 +0100
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=SHA1; boundary="----=_NextPart_000_00B2_01CAC6D5.35B92420"
Message-ID: <70524A4436C03E43A293958B505008B60281DF55@EXCHSRV.fokus.fraunhofer.de>
In-Reply-To: <4B87A9DE.1030902@tik.ee.ethz.ch>
X-MS-Has-Attach: yes
X-MS-TNEF-Correlator: 
Thread-Topic: [IPFIX] Fwd:  I-D Action:draft-ietf-ipfix-anon-02.txt
Thread-Index: Acq20vpgYUVkLKXzQkSpJZ7cu2yG8QP+FxLQ
References: <4B87A9DE.1030902@tik.ee.ethz.ch>
From: "Schmoll, Carsten" <Carsten.Schmoll@fokus.fraunhofer.de>
To: "Elisa Boschi" <boschie@tik.ee.ethz.ch>, "Brian Trammell" <trammell@tik.ee.ethz.ch>
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Fwd:  I-D Action:draft-ietf-ipfix-anon-02.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 18 Mar 2010 18:58:10 -0000

This is a multi-part message in MIME format.

------=_NextPart_000_00B2_01CAC6D5.35B92420
Content-Type: multipart/alternative;
	boundary="----=_NextPart_001_00B3_01CAC6D5.35B92420"


------=_NextPart_001_00B3_01CAC6D5.35B92420
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Dear Elisa, Brian,

=20

I=92ll have a look at the latest version over the weekend. The last =
version
was already quite mature so I expect only a few comments will come out =
of
this review from my side.

More later =85

=20

Best regards

Carsten

=20

=20

Von: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] Im Auftrag =
von
Elisa Boschi
Gesendet: Freitag, 26. Februar 2010 12:01
An: ipfix-chairs@tools.ietf.org
Cc: ipfix@ietf.org
Betreff: [IPFIX] Fwd: I-D Action:draft-ietf-ipfix-anon-02.txt

=20

Nevil, J=FCrgen,

as you've seen we've submitted a new revision of the anonymisation =
draft.
The changes are relatively minor and address all remaining open issues =
and
comments received so far. At this point we think we're ready for Working
Group Last Call. In order to take advantage of the upcoming meeting we
suggest starting WG last call soon so that we can discuss last call =
comments
during the anon draft agenda slot.

thanks,
Elisa & Brian

-------- Original Message --------=20


Subject:=20

[IPFIX] I-D Action:draft-ietf-ipfix-anon-02.txt


Date:=20

Mon, 15 Feb 2010 04:00:01 -0800 (PST)


From:=20

Internet-Drafts@ietf.org


To:=20

i-d-announce@ietf.org


CC:=20

ipfix@ietf.org

=20

A New Internet-Draft is available from the on-line Internet-Drafts
directories.
This draft is a work item of the IP Flow Information Export Working =
Group of
the IETF.
=20
=20
        Title           : IP Flow Anonymisation Support
        Author(s)       : E. Boschi, B. Trammell
        Filename        : draft-ietf-ipfix-anon-02.txt
        Pages           : 37
        Date            : 2010-02-15
=20
This document describes anonymisation techniques for IP flow data and
the export of anonymised data using the IPFIX protocol.  It
categorizes common anonymisation schemes and defines the parameters
needed to describe them.  It provides guidelines for the
implementation of anonymised data export and storage over IPFIX, and
describes an information model and Options-based method for
anonymisation technique metadata export within the IPFIX protocol or
storage in IPFIX Files.
=20
A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-ipfix-anon-02.txt
=20
Internet-Drafts are also available by anonymous FTP at:
ftp://ftp.ietf.org/internet-drafts/
=20
Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.
=20

------=_NextPart_001_00B3_01CAC6D5.35B92420
Content-Type: text/html;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Diso-8859-1">
<html xmlns:v=3D"urn:schemas-microsoft-com:vml" =
xmlns:o=3D"urn:schemas-microsoft-com:office:office" =
xmlns:w=3D"urn:schemas-microsoft-com:office:word" =
xmlns:m=3D"http://schemas.microsoft.com/office/2004/12/omml" =
xmlns=3D"http://www.w3.org/TR/REC-html40">

<head>

<meta name=3DGenerator content=3D"Microsoft Word 12 (filtered medium)">
<style>
<!--
 /* Font Definitions */
 @font-face
	{font-family:"Cambria Math";
	panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
	{font-family:Calibri;
	panose-1:2 15 5 2 2 2 4 3 2 4;}
@font-face
	{font-family:Tahoma;
	panose-1:2 11 6 4 3 5 4 4 2 4;}
@font-face
	{font-family:Consolas;
	panose-1:2 11 6 9 2 2 4 3 2 4;}
 /* Style Definitions */
 p.MsoNormal, li.MsoNormal, div.MsoNormal
	{margin:0cm;
	margin-bottom:.0001pt;
	font-size:12.0pt;
	font-family:"Times New Roman","serif";
	color:black;}
a:link, span.MsoHyperlink
	{mso-style-priority:99;
	color:blue;
	text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
	{mso-style-priority:99;
	color:purple;
	text-decoration:underline;}
pre
	{mso-style-priority:99;
	mso-style-link:"HTML Vorformatiert Zchn";
	margin:0cm;
	margin-bottom:.0001pt;
	font-size:10.0pt;
	font-family:"Courier New";
	color:black;}
span.HTMLVorformatiertZchn
	{mso-style-name:"HTML Vorformatiert Zchn";
	mso-style-priority:99;
	mso-style-link:"HTML Vorformatiert";
	font-family:Consolas;
	color:black;}
span.E-MailFormatvorlage19
	{mso-style-type:personal-reply;
	font-family:"Calibri","sans-serif";
	color:#1F497D;}
.MsoChpDefault
	{mso-style-type:export-only;
	font-size:10.0pt;}
@page Section1
	{size:612.0pt 792.0pt;
	margin:70.85pt 70.85pt 2.0cm 70.85pt;}
div.Section1
	{page:Section1;}
-->
</style>
<!--[if gte mso 9]><xml>
 <o:shapedefaults v:ext=3D"edit" spidmax=3D"1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
 <o:shapelayout v:ext=3D"edit">
  <o:idmap v:ext=3D"edit" data=3D"1" />
 </o:shapelayout></xml><![endif]-->
</head>

<body bgcolor=3Dwhite lang=3DDE link=3Dblue vlink=3Dpurple>

<div class=3DSection1>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>Dear Elisa, Brian,<o:p></o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>I&#8217;ll have a look at the latest version over the =
weekend. The
last version was already quite mature so I expect only a few comments =
will come
out of this review from my side.<o:p></o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>More later &#8230;<o:p></o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>Best regards<o:p></o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>Carsten<o:p></o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<div>

<div style=3D'border:none;border-top:solid #B5C4DF 1.0pt;padding:3.0pt =
0cm 0cm 0cm'>

<p class=3DMsoNormal><b><span =
style=3D'font-size:10.0pt;font-family:"Tahoma","sans-serif";
color:windowtext'>Von:</span></b><span =
style=3D'font-size:10.0pt;font-family:
"Tahoma","sans-serif";color:windowtext'> ipfix-bounces@ietf.org
[mailto:ipfix-bounces@ietf.org] <b>Im Auftrag von </b>Elisa Boschi<br>
<b>Gesendet:</b> Freitag, 26. Februar 2010 12:01<br>
<b>An:</b> ipfix-chairs@tools.ietf.org<br>
<b>Cc:</b> ipfix@ietf.org<br>
<b>Betreff:</b> [IPFIX] Fwd: I-D =
Action:draft-ietf-ipfix-anon-02.txt<o:p></o:p></span></p>

</div>

</div>

<p class=3DMsoNormal><o:p>&nbsp;</o:p></p>

<p class=3DMsoNormal>Nevil, J=FCrgen,<br>
<br>
as you've seen we've submitted a new revision of the anonymisation =
draft. The
changes are relatively minor and address all remaining open issues and =
comments
received so far. At this point we think we're ready for Working Group =
Last
Call. In order to take advantage of the upcoming meeting we suggest =
starting WG
last call soon so that we can discuss last call comments during the anon =
draft
agenda slot.<br>
<br>
thanks,<br>
Elisa &amp; Brian<br>
<br>
-------- Original Message -------- <o:p></o:p></p>

<table class=3DMsoNormalTable border=3D0 cellspacing=3D0 =
cellpadding=3D0>
 <tr>
  <td nowrap valign=3Dtop style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal align=3Dright =
style=3D'text-align:right'><b>Subject: <o:p></o:p></b></p>
  </td>
  <td style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal>[IPFIX] I-D =
Action:draft-ietf-ipfix-anon-02.txt<o:p></o:p></p>
  </td>
 </tr>
 <tr>
  <td nowrap valign=3Dtop style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal align=3Dright style=3D'text-align:right'><b>Date: =
<o:p></o:p></b></p>
  </td>
  <td style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal>Mon, 15 Feb 2010 04:00:01 -0800 =
(PST)<o:p></o:p></p>
  </td>
 </tr>
 <tr>
  <td nowrap valign=3Dtop style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal align=3Dright style=3D'text-align:right'><b>From: =
<o:p></o:p></b></p>
  </td>
  <td style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal><a =
href=3D"mailto:Internet-Drafts@ietf.org">Internet-Drafts@ietf.org</a><o:p=
></o:p></p>
  </td>
 </tr>
 <tr>
  <td nowrap valign=3Dtop style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal align=3Dright style=3D'text-align:right'><b>To: =
<o:p></o:p></b></p>
  </td>
  <td style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal><a =
href=3D"mailto:i-d-announce@ietf.org">i-d-announce@ietf.org</a><o:p></o:p=
></p>
  </td>
 </tr>
 <tr>
  <td nowrap valign=3Dtop style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal align=3Dright style=3D'text-align:right'><b>CC: =
<o:p></o:p></b></p>
  </td>
  <td style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal><a =
href=3D"mailto:ipfix@ietf.org">ipfix@ietf.org</a><o:p></o:p></p>
  </td>
 </tr>
</table>

<p class=3DMsoNormal =
style=3D'margin-bottom:12.0pt'><o:p>&nbsp;</o:p></p>

<pre>A New Internet-Draft is available from the on-line Internet-Drafts =
directories.<o:p></o:p></pre><pre>This draft is a work item of the IP =
Flow Information Export Working Group of the =
IETF.<o:p></o:p></pre><pre><o:p>&nbsp;</o:p></pre><pre><o:p>&nbsp;</o:p><=
/pre><pre>=A0=A0=A0=A0=A0=A0=A0 Title=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0 : IP =
Flow Anonymisation Support<o:p></o:p></pre><pre>=A0=A0=A0=A0=A0=A0=A0 =
Author(s)=A0=A0=A0=A0=A0=A0 : E. Boschi, B. =
Trammell<o:p></o:p></pre><pre>=A0=A0=A0=A0=A0=A0=A0 =
Filename=A0=A0=A0=A0=A0=A0=A0 : =
draft-ietf-ipfix-anon-02.txt<o:p></o:p></pre><pre>=A0=A0=A0=A0=A0=A0=A0 =
Pages=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0 : =
37<o:p></o:p></pre><pre>=A0=A0=A0=A0=A0=A0=A0 Date =
=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0: =
2010-02-15<o:p></o:p></pre><pre><o:p>&nbsp;</o:p></pre><pre>This =
document describes anonymisation techniques for IP flow data =
and<o:p></o:p></pre><pre>the export of anonymised data using the IPFIX =
protocol.=A0 It<o:p></o:p></pre><pre>categorizes common anonymisation =
schemes and defines the parameters<o:p></o:p></pre><pre>needed to =
describe them.=A0 It provides guidelines for =
the<o:p></o:p></pre><pre>implementation of anonymised data export and =
storage over IPFIX, and<o:p></o:p></pre><pre>describes an information =
model and Options-based method for<o:p></o:p></pre><pre>anonymisation =
technique metadata export within the IPFIX protocol =
or<o:p></o:p></pre><pre>storage in IPFIX =
Files.<o:p></o:p></pre><pre><o:p>&nbsp;</o:p></pre><pre>A URL for this =
Internet-Draft is:<o:p></o:p></pre><pre><a
href=3D"http://www.ietf.org/internet-drafts/draft-ietf-ipfix-anon-02.txt"=
>http://www.ietf.org/internet-drafts/draft-ietf-ipfix-anon-02.txt</a><o:p=
></o:p></pre><pre><o:p>&nbsp;</o:p></pre><pre>Internet-Drafts are also =
available by anonymous FTP at:<o:p></o:p></pre><pre><a
href=3D"ftp://ftp.ietf.org/internet-drafts/">ftp://ftp.ietf.org/internet-=
drafts/</a><o:p></o:p></pre><pre><o:p>&nbsp;</o:p></pre><pre>Below is =
the data which will enable a MIME compliant mail =
reader<o:p></o:p></pre><pre>implementation to automatically retrieve the =
ASCII version of =
the<o:p></o:p></pre><pre>Internet-Draft.<o:p></o:p></pre><pre><o:p>&nbsp;=
</o:p></pre></div>

</body>

</html>

------=_NextPart_001_00B3_01CAC6D5.35B92420--

------=_NextPart_000_00B2_01CAC6D5.35B92420
Content-Type: application/x-pkcs7-signature;
	name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment;
	filename="smime.p7s"
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------=_NextPart_000_00B2_01CAC6D5.35B92420--

From Thomas.Dietz@nw.neclab.eu  Fri Mar 19 06:58:53 2010
Return-Path: <Thomas.Dietz@nw.neclab.eu>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 756043A6A1A for <ipfix@core3.amsl.com>; Fri, 19 Mar 2010 06:58:53 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.548
X-Spam-Level: 
X-Spam-Status: No, score=-0.548 tagged_above=-999 required=5 tests=[AWL=-1.679, BAYES_50=0.001, DNS_FROM_OPENWHOIS=1.13]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id BOEFoJHCirU7 for <ipfix@core3.amsl.com>; Fri, 19 Mar 2010 06:58:52 -0700 (PDT)
Received: from smtp0.neclab.eu (smtp0.neclab.eu [195.37.70.41]) by core3.amsl.com (Postfix) with ESMTP id A18063A695B for <ipfix@ietf.org>; Fri, 19 Mar 2010 06:58:51 -0700 (PDT)
Received: from localhost (localhost.localdomain [127.0.0.1]) by smtp0.neclab.eu (Postfix) with ESMTP id 71FFA2C0204FC; Fri, 19 Mar 2010 14:59:04 +0100 (CET)
X-Virus-Scanned: Amavisd on Debian GNU/Linux (atlas2.office)
Received: from smtp0.neclab.eu ([127.0.0.1]) by localhost (atlas2.office [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id r64ozJeziXq9; Fri, 19 Mar 2010 14:59:04 +0100 (CET)
Received: from VENUS.office (mx2.office [192.168.24.15]) by smtp0.neclab.eu (Postfix) with ESMTP id 56BED2C01D47A; Fri, 19 Mar 2010 14:58:54 +0100 (CET)
Content-class: urn:content-classes:message
MIME-Version: 1.0
X-MimeOLE: Produced By Microsoft Exchange V6.5
Date: Fri, 19 Mar 2010 14:58:53 +0100
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=SHA1; boundary="----=_NextPart_000_0260_01CAC774.B10307A0"
Message-ID: <547F018265F92642B577B986577D671C012A31C5@VENUS.office>
In-reply-to: <4B9FA451.8080802@net.in.tum.de>
X-MS-Has-Attach: yes
X-MS-TNEF-Correlator: 
Thread-Topic: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIXMIB
Thread-Index: AcrFHtfPqU7+NwLLTSi+3e3EtpHV3wCTRm8A
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com><4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com><4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com><4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E10EE.9050406@net.in.tum.de> <547F018265F92642B577B986577D671C012A2D27@VENUS.office> <4B9F9574.1080606@net.in.tum.de> <547F018265F92642B577B986577D671C012A2D5B@VENUS.office> <4B9FA451.8080802@net.in.tum.de>
From: "Thomas Dietz" <Thomas.Dietz@nw.neclab.eu>
To: "Gerhard Muenz" <muenz@net.in.tum.de>
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIXMIB
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 19 Mar 2010 13:58:53 -0000

This is a multi-part message in MIME format.

------=_NextPart_000_0260_01CAC774.B10307A0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Hi Gerhard,

> -----Original Message-----
> From: Gerhard Muenz [mailto:muenz@net.in.tum.de]
> Sent: Dienstag, 16. M=E4rz 2010 16:31
> To: Thomas Dietz
> Cc: IETF IPFIX Working Group
> Subject: Re: [IPFIX] IPFIX-CONFIG: Questions about counters adopted
> from IPFIXMIB
>=20
> Hi Thomas,
>=20
> > Be careful here: the discontinuity time is the time when anything
> went wrong
> > and you cannot rely on the value of the counter anymore because it
> may have
> > missed some packets or whatsoever. This is not the time the counter
> was last
> > updated or started!
> >
> > By the way this is also something you may need?
>  >
> > Fine but what do you do if your counter wraps?
>=20
> Good points.
>=20
> My original point was that I would like to get away from sysUpTime
> because it is not sure from where you can get it with Netconf.
>=20
> Discontinuity times like in the MIB are ok, but then they should
> contain
> absolute time stamps.

If you are sure that you can always get the correct absolute time then =
this
is fine. I am not an Netconf expert, so it's up to you to judge.

>=20
> Maybe it is the easiest to only make this little change and keep the
> definitions of the other counters as they are in the MIB.

This would be the easiest option from my point of view.

Best Regards,

Thomas

>=20
> What do you think?
>=20
> Regards,
> Gerhard
>=20
> --
> Dipl.-Ing. Gerhard M=FCnz
> Chair for Network Architectures and Services (I8)
> Technische Universit=E4t M=FCnchen - Department of Informatics
> Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
> Phone:  +49 89 289-18008       Fax: +49 89 289-18033
> E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz
>=20


--=20
Thomas Dietz                 E-mail: Thomas.Dietz@nw.neclab.eu
NEC Europe Ltd.              Phone:  +49 6221 4342-128
NEC Laboratories Europe      Fax:    +49 6221 4342-155
Network Research Division
Kurfuersten-Anlage 36
69115 Heidelberg, Germany    http://www.nw.neclab.eu

NEC Europe Limited           Registered in England 2832014
Registered Office: NEC House, 1 Victoria Road, London W3 6BL

------=_NextPart_000_0260_01CAC774.B10307A0
Content-Type: application/x-pkcs7-signature;
	name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment;
	filename="smime.p7s"
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------=_NextPart_000_0260_01CAC774.B10307A0--

From Carsten.Schmoll9ab33xy531fokus.fraunhofer.de@bounce.antispameurope.com  Fri Mar 19 07:31:24 2010
Return-Path: <Carsten.Schmoll9ab33xy531fokus.fraunhofer.de@bounce.antispameurope.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 279D83A6A91 for <ipfix@core3.amsl.com>; Fri, 19 Mar 2010 07:31:24 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.631
X-Spam-Level: 
X-Spam-Status: No, score=-0.631 tagged_above=-999 required=5 tests=[AWL=-0.577, BAYES_40=-0.185, DNS_FROM_OPENWHOIS=1.13, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-1]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id dUxv87V7+Ctp for <ipfix@core3.amsl.com>; Fri, 19 Mar 2010 07:31:15 -0700 (PDT)
Received: from relay04-haj2.antispameurope.com (relay04-haj2.antispameurope.com [83.246.65.54]) by core3.amsl.com (Postfix) with ESMTP id C51D03A69D7 for <ipfix@ietf.org>; Fri, 19 Mar 2010 07:31:12 -0700 (PDT)
Received: by relay04-haj2.antispameurope.com (ASE-Secure-MTA, from userid 1000) id C977B5EC260; Fri, 19 Mar 2010 15:31:23 +0100 (CET)
Received: from pluto.fokus.fraunhofer.de (pluto.fokus.fraunhofer.de [195.37.77.164]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by relay04-haj2.antispameurope.com (ASE-Secure-MTA) with ESMTP id 59E7F5EC14D; Fri, 19 Mar 2010 15:31:22 +0100 (CET)
Received: from EXCHSRV.fokus.fraunhofer.de (bohr.fokus.fraunhofer.de [10.147.9.231]) by pluto.fokus.fraunhofer.de (8.14.2/8.14.2) with SMTP id o2JEVLhi023840; Fri, 19 Mar 2010 15:31:21 +0100 (MET)
x-mimeole: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01CAC770.D8589671"
Date: Fri, 19 Mar 2010 15:31:21 +0100
Message-ID: <70524A4436C03E43A293958B505008B60281DF5F@EXCHSRV.fokus.fraunhofer.de>
In-Reply-To: <4B87A9DE.1030902@tik.ee.ethz.ch>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [IPFIX] Fwd:  I-D Action:draft-ietf-ipfix-anon-02.txt
Thread-Index: Acq20vpgYUVkLKXzQkSpJZ7cu2yG8QQnAk/Q
References: <4B87A9DE.1030902@tik.ee.ethz.ch>
From: "Schmoll, Carsten" <Carsten.Schmoll@fokus.fraunhofer.de>
To: "Elisa Boschi" <boschie@tik.ee.ethz.ch>, "Brian Trammell" <trammell@tik.ee.ethz.ch>
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Fwd:  I-D Action:draft-ietf-ipfix-anon-02.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 19 Mar 2010 14:31:24 -0000

This is a multi-part message in MIME format.

------_=_NextPart_001_01CAC770.D8589671
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

=20

Dear Elisa, Brian,

=20

I reviewed the latest version of the draft and, apart from a few =
suggestions for improved wording, found only two issues, which I would =
consider to address:

=20

If I'm not completely mistaken then there exists no section about "IP =
address enumeration" yet. Since this is a commonly used technique, I =
would give it some text (see below).

The second is similar; there is no explicit notion of hashing of IP =
addresses, which is often used as well, but has some deviations from a =
perfect permutation, namely collisions.

=20

I wrote some new text as a suggestion, following below. Maybe it could =
be a subsection of 4.1.3 Permutation. I also think it could be a bit =
shorter but would leave this open to you. Comments are of course =
welcome.

=20

<Current text:>

4.1.3.  Permutation

=20

   Permutation is a direct substitution technique, replacing each IP

   address with an address selected from the set of possible IP

   addresses, guaranteeing that each anonymised address represents a

   unique original address.  The selection function is often random,

   though it is not necessarily so.  Permutation does not preserve any

   structural information about a network, but it does preserve the

   unique count of IP addresses.  Any application that requires more

   structure than host-uniqueness will not be able to use permuted IP

   addresses.

=20

<New test could start here:>

As subcategories of permutation exist stateful and stateless permutation =
techniques: Stateless permutation (e.g., by exchange of address bits in =
a fixed way) yields a consistent, deterministic mapping of addresses, =
while stateful permutation (e.g., by renumbering found IP addresses in a =
packet trace by giving each a new number in a consecutive way) does not =
yield a consistent anonymisation across multiple packet traces. See =
section xxx for details on the definition of anonymisation mapping =
stability.

=20

4.1.3.1 Enumeration

=20

   Enumeration is a substitution function that retains the chronological =
order in which different IP addresses occurred on the network or in an =
anonymised packet data trace while eliminating the original IP =
addresses. IP Addresses are usually replaced by consecutive addresses, =
e.g. counting up from 1.1.1.1, 1.1.1.2, and so on in the IPv4 case. =
During this process a map is extended whenever a previously unseen IP =
address is observed. This kind of permutation produces a unique mapping, =
which is usually not stable across invocations of this process as the =
mapping is rarely kept, after an anonymisation run based on enumeration =
is finished.=20

=20

4.1.3.2 Hashing

=20

In this permutation process a hash function is applied to the IP =
addresses, which usually outputs numbers (values?) in the same dimension =
as the input, i.e. 32bit output for IPv4 and 128bit output for IPv6. A =
hash function computes this output number in a deterministic way from =
the IP address input. It therefore belongs to the stable class of =
anonymisation techniques. The hash function may also take an additional =
salt (seed?) parameter, so that not only the hash function type alone =
determines the actual mapping. This salt value must be kept secret in =
order to protect the non-reversibility of the hash function. Note that =
hash functions usually produce a minor amount of collisions in the =
output numbers, so that there does not exist an exact one-to-one mapping =
to the original IP addresses. The exception is the choice of a perfect =
hashing function, which does not show collisions. However such hash =
function needs to be tailored to the input data set (the set of IP =
addresses) and cannot be applied to a live stream of IP addresses, since =
one needs to know the complete set of different IP addresses in order to =
define the perfect hash function for this set.

=20

What do you think?

=20

Best regards

Carsten

=20

=20

Von: ipfix-bounces@ietf.org [mailto:ipfix-bounces@ietf.org] Im Auftrag =
von Elisa Boschi
Gesendet: Freitag, 26. Februar 2010 12:01
An: ipfix-chairs@tools.ietf.org
Cc: ipfix@ietf.org
Betreff: [IPFIX] Fwd: I-D Action:draft-ietf-ipfix-anon-02.txt

=20

Nevil, J=FCrgen,

as you've seen we've submitted a new revision of the anonymisation =
draft. The changes are relatively minor and address all remaining open =
issues and comments received so far. At this point we think we're ready =
for Working Group Last Call. In order to take advantage of the upcoming =
meeting we suggest starting WG last call soon so that we can discuss =
last call comments during the anon draft agenda slot.

thanks,
Elisa & Brian

-------- Original Message --------=20


Subject:=20

[IPFIX] I-D Action:draft-ietf-ipfix-anon-02.txt


Date:=20

Mon, 15 Feb 2010 04:00:01 -0800 (PST)


From:=20

Internet-Drafts@ietf.org


To:=20

i-d-announce@ietf.org


CC:=20

ipfix@ietf.org

=20

A New Internet-Draft is available from the on-line Internet-Drafts =
directories.
This draft is a work item of the IP Flow Information Export Working =
Group of the IETF.
=20
=20
        Title           : IP Flow Anonymisation Support
        Author(s)       : E. Boschi, B. Trammell
        Filename        : draft-ietf-ipfix-anon-02.txt
        Pages           : 37
        Date            : 2010-02-15
=20
This document describes anonymisation techniques for IP flow data and
the export of anonymised data using the IPFIX protocol.  It
categorizes common anonymisation schemes and defines the parameters
needed to describe them.  It provides guidelines for the
implementation of anonymised data export and storage over IPFIX, and
describes an information model and Options-based method for
anonymisation technique metadata export within the IPFIX protocol or
storage in IPFIX Files.
=20
A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-ipfix-anon-02.txt
=20
Internet-Drafts are also available by anonymous FTP at:
ftp://ftp.ietf.org/internet-drafts/
=20
Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.
=20

------_=_NextPart_001_01CAC770.D8589671
Content-Type: text/html;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<html xmlns:v=3D"urn:schemas-microsoft-com:vml" =
xmlns:o=3D"urn:schemas-microsoft-com:office:office" =
xmlns:w=3D"urn:schemas-microsoft-com:office:word" =
xmlns:m=3D"http://schemas.microsoft.com/office/2004/12/omml" =
xmlns=3D"http://www.w3.org/TR/REC-html40">

<head>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Diso-8859-1">


<meta name=3DGenerator content=3D"Microsoft Word 12 (filtered medium)">
<style>
<!--
 /* Font Definitions */
 @font-face
	{font-family:"Cambria Math";
	panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
	{font-family:Calibri;
	panose-1:2 15 5 2 2 2 4 3 2 4;}
@font-face
	{font-family:Tahoma;
	panose-1:2 11 6 4 3 5 4 4 2 4;}
@font-face
	{font-family:Consolas;
	panose-1:2 11 6 9 2 2 4 3 2 4;}
 /* Style Definitions */
 p.MsoNormal, li.MsoNormal, div.MsoNormal
	{margin:0cm;
	margin-bottom:.0001pt;
	font-size:12.0pt;
	font-family:"Times New Roman","serif";
	color:black;}
a:link, span.MsoHyperlink
	{mso-style-priority:99;
	color:blue;
	text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
	{mso-style-priority:99;
	color:purple;
	text-decoration:underline;}
pre
	{mso-style-priority:99;
	mso-style-link:"HTML Vorformatiert Zchn";
	margin:0cm;
	margin-bottom:.0001pt;
	font-size:10.0pt;
	font-family:"Courier New";
	color:black;}
span.HTMLVorformatiertZchn
	{mso-style-name:"HTML Vorformatiert Zchn";
	mso-style-priority:99;
	mso-style-link:"HTML Vorformatiert";
	font-family:Consolas;
	color:black;}
span.E-MailFormatvorlage19
	{mso-style-type:personal-reply;
	font-family:"Calibri","sans-serif";
	color:#1F497D;}
.MsoChpDefault
	{mso-style-type:export-only;
	font-size:10.0pt;}
@page Section1
	{size:612.0pt 792.0pt;
	margin:70.85pt 70.85pt 2.0cm 70.85pt;}
div.Section1
	{page:Section1;}
-->
</style>
<!--[if gte mso 9]><xml>
 <o:shapedefaults v:ext=3D"edit" spidmax=3D"1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
 <o:shapelayout v:ext=3D"edit">
  <o:idmap v:ext=3D"edit" data=3D"1" />
 </o:shapelayout></xml><![endif]-->
</head>

<body bgcolor=3Dwhite lang=3DDE link=3Dblue vlink=3Dpurple>

<div class=3DSection1>

<p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>Dear Elisa, Brian,<o:p></o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>I reviewed the latest version of the draft and, apart =
from a few
suggestions for improved wording, found only two issues, which I would =
consider
to address:<o:p></o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>If I&#8217;m not completely mistaken then there exists no
section about &#8220;IP address enumeration&#8221; yet. Since this is a =
commonly
used technique, I would give it some text (see =
below).<o:p></o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>The second is similar; there is no explicit notion of =
hashing of
IP addresses, which is often used as well, but has some deviations from =
a
perfect permutation, namely collisions.<o:p></o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>I wrote some new text as a suggestion, following below. =
Maybe it
could be a subsection of 4.1.3 Permutation. I also think it could be a =
bit
shorter but would leave this open to you. Comments are of course =
welcome.<o:p></o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>&lt;Current text:&gt;<o:p></o:p></span></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>4.1.3.=A0 Permutation<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'><o:p>&nbsp;</o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>=A0=A0 Permutation is a direct substitution technique, =
replacing
each IP<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>=A0=A0 address with an address selected from the set of =
possible IP<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>=A0=A0 addresses, guaranteeing that each anonymised =
address
represents a<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>=A0=A0 unique original address.=A0 The selection function =
is often
random,<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>=A0=A0 though it is not necessarily so. =A0Permutation =
does not
preserve any<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>=A0=A0 structural information about a network, but it =
does preserve
the<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>=A0=A0 unique count of IP addresses.=A0 Any application =
that requires
more<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>=A0=A0 structure than host-uniqueness will not be able to =
use
permuted IP<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>=A0=A0 addresses.<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'><o:p>&nbsp;</o:p></span></b></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>&lt;New test could start here:&gt;<o:p></o:p></span></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>As subcategories of permutation exist stateful and =
stateless
permutation techniques: Stateless permutation (e.g., by exchange of =
address
bits in a fixed way) yields a consistent, deterministic mapping of =
addresses,
while stateful permutation (e.g., by renumbering found IP addresses in a =
packet
trace by giving each a new number in a consecutive way) does not yield a
consistent anonymisation across multiple packet traces. See section xxx =
for
details on the definition of anonymisation mapping =
stability.<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'><o:p>&nbsp;</o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>4.1.3.1 Enumeration<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'><o:p>&nbsp;</o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>=A0=A0 Enumeration is a substitution function that =
retains the chronological
order in which different IP addresses occurred on the network or in an
anonymised packet data trace while eliminating the original IP =
addresses. IP
Addresses are usually replaced by consecutive addresses, e.g. counting =
up from
1.1.1.1, 1.1.1.2, and so on in the IPv4 case. During this process a map =
is
extended whenever a previously unseen IP address is observed. This kind =
of
permutation produces a unique mapping, which is usually not stable =
across
invocations of this process as the mapping is rarely kept, after an
anonymisation run based on enumeration is finished. =
<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'><o:p>&nbsp;</o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>4.1.3.2 Hashing<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'><o:p>&nbsp;</o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'>In this permutation process a hash function is applied to =
the IP
addresses, which usually outputs numbers (values?) in the same dimension =
as the
input, i.e. 32bit output for IPv4 and 128bit output for IPv6. A hash =
function
computes this output number in a deterministic way from the IP address =
input.
It therefore belongs to the stable class of anonymisation techniques. =
The hash
function may also take an additional salt (seed?) parameter, so that not =
only
the hash function type alone determines the actual mapping. This salt =
value
must be kept secret in order to protect the non-reversibility of the =
hash
function. Note that hash functions usually produce a minor amount of =
collisions
in the output numbers, so that there does not exist an exact one-to-one =
mapping
to the original IP addresses. The exception is the choice of a perfect =
hashing
function, which does not show collisions. However such hash function =
needs to
be tailored to the input data set (the set of IP addresses) and cannot =
be
applied to a live stream of IP addresses, since one needs to know the =
complete
set of different IP addresses in order to define the perfect hash =
function for
this set.<o:p></o:p></span></b></p>

<p class=3DMsoNormal><b><span lang=3DEN-US =
style=3D'font-size:10.0pt;font-family:
"Courier New"'><o:p>&nbsp;</o:p></span></b></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>What do you think?<o:p></o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>Best regards<o:p></o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>Carsten<o:p></o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<p class=3DMsoNormal><span lang=3DEN-US =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<div>

<div style=3D'border:none;border-top:solid #B5C4DF 1.0pt;padding:3.0pt =
0cm 0cm 0cm'>

<p class=3DMsoNormal><b><span =
style=3D'font-size:10.0pt;font-family:"Tahoma","sans-serif";
color:windowtext'>Von:</span></b><span =
style=3D'font-size:10.0pt;font-family:
"Tahoma","sans-serif";color:windowtext'> ipfix-bounces@ietf.org
[mailto:ipfix-bounces@ietf.org] <b>Im Auftrag von </b>Elisa Boschi<br>
<b>Gesendet:</b> Freitag, 26. Februar 2010 12:01<br>
<b>An:</b> ipfix-chairs@tools.ietf.org<br>
<b>Cc:</b> ipfix@ietf.org<br>
<b>Betreff:</b> [IPFIX] Fwd: I-D =
Action:draft-ietf-ipfix-anon-02.txt<o:p></o:p></span></p>

</div>

</div>

<p class=3DMsoNormal><o:p>&nbsp;</o:p></p>

<p class=3DMsoNormal>Nevil, J=FCrgen,<br>
<br>
as you've seen we've submitted a new revision of the anonymisation =
draft. The
changes are relatively minor and address all remaining open issues and =
comments
received so far. At this point we think we're ready for Working Group =
Last
Call. In order to take advantage of the upcoming meeting we suggest =
starting WG
last call soon so that we can discuss last call comments during the anon =
draft
agenda slot.<br>
<br>
thanks,<br>
Elisa &amp; Brian<br>
<br>
-------- Original Message -------- <o:p></o:p></p>

<table class=3DMsoNormalTable border=3D0 cellspacing=3D0 =
cellpadding=3D0>
 <tr>
  <td nowrap valign=3Dtop style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal align=3Dright =
style=3D'text-align:right'><b>Subject: <o:p></o:p></b></p>
  </td>
  <td style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal>[IPFIX] I-D =
Action:draft-ietf-ipfix-anon-02.txt<o:p></o:p></p>
  </td>
 </tr>
 <tr>
  <td nowrap valign=3Dtop style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal align=3Dright style=3D'text-align:right'><b>Date: =
<o:p></o:p></b></p>
  </td>
  <td style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal>Mon, 15 Feb 2010 04:00:01 -0800 =
(PST)<o:p></o:p></p>
  </td>
 </tr>
 <tr>
  <td nowrap valign=3Dtop style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal align=3Dright style=3D'text-align:right'><b>From: =
<o:p></o:p></b></p>
  </td>
  <td style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal><a =
href=3D"mailto:Internet-Drafts@ietf.org">Internet-Drafts@ietf.org</a><o:p=
></o:p></p>
  </td>
 </tr>
 <tr>
  <td nowrap valign=3Dtop style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal align=3Dright style=3D'text-align:right'><b>To: =
<o:p></o:p></b></p>
  </td>
  <td style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal><a =
href=3D"mailto:i-d-announce@ietf.org">i-d-announce@ietf.org</a><o:p></o:p=
></p>
  </td>
 </tr>
 <tr>
  <td nowrap valign=3Dtop style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal align=3Dright style=3D'text-align:right'><b>CC: =
<o:p></o:p></b></p>
  </td>
  <td style=3D'padding:0cm 0cm 0cm 0cm'>
  <p class=3DMsoNormal><a =
href=3D"mailto:ipfix@ietf.org">ipfix@ietf.org</a><o:p></o:p></p>
  </td>
 </tr>
</table>

<p class=3DMsoNormal =
style=3D'margin-bottom:12.0pt'><o:p>&nbsp;</o:p></p>

<pre>A New Internet-Draft is available from the on-line Internet-Drafts =
directories.<o:p></o:p></pre><pre>This draft is a work item of the IP =
Flow Information Export Working Group of the =
IETF.<o:p></o:p></pre><pre><o:p>&nbsp;</o:p></pre><pre><o:p>&nbsp;</o:p><=
/pre><pre>=A0=A0=A0=A0=A0=A0=A0 Title=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0 : IP =
Flow Anonymisation Support<o:p></o:p></pre><pre>=A0=A0=A0=A0=A0=A0=A0 =
Author(s)=A0=A0=A0=A0=A0=A0 : E. Boschi, B. =
Trammell<o:p></o:p></pre><pre>=A0=A0=A0=A0=A0=A0=A0 =
Filename=A0=A0=A0=A0=A0=A0=A0 : =
draft-ietf-ipfix-anon-02.txt<o:p></o:p></pre><pre>=A0=A0=A0=A0=A0=A0=A0 =
Pages=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0 : =
37<o:p></o:p></pre><pre>=A0=A0=A0=A0=A0=A0=A0 =
Date=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0 : =
2010-02-15<o:p></o:p></pre><pre><o:p>&nbsp;</o:p></pre><pre>This =
document describes anonymisation techniques for IP flow data =
and<o:p></o:p></pre><pre>the export of anonymised data using the IPFIX =
protocol.=A0 It<o:p></o:p></pre><pre>categorizes common anonymisation =
schemes and defines the parameters<o:p></o:p></pre><pre>needed to =
describe them.=A0 It provides guidelines for =
the<o:p></o:p></pre><pre>implementation of anonymised data export and =
storage over IPFIX, and<o:p></o:p></pre><pre>describes an information =
model and Options-based method for<o:p></o:p></pre><pre>anonymisation =
technique metadata export within the IPFIX protocol =
or<o:p></o:p></pre><pre>storage in IPFIX =
Files.<o:p></o:p></pre><pre><o:p>&nbsp;</o:p></pre><pre>A URL for this =
Internet-Draft is:<o:p></o:p></pre><pre><a
href=3D"http://www.ietf.org/internet-drafts/draft-ietf-ipfix-anon-02.txt"=
>http://www.ietf.org/internet-drafts/draft-ietf-ipfix-anon-02.txt</a><o:p=
></o:p></pre><pre><o:p>&nbsp;</o:p></pre><pre>Internet-Drafts are also =
available by anonymous FTP at:<o:p></o:p></pre><pre><a
href=3D"ftp://ftp.ietf.org/internet-drafts/">ftp://ftp.ietf.org/internet-=
drafts/</a><o:p></o:p></pre><pre><o:p>&nbsp;</o:p></pre><pre>Below is =
the data which will enable a MIME compliant mail =
reader<o:p></o:p></pre><pre>implementation to automatically retrieve the =
ASCII version of =
the<o:p></o:p></pre><pre>Internet-Draft.<o:p></o:p></pre><pre><o:p>&nbsp;=
</o:p></pre></div>

</body>

</html>

------_=_NextPart_001_01CAC770.D8589671--

From muenz@net.in.tum.de  Fri Mar 19 08:27:01 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 9CC413A6A64 for <ipfix@core3.amsl.com>; Fri, 19 Mar 2010 08:27:01 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.554
X-Spam-Level: 
X-Spam-Status: No, score=-1.554 tagged_above=-999 required=5 tests=[AWL=-0.435, BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 45jAzQikmxd8 for <ipfix@core3.amsl.com>; Fri, 19 Mar 2010 08:27:00 -0700 (PDT)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id E59393A6AC7 for <ipfix@ietf.org>; Fri, 19 Mar 2010 08:26:54 -0700 (PDT)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id E382C20B32F9; Fri, 19 Mar 2010 16:27:05 +0100 (CET)
Message-ID: <4BA397E5.1040106@net.in.tum.de>
Date: Fri, 19 Mar 2010 16:27:33 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.24 (Windows/20100228)
MIME-Version: 1.0
To: Thomas Dietz <Thomas.Dietz@nw.neclab.eu>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com><4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com><4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com><4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E10EE.9050406@net.in.tum.de> <547F018265F92642B577B986577D671C012A2D27@VENUS.office> <4B9F9574.1080606@net.in.tum.de> <547F018265F92642B577B986577D671C012A2D5B@VENUS.office> <4B9FA451.8080802@net.in.tum.de> <547F018265F92642B577B986577D671C012A31C5@VENUS.office>
In-Reply-To: <547F018265F92642B577B986577D671C012A31C5@VENUS.office>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms020102070508050503050608"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX-CONFIG: Questions about counters adopted from IPFIXMIB
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 19 Mar 2010 15:27:01 -0000

This is a cryptographically signed message in MIME format.

--------------ms020102070508050503050608
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit


Hi Thomas,

Thomas Dietz wrote:
>> My original point was that I would like to get away from sysUpTime
>> because it is not sure from where you can get it with Netconf.
>>
>> Discontinuity times like in the MIB are ok, but then they should
>> contain
>> absolute time stamps.
> 
> If you are sure that you can always get the correct absolute time then this
> is fine. I am not an Netconf expert, so it's up to you to judge.

Each IPFIX Message contains the export time in UTC.
So, as IPFIX requires an absolute timestamp, I assume that devices have
a more or less synchronized clock.

>> Maybe it is the easiest to only make this little change and keep the
>> definitions of the other counters as they are in the MIB.
> 
> This would be the easiest option from my point of view.

Good.

So, at the end, we keep counter64 (non-zero-based) and have absolute
discontinuity timestamps.

Regards,
Gerhard

--------------ms020102070508050503050608
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms020102070508050503050608--

From bclaise@cisco.com  Sat Mar 20 06:13:44 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id B764C3A6850 for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 06:13:44 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.24
X-Spam-Level: 
X-Spam-Status: No, score=-0.24 tagged_above=-999 required=5 tests=[AWL=-1.371,  BAYES_50=0.001, DNS_FROM_OPENWHOIS=1.13]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 0at-7CepxS+7 for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 06:13:42 -0700 (PDT)
Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id 8B7BF3A689D for <ipfix@ietf.org>; Sat, 20 Mar 2010 06:13:42 -0700 (PDT)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2KCmcPC000824; Sat, 20 Mar 2010 13:48:38 +0100 (CET)
Received: from [10.82.226.43] (rtp-vpn1-555.cisco.com [10.82.226.43]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2KCmWkX019927; Sat, 20 Mar 2010 13:48:36 +0100 (CET)
Message-ID: <4BA4C41F.9020407@cisco.com>
Date: Sat, 20 Mar 2010 13:48:31 +0100
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: Brian Trammell <trammell@tik.ee.ethz.ch>
References: <4AF73525.8050009@net.in.tum.de>	<4AF8F999.3000207@cisco.com>	<F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>	<4B9E33BE.7070907@cisco.com>	<7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch>	<4B9F520E.8060507@cisco.com>	<DB6E59D9-B373-4919-BB58-00EB26014564@tik.ee.ethz.ch>	<4BA0C26D.9070901@cisco.com>	<6CD84D0C-333F-4C65-8E41-B1DF3774BA15@tik.ee.ethz.ch>	<4BA1E4C4.3000204@cisco.com> <4BA20CD5.6010800@cisco.com> <EA4C7E42-1F67-4898-8CA3-D06A4FCD4C2E@tik.ee.ethz.ch>
In-Reply-To: <EA4C7E42-1F67-4898-8CA3-D06A4FCD4C2E@tik.ee.ethz.ch>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] [Sender: ipfix-bounces@ietf.org] Re: Semantic and	structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sat, 20 Mar 2010 13:13:44 -0000

Brian,
> On Mar 18, 2010, at 4:21 AM, Paul Aitken wrote:
>
>    
>> Benoit, Brian,
>>
>>      
>>>> On reflection, I would suggest defining these in terms of multiplicities (as we're dealing with N-element lists) as opposed to logical operations. So I'd propose the following:
>>>>          
>>
>>      
>>>> 0x00: undefined (default) (CP may assume allOf)
>>>>          
>>>    This corresponds to NONE
>>>        
> Ahhh..... okay. This is confusing, though. I'd _definitely_ call this "undefined", because "none" could be interpreted as "not".
>    
"undefined" is a better term, you're right.
>    
>>      
>>>> 0x01: allOf
>>>>          
>>>    This corresponds to AND
>>>        
>>
>>      
>>>> 0x02: oneOrMoreOf
>>>>          
>>>    This corresponds to OR
>>>        
>>
>>      
>>>> 0x03: exactlyOneOf
>>>>          
>>>    Not sure what it corresponds to. At first glance, I don't see a use case for it.
>>>        
>> 0x02 and 0x03 are two different kinds of OR: "at least one of" as opposed to "just one of".
>>      
> With two elements in the list, it corresponds to XOR; exactlyOneOf is the multiplicity extension of this (which lends itself nicely to clarity for lists of more than two).
>
> The use case is when you want to express uncertainty within a certain set: for example, say a mediator knows that a flow was observed on one of a set of interfaces at a MP on an original exporter (by virtue of having come from that exporter), but that the flow ONLY traversed one of them (say, because the exporter is only looking at ingress interfaces at the network border, and the routing configuration precludes balancing across those interfaces). With oneOrMoreOf ("OR"), the mediator cannot exclude the possibility it was seen on both. It's also one of the basic multiplicities {0, 1, 1..n, n}, so should be included for completeness in any case.
>    
Point taken.
I now see the need for exactlyOneOf versus oneOrMoreOf

We're slowly but surely converging. Great.

Regards, Benoit.
>    
>>      
>>>> 0x04: noneOf
>>>>          
>>> This corresponds to NOT
>>>        
>> - which introduces the idea of NOT as a semantic in its own right, rather than simply as a modifier for other semantics.
>>
>> However, "noneOf (x, y, z)" could be "NOT-AND (x, y, z)". Or possibly "NOT-OR (x, y, z)", though the distinction is subtle.
>>
>> So NOT may still be considered as a modifier, rather than a distinct and separate semantic.
>>      
> I think the argument for the flag goes away if you treat these as multiplicities. I'd treat the flag as a set negation: NOT AND = allOf (set) (negated) = noneOf: the represented value set can contain zero or more elements in the universe that do not appear in the list.
>
> Here, NOT OR = oneOrMoreOf (set) (negated) = notAllOf, meaning, the represented value set can contain zero or more elements in the universe but not in the list, as well as up to n-1 elements in the list, but not all n elements of the list; at least one (unspecified) value in the list must be absent from the set of represented values. It is not clear to me that this is useful...
>
> NOT XOR = exactlyOneOf (set) (negated) seems even less useful.
>
>
>    
>>
>>      
>>>> If I understand what you mean by ORDERED, add:
>>>> 0x05: allInOrder
>>>>          
>>> This corresponds to ORDERED
>>>        
>>
>>      
>>>> If we really want to do ranges:
>>>> 0x06: inclusiveRanges (MUST have even number of elements, and elements MUST be numeric; this is then equivalent to inclusive allOf [e_0,e_1], [e_2,e_3], ... , [e_n-1]..[e_n]);
>>>>          
>>>    This corresponds to RANGE
>>> What is required? NONE, AND, OR, potentially NOT, ORDERED
>>>        
>> And RANGE.
>>      
> The case for this one's not clear to me. I see the use case, but I'm undecided on whether it's in scope for this draft...
>
>    
>>      
>>>> Still unanswered is whether we want to force all basicList/subTemplateList/subTemplateMultiList instances to state semantics, or whether we want "semanticless" versions as well...
>>>>
>>>>          
>>> No strong feeling about this one.
>>> Maybe it would be better to have additional semanticBasicList, semanticSubTemplateList, semanticSubTemplateMultiList
>>>        
>> Then there's no need for semantics = NONE.
>>
>> I think it's more straightforward not to have "semanticless" versions.
>>      
> arguments for "semanticless" versions: clean representation and no waste of a byte per list (okay, minor considering how big lists are likely to be, but maybe annoying for alignment purposes) for the (majority of?) applications that don't care about semantics. Arguments against: we need an undefined (again, "NONE" might lead to confusion) semantic flag anyway so now we have two legal ways to represent the same thing, which makes the collector's life harder... plus it doesn't look as clean in the IE registry.
>
> No super strong opinion here either way...
>
> Best regards,
>
> Brian
>    


From bclaise@cisco.com  Sat Mar 20 11:07:44 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id DCD133A683B for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 11:07:44 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.437
X-Spam-Level: 
X-Spam-Status: No, score=-1.437 tagged_above=-999 required=5 tests=[AWL=0.032,  BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id VhSq2FL0G10G for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 11:07:43 -0700 (PDT)
Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id D97243A686A for <ipfix@ietf.org>; Sat, 20 Mar 2010 11:07:30 -0700 (PDT)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2KI7iL4014725; Sat, 20 Mar 2010 19:07:44 +0100 (CET)
Received: from [10.55.89.147] (dhcp-10-55-89-147.cisco.com [10.55.89.147]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2KI7bv4014535; Sat, 20 Mar 2010 19:07:37 +0100 (CET)
Message-ID: <4BA50EE8.4020502@cisco.com>
Date: Sat, 20 Mar 2010 19:07:36 +0100
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: Paul Aitken <paitken@cisco.com>
References: <4AF73525.8050009@net.in.tum.de>	<4AF8F999.3000207@cisco.com>	<F60CA342-F488-4179-8AB8-079D32D26BCD@tik.ee.ethz.ch>	<4B9E33BE.7070907@cisco.com>	<7A1F2B11-B407-4BAF-8481-F867CCDEF5FC@tik.ee.ethz.ch>	<4B9F520E.8060507@cisco.com>	<DB6E59D9-B373-4919-BB58-00EB26014564@tik.ee.ethz.ch>	<4BA0C26D.9070901@cisco.com>	<6CD84D0C-333F-4C65-8E41-B1DF3774BA15@tik.ee.ethz.ch>	<4BA1E4C4.3000204@cisco.com> <4BA20CD5.6010800@cisco.com>
In-Reply-To: <4BA20CD5.6010800@cisco.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] [Sender: ipfix-bounces@ietf.org] Re: Semantic and	structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sat, 20 Mar 2010 18:07:45 -0000

Paul,
>
>>> Still unanswered is whether we want to force all 
>>> basicList/subTemplateList/subTemplateMultiList instances to state 
>>> semantics, or whether we want "semanticless" versions as well...
>> No strong feeling about this one.
>> Maybe it would be better to have additional semanticBasicList, 
>> semanticSubTemplateList, semanticSubTemplateMultiList
>
> Then there's no need for semantics = NONE.
>
> I think it's more straightforward not to have "semanticless" versions.

Ok,. let's shoot for that one.

Regards, Benoit.

From n.brownlee@auckland.ac.nz  Sat Mar 20 16:25:14 2010
Return-Path: <n.brownlee@auckland.ac.nz>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 2BD173A68B2 for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 16:25:14 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.869
X-Spam-Level: 
X-Spam-Status: No, score=-2.869 tagged_above=-999 required=5 tests=[BAYES_50=0.001, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id L6DPIvV970Mz for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 16:25:13 -0700 (PDT)
Received: from mailhost.auckland.ac.nz (moe.its.auckland.ac.nz [130.216.12.37]) by core3.amsl.com (Postfix) with ESMTP id 4DC983A6A1F for <ipfix@ietf.org>; Sat, 20 Mar 2010 16:24:10 -0700 (PDT)
Received: from localhost (localhost.localdomain [127.0.0.1]) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 38D28173023; Sun, 21 Mar 2010 12:24:24 +1300 (NZDT)
X-Virus-Scanned: by amavisd-new at mailhost.auckland.ac.nz
Received: from mailhost.auckland.ac.nz ([127.0.0.1]) by localhost (moe.its.auckland.ac.nz [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id JURLsooHefaK; Sun, 21 Mar 2010 12:24:24 +1300 (NZDT)
Received: from [172.27.171.231] (207.88.181.2.ptr.us.xo.net [207.88.181.2]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 84792170A95; Sun, 21 Mar 2010 12:24:22 +1300 (NZDT)
Message-ID: <4BA55924.7050905@auckland.ac.nz>
Date: Sat, 20 Mar 2010 16:24:20 -0700
From: Nevil Brownlee <n.brownlee@auckland.ac.nz>
User-Agent: Thunderbird 2.0.0.24 (X11/20100317)
MIME-Version: 1.0
To: Brian Trammell <trammell@tik.ee.ethz.ch>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: IPFIX list <ipfix@ietf.org>
Subject: [IPFIX] Errata 1988 and 2030 for RFC 5655
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sat, 20 Mar 2010 23:25:14 -0000

Hi Brian:

These two errata are Technical, would you have a look at them please,
and let us know whether we should verify them.

Thanks, Nevil

-- 
---------------------------------------------------------------------
  Nevil Brownlee                    Computer Science Department | ITS
  Phone: +64 9 373 7599 x88941             The University of Auckland
  FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand


From n.brownlee@auckland.ac.nz  Sat Mar 20 16:52:08 2010
Return-Path: <n.brownlee@auckland.ac.nz>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 95D333A6832 for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 16:52:08 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -4.169
X-Spam-Level: 
X-Spam-Status: No, score=-4.169 tagged_above=-999 required=5 tests=[AWL=1.300,  BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 1tjlQvkrSv0D for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 16:52:07 -0700 (PDT)
Received: from mailhost.auckland.ac.nz (larry.its.auckland.ac.nz [130.216.12.34]) by core3.amsl.com (Postfix) with ESMTP id 3AB7A3A6889 for <ipfix@ietf.org>; Sat, 20 Mar 2010 16:52:06 -0700 (PDT)
Received: from localhost (localhost.localdomain [127.0.0.1]) by mailhost.auckland.ac.nz (Postfix) with ESMTP id CC0771BBB3; Sun, 21 Mar 2010 12:20:52 +1300 (NZDT)
X-Virus-Scanned: by amavisd-new at mailhost.auckland.ac.nz
Received: from mailhost.auckland.ac.nz ([127.0.0.1]) by localhost (larry.its.auckland.ac.nz [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id KHPfjqnR8ZEq; Sun, 21 Mar 2010 12:20:52 +1300 (NZDT)
Received: from [172.27.171.231] (207.88.181.2.ptr.us.xo.net [207.88.181.2]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mailhost.auckland.ac.nz (Postfix) with ESMTP id AF1B618D89; Sun, 21 Mar 2010 12:20:48 +1300 (NZDT)
Message-ID: <4BA5584E.3060508@auckland.ac.nz>
Date: Sat, 20 Mar 2010 16:20:46 -0700
From: Nevil Brownlee <n.brownlee@auckland.ac.nz>
User-Agent: Thunderbird 2.0.0.24 (X11/20100317)
MIME-Version: 1.0
To: Gerhard Muenz <muenz@net.in.tum.de>
References: <4B9AC2F8.1090504@cisco.com> <4B9B87BD.7020201@net.in.tum.de>
In-Reply-To: <4B9B87BD.7020201@net.in.tum.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: "ipfix@ietf.org" <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX errata 2052 to 2059
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sat, 20 Mar 2010 23:52:08 -0000

Hi Gerhard & Paul:

I'm looking through our 'unverified' errata - do we have consensus
for Paul's Errata 2052, i.e. change SelectorId to unsigned64?

Cheers, Nevil


Gerhard Muenz wrote:
> Hi Paul,
> 
> Still, I have not heard any good arguments why we need to change it to 
> unsigned64. Looking at the examples in RFC5476, it seems that the 
> original intention was unsigned32. This is inline with 
> ipfixSelectionProcessSelectorIndex in IPFIX-MIB.
> 
> So, why do you prefer 64bit, which does not only require changes in 
> RFC5477, but in RFC5476 as well, and which is inconsistent with IPFIX-MIB?
> 
> See also my previous mail:
> http://www.ietf.org/mail-archive/web/ipfix/current/msg05311.html
> 
> Regards,
> Gerhard
> 
> Paul Aitken wrote:
>> Dear all,
>>
>> Discussion of these errata ended ~ 2 weeks ago. Presumably nobody has 
>> any objections?
>>
>> http://www.rfc-editor.org/errata_search.php?rfc=5477
>> http://www.rfc-editor.org/errata_search.php?rfc=5476
>>
>> http://www.ietf.org/mail-archive/web/ipfix/current/msg05314.html
>> http://www.ietf.org/mail-archive/web/ipfix/current/msg05315.html
>>
>>
>> Dan, what's the next step? Ideally an update to the IANA registry.
>>
>>
>> Thanks,
>> P.

-- 
---------------------------------------------------------------------
  Nevil Brownlee                    Computer Science Department | ITS
  Phone: +64 9 373 7599 x88941             The University of Auckland
  FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand


From paitken@cisco.com  Sat Mar 20 16:58:10 2010
Return-Path: <paitken@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id E2E213A686D for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 16:58:10 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -3.491
X-Spam-Level: 
X-Spam-Status: No, score=-3.491 tagged_above=-999 required=5 tests=[AWL=-2.022, BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id FqS6Nq6WtBp0 for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 16:58:10 -0700 (PDT)
Received: from ams-iport-2.cisco.com (ams-iport-2.cisco.com [144.254.224.141]) by core3.amsl.com (Postfix) with ESMTP id 9DF5C3A6860 for <ipfix@ietf.org>; Sat, 20 Mar 2010 16:58:09 -0700 (PDT)
Authentication-Results: ams-iport-2.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AkUBAKr9pEuQ/uCWe2dsb2JhbACbPhUBAQsLJAYcoi+YM4R9BI5T
X-IronPort-AV: E=Sophos;i="4.51,280,1267401600";  d="scan'208";a="4594475"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-2.cisco.com with ESMTP; 20 Mar 2010 23:24:24 +0000
Received: from [10.61.98.100] (dhcp-10-61-98-100.cisco.com [10.61.98.100]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2KNwN9L013634; Sat, 20 Mar 2010 23:58:23 GMT
Message-ID: <4BA5611F.90309@cisco.com>
Date: Sat, 20 Mar 2010 23:58:23 +0000
From: Paul Aitken <paitken@cisco.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-GB; rv:1.8.1.22) Gecko/20090605 SeaMonkey/1.1.17 (Ubuntu-1.1.17+nobinonly-0ubuntu0.9.04.1)
MIME-Version: 1.0
To: Nevil Brownlee <n.brownlee@auckland.ac.nz>
References: <4B9AC2F8.1090504@cisco.com> <4B9B87BD.7020201@net.in.tum.de> <4BA5584E.3060508@auckland.ac.nz>
In-Reply-To: <4BA5584E.3060508@auckland.ac.nz>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: "ipfix@ietf.org" <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX errata 2052 to 2059
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sat, 20 Mar 2010 23:58:11 -0000

Nevil,

> I'm looking through our 'unverified' errata - do we have consensus
> for Paul's Errata 2052, i.e. change SelectorId to unsigned64?

The consensus seems to be a change to unsigned32.

Thanks,
P.

-- 
Paul Aitken
Cisco Systems Ltd, Edinburgh, Scotland.
http://www.cisco.com/web/about/doing_business/legal/cri/index.html

From akoba@nttv6.net  Sat Mar 20 22:25:41 2010
Return-Path: <akoba@nttv6.net>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id C5A0C3A683C for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 22:25:39 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 3.929
X-Spam-Level: ***
X-Spam-Status: No, score=3.929 tagged_above=-999 required=5 tests=[BAYES_50=0.001, DNS_FROM_OPENWHOIS=1.13, NO_RELAYS=-0.001, SUBJ_RE_NUM=2.799]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id JrFv-U1WsxcC for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 22:25:33 -0700 (PDT)
Received: from mail.nttv6.net (mail.nttv6.net [IPv6:2001:fa8::25]) by core3.amsl.com (Postfix) with ESMTP id 747803A682C for <ipfix@ietf.org>; Sat, 20 Mar 2010 22:25:05 -0700 (PDT)
Received: from [192.168.0.134] (mail.nttv6.net [IPv6:2001:fa8::25]) by mail.nttv6.net (8.14.3/8.14.3) with ESMTP id o2L5P6SQ064029; Sun, 21 Mar 2010 14:25:08 +0900 (JST) (envelope-from akoba@nttv6.net)
Date: Sun, 21 Mar 2010 14:25:12 +0900
From: Atsushi Kobayashi <akoba@nttv6.net>
To: Benoit Claise <bclaise@cisco.com>
In-Reply-To: <4BA1E9C4.9030204@cisco.com>
References: <20100318051549.AFC9.17391CF2@nttv6.net> <4BA1E9C4.9030204@cisco.com>
Message-Id: <20100321142504.9220.17391CF2@nttv6.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
X-Mailer: Becky! ver. 2.52.03 [ja] (Unregistered)
X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.2.3 (mail.nttv6.net [IPv6:2001:fa8::25]); Sun, 21 Mar 2010 14:25:17 +0900 (JST)
Cc: ipfix@ietf.org
Subject: Re: [IPFIX] Semantic and structured data
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sun, 21 Mar 2010 05:25:41 -0000

Hi Benoit,

On Thu, 18 Mar 2010 09:52:20 +0100
Benoit Claise <bclaise@cisco.com> wrote:

> Kobayashi-san,
> > Hi Benoit, and all,
> >
> > I agree solution3 rather than new semantic IE.
> > As Benoit mentioned, semantic IE seems to increase the difficulty to
> > interpret them.
> >
> > As Brian mentioned, I think it should avoid to have unnecessary
> > complexity. But, I am not sure what type, e.g., NONE, OR, AND, NOR,
> > RANGE, and ORDERRED, is needed. I would like to avoid that one data
> > structure is represented in multiple ways.
> So do I.
>  From this, I conclude that a simple enumeration type is better compared 
> to the bitvector encoding.
> If we include a bitvector with one bit for NOT, we end up in this situation.
> 
>     NOT OR = NOR
>     NOT AND = NAND
>     NOT ORDERED = UNORDERED (*)
>     NOT UNORDERED = ORDERED (*)
>     NOT NONE => doesn't make sense  (**)
>     NOT RANGE => doesn't make sense  (**)
>     NOT NOT => doesn't make sense  (**)
> 
> 
> (*) implies that we can represent the same semantic different ways. Now, 
> we can argue that we don't need UNORDERED

If using "UNORDERED", we can use "AND" or "OR" alternatively.
In my conclusion, we can remove it.

> (**) add non possible and confusing semantics
> > To examine it, we needs more
> > practical examples in addition to: Selector Report
> > Interpretation and interface lists on aggregated Flow Records.
> >

That make sense.

> > How about BGP AS Path, Community? Anything else?
> >
> > When I have the following BGP AS Path mixing as-sequence and as-set, how
> > to present it?
> >
> > 10 20 30 40 {50,60}
> >
> > (basicList, ORDERED, (basicList, ORDERED, AS10,AS20,AS30,AS40),
> > (basicList, OR, AS50, AS60))
> >
> > Is it correct?
> >    
> Exactly, which implies that the ORDER semantic is required for this use 
> case.
> 

We needs more other examples.

Regards,
Atsushi


> Regards, Benoit.
> > Regards,
> > Atsushi
> >
> > On Wed, 17 Mar 2010 14:10:35 +0100
> > Benoit Claise<bclaise@cisco.com>  wrote:
> >
> >    
> >> Thanks Gerhard for your feedback.
> >> See inline.
> >>      
> >>> Hi all,
> >>>
> >>> Some general thoughts from my side:
> >>>
> >>> - I appreciate that you want to add a basic notion of semantic to the
> >>>    structured data.
> >>>        
> >> Great.
> >>      
> >>> - Up to now, semantic was not in the protocol but in the info model.
> >>>
> >>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >>> |0|               Field ID    |       Element Length            |
> >>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >>> | Semantic  |             BasicList Content ...                 |
> >>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >>> |                           ...                                 |
> >>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >>>
> >>>    Can't we encode Semantic in an IE?
> >>>    Then, we could do without a new IANA registry.
> >>>    And: This IE could be used for other purposes (e.g., in Templates)
> >>>        
> >> We thought about this one. However, one of the IPFIX principle is that
> >> the semantic of one IE can't depend on the value or position of some
> >> other IEs.
> >> One example of this is the (MPLS label position, MPLS label). Way before
> >> structured data, our initial implementation contained: MPLS label
> >> position, MPLS label.
> >> So the collector had to look first at the MPLS label position value in
> >> order to correctly understand the following  MPLS label.
> >> Somehow, with these conventions, we were having an information model on
> >> the top of the IE. This was wrong.
> >> Also it was not obvious for a collector, i.e. without hardcoding the
> >> information.
> >> Finally, what if the IE order changed within the flow record?
> >> Conclusion: we had to change back our implementation mplslabelposition1,
> >> mplslabelposition2, etc... Like we did in IPIFX ;-)
> >>
> >>
> >>      
> >>> - I'm not sure whether the bitvector encoding has an advantage over an
> >>>    enumeration type.
> >>>
> >>> Unfortunately, I do not have the time to participate in a deep
> >>> discussion - I'm busy with IPFIX-CONFIG and other stuff.
> >>>        
> >> Thanks and regards, Benoit.
> >>      
> >>> Regards,
> >>> Gerhard
> >>>
> >>> Benoit Claise wrote:
> >>>        
> >>>> Hi Brian,
> >>>>          
> >>>>> Hi, Benoit,
> >>>>>
> >>>>> Replies inline...
> >>>>>
> >>>>> On Mar 16, 2010, at 2:40 AM, Benoit Claise wrote:
> >>>>>
> >>>>>            
> >>>>>> Hi Brian,
> >>>>>>
> >>>>>> Thanks for your feedback.
> >>>>>>              
> >>>>>>> hi, Benoit, all...
> >>>>>>>
> >>>>>>> Agreed that we should do something about this (i.e., that solution
> >>>>>>> 1 is no solution.); that said, a few comments in no particular order:
> >>>>>>>
> >>>>>>> 1. In considering adding explicit semantics to structured data, we
> >>>>>>> as a WG are taking on the task of defining semantics for IPFIX as
> >>>>>>> a whole. Semantics, as I understand them, are in IPFIX largely
> >>>>>>> contextual and template dependent, but in almost all cases I can
> >>>>>>> think of it seems like these are implicitly "AND" semnantics (this
> >>>>>>> flow has source IP A _AND_ destination IP B _AND_...).
> >>>>>>>
> >>>>>>>                
> >>>>>> Agreed.
> >>>>>>              
> >>>>>>> We will need to make an explicit statement on this.
> >>>>>>>                
> >>>>>> Not sure why.
> >>>>>>              
> >>>>>>> We will need to determine whether these implicit semantics are a
> >>>>>>> property of the protocol (in which case Structured Data is really
> >>>>>>> a protocol-level extension), or a property of each information
> >>>>>>> element (in which case all 5103 IEs have implicit AND semantics;
> >>>>>>> question 2: will we need to add semantics to the IANA registry in
> >>>>>>> this case?). We will need to be quite careful about this. It's not
> >>>>>>> as simple as defining semantics within structured data then
> >>>>>>> calling it done.
> >>>>>>>
> >>>>>>>
> >>>>>>>                
> >>>>>> I'm not sure why each individual IE should have a semantic ... in
> >>>>>> the IANA registry.
> >>>>>>              
> >>>>> I'm not either. On reflection it seems like overkill. I'm just
> >>>>> saying that if we, as a WG, are moving from stating that semantics
> >>>>> are explicitly out of scope to defining them as in-scope, we need to
> >>>>> have a consistent approach, and answers to all the questions that
> >>>>> arise when we consider moving the protocol from a simple framing
> >>>>> mechanism to a framing mechanism with some logic behind it, so we
> >>>>> answer them once, so that all future efforts having to do with
> >>>>> semantics are consistent. This, I acknowledge, is an argument in
> >>>>> favor of solution 3...)
> >>>>>
> >>>>> One very simple new question that arises here, to illustrate my
> >>>>> point: Is it legal to export a record that has sourceIPAddress X AND
> >>>>> NOT sourceIPAddress X?
> >>>>>            
> >>>>    From a protocol point of view, yes
> >>>>   From a semantic point of view, I don't see a use case for that.
> >>>> Now, this question is not different that: with RFC5101,  is it legal
> >>>> to export a record that has two instances of sourceIPAddress?
> >>>>          
> >>>>>> As you wrote, "in almost all cases I can think of it seems like
> >>>>>> these are implicitly "AND" semnantics", so the case we try to solve
> >>>>>> is when there are multiple instances of a single IE. We know that
> >>>>>> RFC 5101 foresaw that case " The Collector MUST support the use of
> >>>>>> Templates containing multiple occurrences of the similar
> >>>>>> Information Elements", but the idea is not to change RFC5101.  If
> >>>>>> we put some semantic in the IPFIX structured data, that would solve
> >>>>>> the vast majority of our cases. Also, we could say: if you want
> >>>>>> some semantic when exporting multiple IEs, then you SHOULD use
> >>>>>> IPFIX structured data.
> >>>>>>
> >>>>>> Also, the default value for the semantic field in the IPFIX
> >>>>>> structured data SHOULD be "NONE", to express that flow record
> >>>>>> doesn't include any semantic.... like in RFC5101. You might draw
> >>>>>> your own conclusion, maybe because you know your network, maybe
> >>>>>> because you have configured the exporter, but then it's your decision.
> >>>>>> The way I see the proposed solution is: in IPFIX structured data,
> >>>>>> you MAY use the semantic field as a way to express the relationship
> >>>>>> between IEs within the structure.
> >>>>>>              
> >>>>>>> 2. I don't consider draft-sommer-ipfix-mediator-ext-01 a valid
> >>>>>>> argument against Solution 2, as it's trying to solve a somewhat
> >>>>>>> different and more limited problem than structured data. Solution
> >>>>>>> 2 _might_ cause a problem for this draft, but certainly not the
> >>>>>>> other way around (unless we as a WG want to subsume that work into
> >>>>>>> this draft, which would probably require rechartering...). Also, I
> >>>>>>> don't think we need semantics for all of the list types, just
> >>>>>>> basicList (Illustrative question: How do I meaningfully interpret
> >>>>>>> two "or" subTemplateMultiLists with disjoint IE sets? Two "and"
> >>>>>>> subTemplateMultiLists? Nestings thereof?)... So, we don't really
> >>>>>>> have an explosion to deal with if we do Solution 2 correctly:
> >>>>>>> andBasicList, orBasicList, xorBasicList, notBasicList; Four IEs,
> >>>>>>> nestable, done. What can't we do with those four IEs? In this
> >>>>>>> case, we could even step back and say that semantics outside these
> >>>>>>> four are within the protocol explicitly _undefined_, and to be
> >>>>>>> interpreted withi
> >>>>>>>                
> >>> n the
> >>>        
> >>>>>>>    context of each Template.
> >>>>>>>
> >>>>>>>
> >>>>>>>                
> >>>>>> If I translate some more what I wrote "While it's solved the router
> >>>>>> and most mediation function needs today", this would be "Me,
> >>>>>> myself, and I don't need more that OR and AND now ;-)".
> >>>>>> However, who am I to tell that others don't need it now... and that
> >>>>>> the logical solution is to use the IPFIX structured data
> >>>>>> Furthermore, if we think a little bit longer term, the next big
> >>>>>> step in IPFIX is the mediation function. In my company, every
> >>>>>> features want to export his own data with NetFlow/IPFIX... up to
> >>>>>> the point where a CPE would not have enough bandwidth across the
> >>>>>> WAN to export all the "management" information. So we'll need more
> >>>>>> and more of aggregated flow records (both in time and space) even
> >>>>>> in the router. Again, the logical solution will be to use the IPFIX
> >>>>>> structured data. At this point, we will most probably need
> >>>>>> something else than OR and AND, i.e. RANGE, ORDERRED, etc...
> >>>>>>
> >>>>>> An example of subTemplateMultiLists with disjoint IE sets, let's
> >>>>>> imagine that you have to export an aggregated observation point,
> >>>>>> composed of multiple template records
> >>>>>>       template 1: exporterIPaddress
> >>>>>>       template 2: exporterIPaddress, basicList of interfaces
> >>>>>>       template 3: exporterIPaddress, LC
> >>>>>>              
> >>>>> and then you'd want to OR these... Okay, makes sense...
> >>>>>
> >>>>>            
> >>>>>>> 3. If we really _do_ want ranges and so on (which, again, we'd
> >>>>>>> need to get WG consensus on; this is explicitly out of scope in my
> >>>>>>> reading of the present charter), then we could do them in the
> >>>>>>> scope of Solution 3.
> >>>>>>>                
> >>>>>> Btw, acknowledging that one day we will have to solve this is good
> >>>>>> enough for solution 3. I mean, we don't have to populate the
> >>>>>> semantic IANA now... even if that would be more efficient.
> >>>>>>              
> >>>>>>> However, this seems a little not-quite-fleshed-out-enough for me
> >>>>>>> to say whether I like it or not. Could you present an example of
> >>>>>>> how you would use the proposed Semantic field to model your
> >>>>>>> example? "(eth1 OR eth2) AND (NOT (eth3 OR eth4)) OR linecard2"
> >>>>>>>
> >>>>>>> (FWIW, I would do this with my proposal to Solution 2 as follows:)
> >>>>>>>
> >>>>>>> (orBasicList (andBasicList (orBasicList ingressInterface eth1
> >>>>>>> eth2) (notBasicList (orBasicList ingressInterface eth3 eth4))
> >>>>>>> (andBasicList lineCardID 2))
> >>>>>>>
> >>>>>>>
> >>>>>>>                
> >>>>>> (BasicList, OR, (basicList, AND, (basicList, OR, eth1, eth2),
> >>>>>> (basicList, NOR, eth3, eth4)), (basicList, NONE, linecard2)))
> >>>>>>              
> >>>>> Hm. Okay. This makes sense. A couple of questions then, about
> >>>>> solution 3:
> >>>>>
> >>>>> 1. Would we want to try and bitfield this, to define the semantics
> >>>>> we _know_ we have, then leave the rest of it reserved? Something like:
> >>>>>
> >>>>> MSb                      LSb
> >>>>> +---+------+---------------+
> >>>>> | ! | multi| reserved      |
> >>>>> +---+------+---------------+
> >>>>>
> >>>>> ! = negate sense of semantics if 1 (this is the NOT flag)
> >>>>> multi (multiplicity) = 00: undefined, 01: or/oneOrMore, 10:
> >>>>> xor/exactlyOne, 11: and/exactlyAll
> >>>>> reserved = place for adding other bells and whistles like RANGE and
> >>>>> so on in the future.
> >>>>>            
> >>>> Maybe we want to start by asking the question: which semantic do we
> >>>> need now?
> >>>> Is a need for NONE, OR, AND, ORDERED for now. Anything else?
> >>>>          
> >>>>> 2. Are we sure we want to do this in one byte? If we do bitfielding
> >>>>> as above this gives us 32 possible extensions, which seems like
> >>>>> _way_ more than enough, but does stick another odd offset in there,
> >>>>> which slows things down on machines that need aligned access.
> >>>>> Probably one byte is okay and we let the implementation use
> >>>>> paddingOctets and set padding to fix this...
> >>>>>            
> >>>> With structured data, is the alignment still important? When I look
> >>>> at the examples throughout the draft... As you wrote, paddingOctets
> >>>> might be the solution in this case.
> >>>>          
> >>>>> 3. Would it make sense maybe to have two sets of structured data
> >>>>> elements, one with the semantics byte, and one without (which is
> >>>>> then explicitly undefined)? Then exporters who don't need it don't
> >>>>> have to bother sticking an extra odd-aligned zero in the stream for
> >>>>> every list.
> >>>>>
> >>>>> I'm sure I'l have more questions, but none come to mind now... But
> >>>>> it seems like we're converging on the least-unnecessarily-complex
> >>>>> solution here, which is good. :)
> >>>>>            
> >>>> Happy about that. ;-)
> >>>> Note: I was envisioning even something more simpler: one byte,
> >>>> containing all the semantic possibilities, administered by IANA.  So
> >>>> no reserved, no !
> >>>>
> >>>> Regards, Benoit.
> >>>>          
> >>>        
> >> _______________________________________________
> >> IPFIX mailing list
> >> IPFIX@ietf.org
> >> https://www.ietf.org/mailman/listinfo/ipfix
> >>
> >>      
> > ---
> > Atsushi KOBAYASHI<akoba@nttv6.net>
> > NTT Information Sharing Platform Lab.
> > tel:+81-(0)422-59-3978 fax:+81-(0)422-59-5637
> >
> >    
> 

-- 
Atsushi Kobayashi <akoba@nttv6.net>


From akoba@nttv6.net  Sat Mar 20 23:10:23 2010
Return-Path: <akoba@nttv6.net>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 54E493A686D for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 23:10:23 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 4.029
X-Spam-Level: ****
X-Spam-Status: No, score=4.029 tagged_above=-999 required=5 tests=[AWL=-0.101,  BAYES_50=0.001, DNS_FROM_OPENWHOIS=1.13, J_CHICKENPOX_35=0.6, J_CHICKENPOX_37=0.6, J_CHICKENPOX_41=0.6, J_CHICKENPOX_51=0.6, J_CHICKENPOX_91=0.6, NO_RELAYS=-0.001]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id iw263Zjz0n0Q for <ipfix@core3.amsl.com>; Sat, 20 Mar 2010 23:10:21 -0700 (PDT)
Received: from mail.nttv6.net (mail.nttv6.net [IPv6:2001:fa8::25]) by core3.amsl.com (Postfix) with ESMTP id B84023A6845 for <ipfix@ietf.org>; Sat, 20 Mar 2010 23:10:20 -0700 (PDT)
Received: from [192.168.0.134] (mail.nttv6.net [IPv6:2001:fa8::25]) by mail.nttv6.net (8.14.3/8.14.3) with ESMTP id o2L6AWTm064164 for <ipfix@ietf.org>; Sun, 21 Mar 2010 15:10:33 +0900 (JST) (envelope-from akoba@nttv6.net)
Date: Sun, 21 Mar 2010 15:10:32 +0900
From: Atsushi Kobayashi <akoba@nttv6.net>
To: ipfix@ietf.org
Message-Id: <20100321151029.9224.17391CF2@nttv6.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
X-Mailer: Becky! ver. 2.52.03 [ja] (Unregistered)
X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.2.3 (mail.nttv6.net [IPv6:2001:fa8::25]); Sun, 21 Mar 2010 15:10:34 +0900 (JST)
Subject: [IPFIX] What kind of flow record does LSN export?
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sun, 21 Mar 2010 06:10:23 -0000

Hi all,

I am considering what kind of Flow Records LSN(Large Scale NAT) exports.
We had called it CGN(Carrier Grade NAT) at beginning.
http://tools.ietf.org/html/draft-nishitani-cgn-04

Section 7 in RFC5153 mentions the guideline for middlebox. In conclusion,
it recommends that we take side external or internal as Observation
Point, and then each field is distinguished to be changed or not to be
changed by prefix "post".

Generally, a NAT box has been located in a network periphery, in that
case, it is better to monitor internal or external side only.
However, LSN will be located in access side or in core side in carrier
network, the network operators should monitor both sides of network, and
also they needs to keep track Flows on both sides, and there are flows
without passing CGN box. It make more complicated situation.

I consider several Flow Records as follows.

         +----------------+
         |                |
-------->|    LSN Box     |--------->
     OP#1|                |OP#2
         +----------------+ 
     SrcIP#a               SrcIP#b 
     DstIP#d               DstIP#d 

To simplify my question, I omit src/dst port numbers on both side. 

a) Export two Flow Records 
- Flow#A {SrcIP#a, DstIP#d} from OP#1
- Flow#B {SrcIP#b, DstIP#d} from OP#2 
However, operator can not identify that both flows are identical. How to
correlate both Flow Records is not considered yet?


b) Export two Flow Records with flowId
- Flow#A {SrcIP#a, DstIP#d, flowId#b} from OP#1
- Flow#B {SrcIP#b, DstIP#d, flowId#b} from OP#2

We can correlate both Flow Records with flowId. 
However, "flowId" seem be assumed related Flow Records have same Flow
Key. Is it correct to use flowId? And if each entity on cache in
Metering Process is individual, it seem difficult to extract flowId
value.

c) Single Flow Record with "post" IEs
A CGN box originally has a cache for NAT. Thus, it becomes quite simple
to export its entity on cache directly. Several IEs have prefix "post"
IEs, and then we can use them.

- Flow { SrcIP#a, postSrcIP#b, DstIP#d }

But, actually, I can not find "postSourceIPvXAddress" and
"postDestinationIPvXAddress" in RFC5103.  

d) Single Flow Record by using structure data format
Can structured format be applited to it?

- Flow {(basicList, ORDERED, SrcIP#a, SrcIP#b), DstIP#d }

Let me know your thought. Do you have better solution?

Regards,
Atsushi

-- 
Atsushi Kobayashi <akoba@nttv6.net>


From bclaise@cisco.com  Sun Mar 21 16:01:20 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id F422F3A68F7 for <ipfix@core3.amsl.com>; Sun, 21 Mar 2010 16:01:19 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.509
X-Spam-Level: 
X-Spam-Status: No, score=-0.509 tagged_above=-999 required=5 tests=[AWL=-0.899, BAYES_20=-0.74, DNS_FROM_OPENWHOIS=1.13]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id vX9UvbrEJdcb for <ipfix@core3.amsl.com>; Sun, 21 Mar 2010 16:01:19 -0700 (PDT)
Received: from av-tac-bru.cisco.com (odd-brew.cisco.com [144.254.15.119]) by core3.amsl.com (Postfix) with ESMTP id 054FB3A68AE for <ipfix@ietf.org>; Sun, 21 Mar 2010 16:01:18 -0700 (PDT)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2LN1SLc006387; Mon, 22 Mar 2010 00:01:28 +0100 (CET)
Received: from [10.21.72.179] ([10.21.72.179]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2LN1O1N006341; Mon, 22 Mar 2010 00:01:25 +0100 (CET)
Message-ID: <4BA6A547.7050503@cisco.com>
Date: Sun, 21 Mar 2010 16:01:27 -0700
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: Paul Aitken <paitken@cisco.com>
References: <4B9AC2F8.1090504@cisco.com> <4B9B87BD.7020201@net.in.tum.de>	<4BA5584E.3060508@auckland.ac.nz> <4BA5611F.90309@cisco.com>
In-Reply-To: <4BA5611F.90309@cisco.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: Nevil Brownlee <n.brownlee@auckland.ac.nz>, "ipfix@ietf.org" <ipfix@ietf.org>
Subject: Re: [IPFIX] IPFIX errata 2052 to 2059
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sun, 21 Mar 2010 23:01:20 -0000

Nevil,
> Nevil,
>
>> I'm looking through our 'unverified' errata - do we have consensus
>> for Paul's Errata 2052, i.e. change SelectorId to unsigned64?
>
> The consensus seems to be a change to unsigned32.
It's true that unsigned64 would have been better, and Thomas highlighted 
a limitation with the MIB.
So let's go for unsigned32 then.

Regards, Benoit.
>
> Thanks,
> P.
>


From bclaise@cisco.com  Sun Mar 21 17:52:31 2010
Return-Path: <bclaise@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id D2A583A6987 for <ipfix@core3.amsl.com>; Sun, 21 Mar 2010 17:52:31 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.102
X-Spam-Level: 
X-Spam-Status: No, score=-0.102 tagged_above=-999 required=5 tests=[AWL=-1.234, BAYES_50=0.001, DNS_FROM_OPENWHOIS=1.13, HTML_MESSAGE=0.001]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 5rV+eZG1Nh0D for <ipfix@core3.amsl.com>; Sun, 21 Mar 2010 17:52:30 -0700 (PDT)
Received: from av-tac-bru.cisco.com (odd-brew.cisco.com [144.254.15.119]) by core3.amsl.com (Postfix) with ESMTP id 9B6DA3A67BD for <ipfix@ietf.org>; Sun, 21 Mar 2010 17:52:30 -0700 (PDT)
X-TACSUNS: Virus Scanned
Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2M0qkmg008527 for <ipfix@ietf.org>; Mon, 22 Mar 2010 01:52:46 +0100 (CET)
Received: from [10.21.72.179] ([10.21.72.179]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id o2M0qbF6008362 for <ipfix@ietf.org>; Mon, 22 Mar 2010 01:52:38 +0100 (CET)
Message-ID: <4BA6BF63.2090800@cisco.com>
Date: Sun, 21 Mar 2010 17:52:51 -0700
From: Benoit Claise <bclaise@cisco.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: "ipfix@ietf.org" <ipfix@ietf.org>
Content-Type: multipart/alternative; boundary="------------030609020708030602060504"
Subject: [IPFIX] Semantic and structured data - summary
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 22 Mar 2010 00:52:31 -0000

This is a multi-part message in MIME format.
--------------030609020708030602060504
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Dear all,

Thanks for the lively discussion on the mailing list.
Summarizing all the thoughts, below is what we believe the consensus is 
regarding the semantic. This should also help the others to give their views
The authors goal is get the agreement by the end of the WG, so that we 
can create a new draft version. So please express your views, either on 
the mailing list, or during the WG meeting.

- completely ignoring the semantic is not an option
- the solution is for the structured data only, so the semantic of the 
IEs within a structured data
- not change to RFC5101.
- add a semantic field in the 3 list types.

     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|0|               Field ID    |       Element Length            |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Semantic  |             BasicList Content ...                 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|                           ...                                 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+


- the Semantic is a single list, as opposed to a bit vector
- the required semantics are:

	0x00: undefined (default)
	0x01: allOf
	0x02: oneOrMoreOf
	0x03: exactlyOneOf
	0x04: noneOf
	0x05: ordered

     An example of ORDERED is the BGP AS-PATH 10 20 30 40 {50,60}
     (basicList, ORDERED, (basicList, ORDERED, AS10,AS20,AS30,AS40), 
(basicList, exactlyOneOf, AS50, AS60))

     Open issue:  RANGE. Required by Gerhard, but is is in scope for 
this draft?

- an IANA registry is required, to have an extensible solution, exactly 
like for the "MPLS label type" registry.
- no "semanticless" versions of the IPFIX structured data types.
- A few statements are required.
     * "semantic interpretations of recieved Data Records at or beyond 
the CP remain explicitly undefined, unless that data is transmitted 
using this extension with explicit semantic information" (or something 
like it) reduces the scope for confusion.
     * So in Structured Data we should have a statement as well that 
semantically meaningless combinations are legal, and up to the CP to 
interpret.

Example: an aggregated observation point, composed of multiple template 
records
template 1: exporterIPaddress
template 2: exporterIPaddress, basicList of interfaces
     template 3: exporterIPaddress, LC
Those three templates are linked with the semantic "exactlyOneOf"



I hope this summary helps.

Regards, Benoit.





--------------030609020708030602060504
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>

<meta http-equiv="content-type" content="text/html; charset=ISO-8859-1">
</head>
<body bgcolor="#ffffff" text="#000000">
Dear all,<br>
<br>
Thanks for the lively discussion on the mailing list.<br>
Summarizing all the thoughts, below is what we believe the consensus is
regarding the semantic. This should also help the others to give their
views<br>
The authors goal is get the agreement by the end of the WG, so that we
can create a new draft version. So please express your views, either on
the mailing list, or during the WG meeting.<br>
<br>
- completely ignoring the semantic is not an option<br>
- the solution is for the structured data only, so the semantic of the
IEs within a structured data <br>
- not change to RFC5101. <br>
- add a semantic field in the 3 list types.<br>
<div
 style="line-height: 75%; margin-top: 9pt; margin-bottom: 0pt; margin-left: 0.32in; text-indent: -0.32in; text-align: left; direction: ltr; unicode-bidi: embed; vertical-align: baseline;">
<pre>&nbsp;&nbsp;&nbsp; +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|0| &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Field ID &nbsp;&nbsp;&nbsp;|&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Element Length&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp;&nbsp; |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Semantic  | &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;BasicList Content ... &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;...&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

</pre>
</div>
<div
 style="line-height: 75%; margin-top: 9pt; margin-bottom: 0pt; margin-left: 0.32in; text-indent: -0.32in; text-align: left; direction: ltr; unicode-bidi: embed; vertical-align: baseline;"><!--[if ppt]--></div>
- the Semantic is a single list, as opposed to a bit vector<br>
- the required semantics are: <br>
<pre wrap="">	0x00: undefined (default) 
	0x01: allOf
	0x02: oneOrMoreOf
	0x03: exactlyOneOf
	0x04: noneOf
	0x05: ordered
</pre>
&nbsp;&nbsp;&nbsp; An example of ORDERED is the BGP AS-PATH 10 20 30 40 {50,60}<br>
&nbsp;&nbsp;&nbsp; (basicList, ORDERED, (basicList, ORDERED, AS10,AS20,AS30,AS40),
(basicList, exactlyOneOf, AS50, AS60))
<br>
&nbsp;&nbsp; <br>
&nbsp;&nbsp;&nbsp; Open issue:&nbsp; RANGE. Required by Gerhard, but is is in scope for
this draft?&nbsp;&nbsp;&nbsp; <br>
<br>
- an IANA registry is required, to have an extensible solution, exactly
like for the "MPLS label type" registry.
<br>
- no "semanticless" versions of the IPFIX structured data types.<br>
- A few statements are required.<br>
&nbsp;&nbsp;&nbsp; * "semantic interpretations of recieved Data Records at or beyond
the CP
remain explicitly undefined, unless that data is transmitted using this
extension with explicit semantic information" (or something like it)
reduces the scope for confusion.
<br>
&nbsp;&nbsp;&nbsp; * So in Structured Data we should have a statement as well that
semantically meaningless combinations are legal, and up to the CP to
interpret.<br>
<br>
Example: an aggregated observation point, composed of multiple template
records<br>
<span class="moz-txt-citetags">&nbsp;&nbsp;&nbsp; </span>template 1: exporterIPaddress<br>
<span class="moz-txt-citetags">&nbsp;&nbsp;&nbsp; </span>template 2:
exporterIPaddress, basicList of interfaces<span class="moz-txt-citetags"><br>
&nbsp;&nbsp;&nbsp; t</span>emplate 3: exporterIPaddress, LC<br>
Those three templates are linked with the semantic "exactlyOneOf"<br>
<br>
<br>
<br>
I hope this summary helps. <br>
<br>
Regards, Benoit.<br>
<br>
<br>
<br>
<br>
</body>
</html>

--------------030609020708030602060504--

From n.brownlee@auckland.ac.nz  Sun Mar 21 21:40:51 2010
Return-Path: <n.brownlee@auckland.ac.nz>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 057923A69B7 for <ipfix@core3.amsl.com>; Sun, 21 Mar 2010 21:40:51 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.869
X-Spam-Level: 
X-Spam-Status: No, score=-2.869 tagged_above=-999 required=5 tests=[BAYES_50=0.001, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id U0kAlFdTicrk for <ipfix@core3.amsl.com>; Sun, 21 Mar 2010 21:40:49 -0700 (PDT)
Received: from mailhost.auckland.ac.nz (larry.its.auckland.ac.nz [130.216.12.34]) by core3.amsl.com (Postfix) with ESMTP id 1C2D13A6889 for <ipfix@ietf.org>; Sun, 21 Mar 2010 21:40:47 -0700 (PDT)
Received: from localhost (localhost.localdomain [127.0.0.1]) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 77303415137 for <ipfix@ietf.org>; Mon, 22 Mar 2010 17:40:54 +1300 (NZDT)
X-Virus-Scanned: by amavisd-new at mailhost.auckland.ac.nz
Received: from mailhost.auckland.ac.nz ([127.0.0.1]) by localhost (larry.its.auckland.ac.nz [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id jrwYoixD4FxM for <ipfix@ietf.org>; Mon, 22 Mar 2010 17:40:54 +1300 (NZDT)
Received: from [130.129.24.159] (unknown [130.129.24.159]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 08AE5415199 for <ipfix@ietf.org>; Mon, 22 Mar 2010 17:40:51 +1300 (NZDT)
Message-ID: <4BA6F4CD.5060704@auckland.ac.nz>
Date: Sun, 21 Mar 2010 21:40:45 -0700
From: Nevil Brownlee <n.brownlee@auckland.ac.nz>
User-Agent: Thunderbird 2.0.0.24 (X11/20100317)
MIME-Version: 1.0
To: IPFIX list <ipfix@ietf.org>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Subject: [IPFIX] Please email slides for Tuesday IPFIX meeting
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 22 Mar 2010 04:40:51 -0000

To everyone presenting slides in our session:

Please don't forget to email a copy of your slides (preferable as a
.pdf file) to  ipfix-chairs@ietf.org  in time for us to get it
onto the Meeting Materials page in time for the meeting.

Thanks, Nevil

-- 
---------------------------------------------------------------------
  Nevil Brownlee                    Computer Science Department | ITS
  Phone: +64 9 373 7599 x88941             The University of Auckland
  FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand


From trammell@tik.ee.ethz.ch  Mon Mar 22 14:28:16 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 7D3E33A68B6; Mon, 22 Mar 2010 14:28:16 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.004
X-Spam-Level: 
X-Spam-Status: No, score=-5.004 tagged_above=-999 required=5 tests=[AWL=0.465,  BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id klRF5TFgNzp2; Mon, 22 Mar 2010 14:28:15 -0700 (PDT)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id 72B4A3A681B; Mon, 22 Mar 2010 14:28:14 -0700 (PDT)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id 94111D9387; Mon, 22 Mar 2010 22:28:30 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id dPdTv99vOnzs; Mon, 22 Mar 2010 22:28:29 +0100 (MET)
Received: from dhcp-wireless-open-abg-29-170.meeting.ietf.org (unknown [130.129.29.170]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id DC9A5D9381; Mon, 22 Mar 2010 22:28:28 +0100 (MET)
From: Brian Trammell <trammell@tik.ee.ethz.ch>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: quoted-printable
Date: Mon, 22 Mar 2010 14:08:26 -0700
Message-Id: <2880BFB8-2DF6-4E3A-8529-9BA401E5E689@tik.ee.ethz.ch>
To: tsvwg@ietf.org
Mime-Version: 1.0 (Apple Message framework v1077)
X-Mailer: Apple Mail (2.1077)
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: [IPFIX] SCTP Stream Reset and IPFIX Per Stream Export
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 22 Mar 2010 21:28:16 -0000

Greetings, all,

Per my recent comment in the TSVWG meeting, I've looked over =
draft-ietf-tsvwg-sctp-strrst-04 primarily from the point of view of an =
IPFIX implementor, especially with respect to =
draft-ietf-ipfix-export-per-sctp-stream; I'm not really qualified to do =
a more in-depth review (i.e., verifying that sections 3-5 properly =
provide the functionality behind the interface described in section 6).

The primary functionality required for per-stream is described, from the =
API level, in 6.3.4 "Add Outgoing Streams" (and, of course, 6.3.1 =
"Enable/Disable Stream Reset"). These sections are complete for our =
purposes. To the extent that the rest of the document accurately =
describes the protocol-level interactions required to support the =
functionality described therein, I'd say that this document is exactly =
what is needed for IPFIX, and is ready to go.

Best regards,

Brian=

From trammell@tik.ee.ethz.ch  Mon Mar 22 14:28:18 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 1BE533A68E5; Mon, 22 Mar 2010 14:28:18 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.004
X-Spam-Level: 
X-Spam-Status: No, score=-5.004 tagged_above=-999 required=5 tests=[AWL=0.465,  BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id hopmwpJ2fNGN; Mon, 22 Mar 2010 14:28:17 -0700 (PDT)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id 4A7E43A681B; Mon, 22 Mar 2010 14:28:14 -0700 (PDT)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id E2346D9381; Mon, 22 Mar 2010 22:28:30 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id eD2ASr9HQDAA; Mon, 22 Mar 2010 22:28:30 +0100 (MET)
Received: from dhcp-wireless-open-abg-29-170.meeting.ietf.org (unknown [130.129.29.170]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id 0B15BD9382; Mon, 22 Mar 2010 22:28:29 +0100 (MET)
From: Brian Trammell <trammell@tik.ee.ethz.ch>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: quoted-printable
Date: Mon, 22 Mar 2010 14:25:47 -0700
Message-Id: <2FB61AAF-D082-4C1A-A66A-556FCFC954E8@tik.ee.ethz.ch>
To: tsvwg@ietf.org
Mime-Version: 1.0 (Apple Message framework v1077)
X-Mailer: Apple Mail (2.1077)
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: [IPFIX] SCTP Stream Reset and IPFIX Per Stream Export
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 22 Mar 2010 21:28:18 -0000

Greetings, all,

Per my recent comment in the TSVWG meeting, I've looked over =
draft-ietf-tsvwg-sctp-strrst-04 primarily from the point of view of an =
IPFIX implementor, especially with respect to =
draft-ietf-ipfix-export-per-sctp-stream; I'm not really qualified to do =
a more in-depth review (i.e., verifying that sections 3-5 properly =
provide the functionality behind the interface described in section 6).

The primary functionality required for per-stream is described, from the =
API level, in 6.3.4 "Add Outgoing Streams" (and, of course, 6.3.1 =
"Enable/Disable Stream Reset"). These sections are complete for our =
purposes. To the extent that the rest of the document accurately =
describes the protocol-level interactions required to support the =
functionality described therein, I'd say that this document is exactly =
what is needed for IPFIX, and is ready to go.

Best regards,

Brian=

From trammell@tik.ee.ethz.ch  Mon Mar 22 14:54:20 2010
Return-Path: <trammell@tik.ee.ethz.ch>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id B955728C18A for <ipfix@core3.amsl.com>; Mon, 22 Mar 2010 14:54:20 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -4.23
X-Spam-Level: 
X-Spam-Status: No, score=-4.23 tagged_above=-999 required=5 tests=[AWL=-0.620,  BAYES_20=-0.74, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 6RgaaIoKTDNg for <ipfix@core3.amsl.com>; Mon, 22 Mar 2010 14:54:19 -0700 (PDT)
Received: from smtp.ee.ethz.ch (smtp.ee.ethz.ch [129.132.2.219]) by core3.amsl.com (Postfix) with ESMTP id AD79428C1A1 for <ipfix@ietf.org>; Mon, 22 Mar 2010 14:54:19 -0700 (PDT)
Received: from localhost (localhost [127.0.0.1]) by smtp.ee.ethz.ch (Postfix) with ESMTP id 3F959D9382; Mon, 22 Mar 2010 22:54:35 +0100 (MET)
X-Virus-Scanned: by amavisd-new on smtp.ee.ethz.ch
Received: from smtp.ee.ethz.ch ([127.0.0.1]) by localhost (.ee.ethz.ch [127.0.0.1]) (amavisd-new, port 10024) with LMTP id h31Lkun9vHMr; Mon, 22 Mar 2010 22:54:34 +0100 (MET)
Received: from dhcp-wireless-open-abg-29-170.meeting.ietf.org (unknown [130.129.29.170]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) (Authenticated sender: briant) by smtp.ee.ethz.ch (Postfix) with ESMTPSA id 6C0ECD9387; Mon, 22 Mar 2010 22:54:34 +0100 (MET)
From: Brian Trammell <trammell@tik.ee.ethz.ch>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: quoted-printable
Date: Mon, 22 Mar 2010 14:53:07 -0700
Message-Id: <DD230A95-9BA5-4A54-9396-2408AE84F338@tik.ee.ethz.ch>
To: iana@iana.org
Mime-Version: 1.0 (Apple Message framework v1077)
X-Mailer: Apple Mail (2.1077)
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: [IPFIX] IPFIX Assignments (http://www.iana.org/assignments/ipfix/ipfix.xml)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 22 Mar 2010 21:54:20 -0000

Greetings,

The IPFIX assignments registry has one (minor) issue, which I'm =
reasonably certain is my fault, since it came from a bit of RFC 5655 =
that I wrote. The Data Type for IE 262, messageMD5Checksum contains the =
string "octetArray (16 bytes)", which due to the note on the size is not =
a valid data type description string in the informationElementDataTypes =
registry as per RFC 5610.=20

For consistency, I would propose to update this registry entry to have =
Data Type "octetArray", with a note added to the description "This =
Information Element has a fixed length of 16 octets."

This would not be an issue at all if the registry were used only for =
human consumption; however, I discovered it while building an IPFIX =
implementation that bootstraps its information model from the IANA =
registry at runtime. The proposed update (and a future policy of only =
having Data Type values that are valid data type description strings) =
would remove the need to code a special-case for this entry.

Incidentally, I'd like to thank everyone at IANA and within the WG =
involved in bringing the XML registry up; it's made it possible to =
implement completely automated registry updates, which is one less thing =
for implementors to worry about. :)

Best regards,

Brian=

From muenz@net.in.tum.de  Tue Mar 23 06:17:05 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id C0C183A6908 for <ipfix@core3.amsl.com>; Tue, 23 Mar 2010 06:17:05 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 4.98
X-Spam-Level: ****
X-Spam-Status: No, score=4.98 tagged_above=-999 required=5 tests=[BAYES_99=3.5, DNS_FROM_OPENWHOIS=1.13, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id LoxpVTZVh7SZ for <ipfix@core3.amsl.com>; Tue, 23 Mar 2010 06:17:03 -0700 (PDT)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 254133A6868 for <ipfix@ietf.org>; Tue, 23 Mar 2010 06:17:02 -0700 (PDT)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id 3AF9B211033F; Tue, 23 Mar 2010 14:17:20 +0100 (CET)
Message-ID: <4BA8BF7F.2060901@net.in.tum.de>
Date: Tue, 23 Mar 2010 14:17:51 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.24 (Windows/20100228)
MIME-Version: 1.0
To: Andrew Johnson <andrjohn@cisco.com>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com> <4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com> <4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E06B4.6090203@net.in.tum.de> <4B9FFE9C.3060908@cisco.com> <4BA0A7C8.9010205@net.in.tum.de> <4BA0E75E.7060507@cisco.com>
In-Reply-To: <4BA0E75E.7060507@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms050704050304070809010809"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] Review (was Re: WG last call on	draft-ietf-ipfix-configuration-model-04)
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 23 Mar 2010 13:17:05 -0000

This is a cryptographically signed message in MIME format.

--------------ms050704050304070809010809
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable


Hi Andrew,

Andrew Johnson wrote:
> Hi Gerhard
>=20
> One question before I comment inline.  How do you directly attach a
> cache to an observation point?  i.e. no sampling or filtering.

Citation from -05 of the draft:

  "Every Metering Process contains at least one Selection Process and at
   least one Cache.  If all packets in the Observed Packet Stream are
   selected without any Filtering or Sampling, this corresponds to a
   Selection Process implementing a trivial Selector which selects all
   packets."

> Gerhard Muenz wrote:
>> Hi Andrew,
>>
>> Andrew Johnson wrote:
>>> Hi Gerhard
>>>
>>> Gerhard Muenz wrote:
>>> [SNIP]
>>>> To make it clearer, I will add an explicit note that the counters in=
=20
>>>> the Selector class correspond to the counters in the IPFIX MIB.
>>>>
>>>> Just like in the MIB specification, we do not specify how the=20
>>>> counters are related to the PSAMP statistics reports.
>>> I think the current configuration model has a few problems:
>>>    * Implementations seem forced to consider more than one cache part=

>>>      of the same Metering Process to be able to use the cascading
>>>      selection configuration.
>> Not at all!
>>
>> As I have written multiple times here on the list, no implementation i=
s=20
>> expected to support the entire configuration data model. That is just =

>> impossible. Every implementation will have restriction at one end or t=
he=20
>> other. This is not a specific problem to the Selection Process or Cach=
e=20
>> at all.
>>
>> YANG allows you to specify restrictions and extensions in a formal way=
=2E
>>
>> If you do not think that this becomes clear from the text in the draft=
,=20
>> then please provide some better text.
>=20
> You seem to have misunderstood my point.  I will try to clarify it: An
> implementation that treats each cache as a separate Metering Process
> will be forced to use a limited subset of the configuration model.
>=20
> To put this another way, the cascading selector model offers a reasonab=
ly
> concise way to state the configuration and more implementations could u=
se
> this model no attempt was made to map the PSAMP model onto the
> configuration model.

Ok.

Though, I would not call the limited configuration model "the PSAMP model=
".

>>>    * The "SelectionSequenceID" can not be a property of a
>>>      "SelectionProcess Class" because that class does not map to a
>>>      Selection Sequence (i.e. OP, selectors, Metering Process).
>> No. The Selection Process may be composed of multiple Selectors.=20
>> Therefore, the Selection Process represents a Selection Sequence and h=
as=20
>> its Selection Sequence ID as state parameter.
>=20
> Since a "Selection Process" can pass a packet stream on to a second
> "Selection Process", it does not necessarily represent the full
> Selection Sequence and therefore won't necessarily have a Selection
> Sequence ID.
>=20
> In the case where you have:
>=20
>                +-> SP2 -> cache A
>   OP1 -> SP1 --+
>                +-> SP3 -> cache B
>=20
> SP1 is part of two selection sequences.

http://tools.ietf.org/html/draft-ietf-ipfix-configuration-model-05#sectio=
n-7.3

>>>    * Selectors can not be re-used fully across Selection Sequences as=

>>>      they are done in the PSAMP model (so I don't think you should ha=
ve
>>>      a "SelectorID" property either).
>> You have always stressed that every Selection Sequence has its own=20
>> Selector instances with separate selection statistics and that only a =

>> Selector configuration template can be shared.
>>
>> This is reflected in the model!
>>
>> The data model specifies instances, no configuration templates (again,=
=20
>> I'm repeating myself). So, if you need the same Selector config in two=
=20
>> Selection Sequences, you copy the data.
>=20
> Exactly!  Doesn't it strike you as odd that the configuration model
> needs every use of a configuration to be repeated?

No.

Network management is about managed objects which have their own
configuration and state parameters. Two Selection Process represent two
such managed objects which are independent from each other.

>> Just like the Selection Sequence ID, the Selector ID is a state=20
>> parameter assigned by the Monitoring Device. It gives the link to the =

>> information exported via PSAMP protocol. Therefore it is useful to hav=
e=20
>> it in the configuration data model.
>=20
> I'm not convinced any run-time state really belongs in the configuratio=
n
> data model.  Worse, the attempt to make that state available has made i=
t
> really difficult to model simple configurations with configuration mode=
l!

Netconf is a protocol for both monitoring and configuration.
The configuration data model is to cover the monitoring part as well.
Therefore, the MIB parameters have been included.

>> The Monitoring Device may assign the same Selector ID for two Selector=
=20
>> instances if the config is identical. There is no problem. Obviously, =

>> the Monitoring Device needs to compare the configurations of the=20
>> different instances.
>=20
> And you don't think this is a problem?!  Here is one of my assumptions
> about a good configuration model.  It is relatively simple for a Meteri=
ng
> Device to generate it's configuration in the model and then read that
> configuration back and produce an internal representation of that
> configuration that is isomorphic to the original.

The mapping from the configuration data model to the
implementation-dependent representation of the data is most important.
The other direction is secondary.

I do not see a problem because the configuration data model can be
quite easily transformed into the configuration file format of our
IPFIX/PSAMP implementation.

>>> The PSAMP model is to define selectors as simply configuration.  Ther=
e=20
>>> are no statistics on a per Selector basis.  Instead, each use of a=20
>>> selector needs separate statistics.  From RFC 5476, section 6.5.1:
>>>
>>>   Example of two Selection Sequences:
>>>
>>>     Selection Sequence 7 (Filter->Sampling):
>>>       ingressInterface    5
>>>       selectorId          5 (Filter, match IPV4SourceAddress 192.0.2.=
1)
>>>       selectorId         10 (Sampler, Random 1 out-of ten)
>>>
>>>     Selection Sequence 9 (Sampling->Filtering):
>>>       ingressInterface    5
>>>       selectorId         10 (Sampler, Random 1 out-of ten)
>>>       selectorId          5 (Filter, match IPV4SourceAddress 192.0.2.=
1)
>>>
>> What if these are two different Observation Domains?
>=20
> The example is very specifically within the same observation domain.
>=20
>=20
>>    * Selector ID
>>
>>       The Selector ID is the unique ID identifying a Primitive Selecto=
r.
>>       The ID is unique within the Observation Domain.
>>
>> It does not say: unique per PSAMP Device.
>=20
> That doesn't matter if you don't put the selector ID in the configurati=
on
> model.

See arguments above why we want to have state parameters in the model.

>> Configuration templates for Selectors would introduce another level of=
=20
>> indirection to the model. This will complicate things a bit.
>>
>> gIf the Selector ID would be a unique state parameter of the=20
>> configuration template, then you force Selector IDs to be unique per=20
>> Monitoring Device. Or you must disallow reusing Selector configuration=
=20
>> templates across Observation Domains. Formally specifying such=20
>> restrictions in YANG results in complicated XPath expresions (or is no=
t=20
>> possible at all).
>>
>> Though, if you insist on configuring templates for Selectors instead o=
f=20
>> instances, we can change the model accordingly.
>>
>> However, this decision will have no effect on the question whether one=
=20
>> Selection Process can feed multiple Caches.
>=20
> For me it will have an impact.  Or rather, I have no objection to a
> "Selection Process" feeding multiple caches, what I object to is the
> forcing of the sharing of selection state between caches, just because
> they share configuration.

There is no such "forcing". This depends on the implementation.

Some e-mails ago, you told me that you could imagine a Selection Process
which keeps separate counters for all Caches which aggregate the
selected packets.

>>> If we're not going to map the PSAMP model in the configuration model =

>>> then I think we should separate them completely.  In fact, I don't=20
>>> think any attempt should be made to directly configure a "Selection=20
>>> Process", instead the selection configuration is provided and the=20
>>> Exporting Device can sort it out.
>>>
>>> This simply avoids the question about whether a Metering Device can=20
>>> (or should) have more than one cache.  It also allows greater=20
>>> flexibility in configuration because a Selector Configuration can hav=
e=20
>>> more than one Packet Stream as input.
>> The questions whether the Selector configuration is organized with=20
>> reusable configuration templates and whether the same Selection Proces=
s=20
>> may feed two Caches are completely orthogonal.
>>
>> You must not mix up these two questions!
>=20
> They aren't completely orthogonal.  An implementation that considers ea=
ch
> cache a separate Metering Process should not be forced to build an
> excessively complicated configuration model.

I assume that you have one specific implementation in mind, which is
Cisco Flexible NetFlow and its CLI, right?

I know that Cisco decided to introduce configuration templates for
Selectors in the FNF CLI. Yet, this has nothing to do with PSAMP.
There are other implementations which do not use such configuration
templates.

Gerhard

>>> I think this is a useful concept.  Imagine this relatively common=20
>>> configuration, where we want to monitor 10 interfaces, doing 1 in 100=
=20
>>> sampling for each.  Rather than define a new sampling selector for=20
>>> each interface, wouldn't it be better to simply do?:
>>>
>>>    Selector MY_SAMPLER: sampling 1 in 100 -> cache: MY_CACHE
>>>    OP A -> MY_SAMPLER
>>>    OP B -> MY_SAMPLER
>>>    ...
>> What do you mean with "simple"?
>> You think that you do not have to copy some lines of XML?
>> I assume that configuration management tools can do this work for you.=

>>
>>> This is closer to the PSAMP model where you would do something like:
>>>
>>>    Selector MY_SAMPLER: sampling 1 in 100
>>>
>>>    Selection Sequence: OP A, MY_SAMPLER, MY_CACHE
>>>    Selection Sequence: OP B, MY_SAMPLER, MY_CACHE
>>>    ...
>> Do you also request configuration templates for Caches? Or is this a=20
>> single instance?
>=20
> It's the same cache.
>=20
>=20
> Andrew

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms050704050304070809010809
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms050704050304070809010809--

From muenz@net.in.tum.de  Tue Mar 23 06:48:43 2010
Return-Path: <muenz@net.in.tum.de>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 62E913A6B3B for <ipfix@core3.amsl.com>; Tue, 23 Mar 2010 06:48:43 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 3.231
X-Spam-Level: ***
X-Spam-Status: No, score=3.231 tagged_above=-999 required=5 tests=[AWL=1.750,  BAYES_50=0.001, DNS_FROM_OPENWHOIS=1.13, HELO_EQ_DE=0.35]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id PT8K8rYtT8Nf for <ipfix@core3.amsl.com>; Tue, 23 Mar 2010 06:48:42 -0700 (PDT)
Received: from mail-out1.informatik.tu-muenchen.de (mail-out1.informatik.tu-muenchen.de [131.159.0.8]) by core3.amsl.com (Postfix) with ESMTP id 44CB73A6B14 for <ipfix@ietf.org>; Tue, 23 Mar 2010 06:48:39 -0700 (PDT)
Received: from [131.159.20.108] (repulse.net.in.tum.de [131.159.20.108]) by mail.net.in.tum.de (Postfix) with ESMTPSA id DF9C2211B627; Tue, 23 Mar 2010 14:48:56 +0100 (CET)
Message-ID: <4BA8C6E8.9050709@net.in.tum.de>
Date: Tue, 23 Mar 2010 14:49:28 +0100
From: Gerhard Muenz <muenz@net.in.tum.de>
User-Agent: Thunderbird 2.0.0.24 (Windows/20100228)
MIME-Version: 1.0
To: Andrew Johnson <andrjohn@cisco.com>
References: <C7A25C73.756E4%Quittek@nw.neclab.eu> <4B7EB360.8060902@cisco.com> <4B8BD178.8030802@net.in.tum.de> <4B918279.4040609@cisco.com> <4B9225A5.5030709@net.in.tum.de> <4B92BA5F.1060601@cisco.com> <4B98C5FC.5030909@net.in.tum.de> <4B9918A6.6010407@cisco.com> <4B9E06B4.6090203@net.in.tum.de> <4B9FFE9C.3060908@cisco.com> <4BA0A7C8.9010205@net.in.tum.de> <4BA0F1C3.5070808@cisco.com>
In-Reply-To: <4BA0F1C3.5070808@cisco.com>
Content-Type: multipart/signed; protocol="application/x-pkcs7-signature"; micalg=sha1; boundary="------------ms080208070008000101070300"
X-Virus-Scanned: ClamAV using ClamSMTP
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] draft-ietf-ipfix-configuration-model by example: Examples 1 - 2
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 23 Mar 2010 13:48:43 -0000

This is a cryptographically signed message in MIME format.

--------------ms080208070008000101070300
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable


Hi Andrew,

Andrew Johnson wrote:
> Hi Gerhard
>=20
> Perhaps we can come to a better understanding if you help me
> see how my implementation would use the configuration data
> model.  I was originally very optimistic about the configuration
> draft because many of the concepts seemed to map very neatly to
> how my implementation models its config via the command line.
>=20
> Would you be willing to map out some real-world configurations
> into the configuration data model?  It doesn't have to be the
> full YANG output, but I think some text showing the basic
> structure would be very useful.

There will never be YANG output. Just XML ;)

> Example 1:
>=20
>   flow record MY_RECORD
>     match ipv4 source address
>     match ipv4 destination address
>     collect counter bytes
>=20
>   flow monitor MY_MON
>     record MY_RECORD
>=20
>   interface ethernet 0/0
>     ip flow monitor MY_MON input
>=20
>   interface ethernet 0/1
>     ip flow monitor MY_MON input
>=20
>=20
> Typically a customer would configure a monitor on many more
> interfaces, perhaps even using a virtual-template or interface
> group to add the monitor to hundreds of interfaces, but for now
> lets look at just two.
>=20
> The above configuration directly applies a single monitor
> to two interfaces.  In our feature each monitor would be
> considered a separate Metering Process, but we don't have to
> worry about that here.

This is similar to the example in the draft:
http://tools.ietf.org/html/draft-ietf-ipfix-configuration-model-05#sectio=
n-7.2

Is it one observation domain? Or two?
The model allows to configure the Observation Domain ID.
Implementations which do not allow this can restrict this parameter.

Please also see the note below the XML part.

<ipfix xmlns=3D"urn:ietf:params:xml:ns:ietf-ipfix-psamp">

  <observationPoint>
    <name>OP at eth0</name>
    <observationDomainId>123</observationDomainId>
    <ifName>eth0</ifName>
    <selectionProcess>All packet selection 1</selectionProcess>
  </observationPoint>

  <observationPoint>
    <name>OP at eth1</name>
    <observationDomainId>123</observationDomainId>
    <ifName>eth1</ifName>
    <selectionProcess>All packet selection 2</selectionProcess>
  </observationPoint>

  <selectionProcess>
    <name>All packet selection 1</name>
    <selector>
      <name>Select all</name>
      <selectAll/>
    </selector>
    <cache>Flow cache</cache>
  </selectionProcess>

  <selectionProcess>
    <name>All packet selection 2</name>
    <selector>
      <name>Select all</name>
      <selectAll/>
    </selector>
    <cache>Flow cache</cache>
  </selectionProcess>

  <cache>
    <name>Flow cache</name>
    <cacheMode>timeout</cacheMode>
    <maxRecords>4096</maxRecords>
    <activeTimeout>5000</activeTimeout>
    <inactiveTimeout>10000</inactiveTimeout>
    <cacheLayout>
      <cacheField>
        <name>Field 1</name>
        <ieName>sourceIPv4Address</ieName>
        <isFlowKey/>
      </cacheField>
      <cacheField>
        <name>Field 2</name>
        <ieName>destinationIPv4Address</ieName>
        <isFlowKey/>
      </cacheField>
      <cacheField>
        <name>Field 8</name>
        <ieName>octetDeltaCount</ieName>
      </cacheField>
    </cacheLayout>
    <exportingProcess>SCTP export with UDP backup</exportingProcess>
  </cache>

  <exportingProcess>
    <name>SCTP export with UDP backup</name>
    ...
  </exportingProcess>

</ipfix>


Note:
I had the idea that we could group multiple interfaces which belong the
same observation domain in one observation point (this possibility is
mentioned in RFC5101):

  <observationPoint>
    <name>OP at eth0 and eth1</name>
    <observationDomainId>123</observationDomainId>
    <ifName>eth0</ifName>
    <ifName>eth1</ifName>
    <selectionProcess>All packet selection 1</selectionProcess>
  </observationPoint>

However, if <direction> was configured, it would affect all interfaces:

  <observationPoint>
    <name>OP at eth0 and eth1</name>
    <observationDomainId>123</observationDomainId>
    <ifName>eth0</ifName>
    <ifName>eth1</ifName>
    <direction>ingress</direction>
    <selectionProcess>All packet selection 1</selectionProcess>
  </observationPoint>

If this makes sense, we can allow multiple occurrences of
ifIndex/ifName/entPhysicalIndex/entPhysicalName per Observation Point.


> Example 2:
>=20
>   sampler MY_SAMPLER
>     mode random 1 out-of 100
>=20
>   flow record MY_RECORD
>     match ipv4 source address
>     match ipv4 destination address
>     collect counter bytes
>=20
>   flow monitor MY_MON
>     record MY_RECORD
>=20
>   interface ethernet 0/0
>     ip flow monitor MY_MON sampler MY_SAMPLER input
>=20
>   interface ethernet 0/1
>     ip flow monitor MY_MON sampler MY_SAMPLER input
>=20
>=20
> Very similar, except that a we have added a sampling selector.

I assume that this configuration results in two independent Selection
Process instances in your system. Otherwise, the selection of packets
from the two interfaces would not be independent from each other.

Then, the configuration is the same as above, but you change the
configuration parameters of the Selection Processes:

      <selector>
         <name>1-out-of-100 sampler</name>
         <sampRandOutOfN>
           <size>1</size>
           <population>100</population>
         </sampRandOutOfN>
       </selector>

> As I think about these common configurations, what strikes
> me as very typical is a reasonably simple configuration that
> is then re-used over many interfaces.  It seems to me that
> in the current configuration data model that leads to a lot
> of repetition.

As you have two independent Selection Process instancees, you have the
same repetition in the running system. So, the configuration reflects
the instances in the system.

Regards,
Gerhard

> Cheers
>=20
> Andrew

--=20
Dipl.-Ing. Gerhard M=FCnz
Chair for Network Architectures and Services (I8)
Technische Universit=E4t M=FCnchen - Department of Informatics
Boltzmannstr. 3, 85748 Garching bei M=FCnchen, Germany
Phone:  +49 89 289-18008       Fax: +49 89 289-18033
E-mail: muenz@net.in.tum.de    WWW: http://www.net.in.tum.de/~muenz



--------------ms080208070008000101070300
Content-Type: application/x-pkcs7-signature; name="smime.p7s"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="smime.p7s"
Content-Description: S/MIME Cryptographic Signature
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--------------ms080208070008000101070300--

From jmpolk@cisco.com  Wed Mar 24 14:02:49 2010
Return-Path: <jmpolk@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 7EC6E3A6DF6; Wed, 24 Mar 2010 14:02:49 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -9.469
X-Spam-Level: 
X-Spam-Status: No, score=-9.469 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Ior7MH3oAHUR; Wed, 24 Mar 2010 14:02:47 -0700 (PDT)
Received: from sj-iport-4.cisco.com (sj-iport-4.cisco.com [171.68.10.86]) by core3.amsl.com (Postfix) with ESMTP id 750593A6DE4; Wed, 24 Mar 2010 14:02:47 -0700 (PDT)
Authentication-Results: sj-iport-4.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AvsEAGoaqkurRN+J/2dsb2JhbACbG3OnMpkUhH4Egx4
X-IronPort-AV: E=Sophos;i="4.51,303,1267401600"; d="scan'208";a="105283107"
Received: from sj-core-3.cisco.com ([171.68.223.137]) by sj-iport-4.cisco.com with ESMTP; 24 Mar 2010 21:03:08 +0000
Received: from jmpolk-wxp01.cisco.com (sjc-vpn2-1061.cisco.com [10.21.116.37]) by sj-core-3.cisco.com (8.13.8/8.14.3) with ESMTP id o2OL38s9011263;  Wed, 24 Mar 2010 21:03:08 GMT
Message-Id: <201003242103.o2OL38s9011263@sj-core-3.cisco.com>
X-Mailer: QUALCOMM Windows Eudora Version 7.1.0.9
Date: Wed, 24 Mar 2010 16:03:07 -0500
To: Brian Trammell <trammell@tik.ee.ethz.ch>, tsvwg@ietf.org
From: "James M. Polk" <jmpolk@cisco.com>
In-Reply-To: <2880BFB8-2DF6-4E3A-8529-9BA401E5E689@tik.ee.ethz.ch>
References: <2880BFB8-2DF6-4E3A-8529-9BA401E5E689@tik.ee.ethz.ch>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"; format=flowed
X-Mailman-Approved-At: Wed, 24 Mar 2010 14:29:22 -0700
Cc: IETF IPFIX Working Group <ipfix@ietf.org>
Subject: Re: [IPFIX] SCTP Stream Reset and IPFIX Per Stream Export
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 24 Mar 2010 21:02:49 -0000

Brian

Thanks for providing the review for the area(s) you do fully 
understand, and for being honest about what how you qualify your knowledge.

James
(TSVWG chair)

At 04:08 PM 3/22/2010, Brian Trammell wrote:
>Greetings, all,
>
>Per my recent comment in the TSVWG meeting, I've looked over 
>draft-ietf-tsvwg-sctp-strrst-04 primarily from the point of view of 
>an IPFIX implementor, especially with respect to 
>draft-ietf-ipfix-export-per-sctp-stream; I'm not really qualified to 
>do a more in-depth review (i.e., verifying that sections 3-5 
>properly provide the functionality behind the interface described in 
>section 6).
>
>The primary functionality required for per-stream is described, from 
>the API level, in 6.3.4 "Add Outgoing Streams" (and, of course, 
>6.3.1 "Enable/Disable Stream Reset"). These sections are complete 
>for our purposes. To the extent that the rest of the document 
>accurately describes the protocol-level interactions required to 
>support the functionality described therein, I'd say that this 
>document is exactly what is needed for IPFIX, and is ready to go.
>
>Best regards,
>
>Brian


From n.brownlee@auckland.ac.nz  Wed Mar 24 23:11:31 2010
Return-Path: <n.brownlee@auckland.ac.nz>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 7E6DA3A6C5B for <ipfix@core3.amsl.com>; Wed, 24 Mar 2010 23:11:31 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.869
X-Spam-Level: 
X-Spam-Status: No, score=-2.869 tagged_above=-999 required=5 tests=[BAYES_50=0.001, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 1yREtmBXzupY for <ipfix@core3.amsl.com>; Wed, 24 Mar 2010 23:11:30 -0700 (PDT)
Received: from mailhost.auckland.ac.nz (larry.its.auckland.ac.nz [130.216.12.34]) by core3.amsl.com (Postfix) with ESMTP id 9EEE53A6C57 for <ipfix@ietf.org>; Wed, 24 Mar 2010 23:11:28 -0700 (PDT)
Received: from localhost (localhost.localdomain [127.0.0.1]) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 054741AE05 for <ipfix@ietf.org>; Thu, 25 Mar 2010 19:11:37 +1300 (NZDT)
X-Virus-Scanned: by amavisd-new at mailhost.auckland.ac.nz
Received: from mailhost.auckland.ac.nz ([127.0.0.1]) by localhost (larry.its.auckland.ac.nz [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id uCdpNVbDXTUR for <ipfix@ietf.org>; Thu, 25 Mar 2010 19:11:36 +1300 (NZDT)
Received: from [130.129.28.177] (dhcp-wireless-open-abg-28-177.meeting.ietf.org [130.129.28.177]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mailhost.auckland.ac.nz (Postfix) with ESMTP id DFC2B1ADF4 for <ipfix@ietf.org>; Thu, 25 Mar 2010 19:11:33 +1300 (NZDT)
Message-ID: <4BAAFE8C.5050203@auckland.ac.nz>
Date: Wed, 24 Mar 2010 23:11:24 -0700
From: Nevil Brownlee <n.brownlee@auckland.ac.nz>
User-Agent: Thunderbird 2.0.0.24 (X11/20100317)
MIME-Version: 1.0
To: IPFIX list <ipfix@ietf.org>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Subject: [IPFIX] ***DRAFT*** minutes for IPFIX meeting at IETF 77
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 25 Mar 2010 06:11:31 -0000

Hi all:

Here are the draft minutes of our meeting on Tuesday afternoon.
Please send me any corrections!

Cheers, Nevil


IPFIX WG meeting at IETF 77, 1520 on 23 March 2010
Minutes by Nevil Brownlee

Nevil Brownlee reported that of 18 currently-open IPFIX errata, 15 are
editorial.  Two are for the IPFIX File Format RFC, Brian Trammel has
verified those.  The last, 2052, changes the type for the SelectorID
IE.  It was unsigned16; mailing list discussion has reached consensus
for unsigned32.  Nevil will summarise all these errata and ask Dan
(our Area Director) to verify them for the RFC Editor.

Benoit Claise presented the Config Model draft version 05, and
reviewed its changes since 04.  Remaining issues are: consistent names
for Cache parameters (the IPFIX MIB RFC is at AUTH-48, these change
should be made there too), and decide on parameters for Selectors
templates.  When we have a new version, it will be passed back to
IESG.

Benoit presented the SCTP per-stream (Standards Track) draft. It's
remaining issue is "how will a collector determine whether an exporter
supports this extension to IPFIX."  We await consensus on the mailing
list.

Atsushi Kobayashi presented the Mediators drafts.  A new version of
the Problem Statement is ready, it resolves issues that arose in IETF
Last Call, it will now be passed to IESG.  The Mediation Framework has
completed WG Last Call, the authors are considering security issues
concerning certificate handling.  The meeting suggested that it should
be done in the same way as in the File Format RFC.  Once we have a new
version it will be submitted to IESG.

Brian Trammell presented the latest version of the Flow Anonymisation
draft, which has had two revisions since its WG LC.  Benoit and Lothar
agreed to review the current version; after any revisions it will be
ready for IETF Last Call.

Tanja Zseby presented the Flow Selection Techniques draft, pointing out
that it follows on from the PSAMP work, selecting flows rather than
packets.  The authors will add more examples, and respond to issues
raised by reviewers.  When we have a new version it will be ready for
WG Last Call.

Benoit presented the Structured Data draft, which has generated a lot
of list traffic lately.  It has one remaining issue, how to specify
the semantics of a structured object.  Discussion will continue on the
list; the authors will produce a new draft for WG Last Call.

Benoit and Juergen presented the PSAMP MIB draft.  This is the last
of the PSAMP drafts, and has been restructured to become an extension
to the IPFIX MIB.  Discussion will continue on the list.

The meeting continued with presentations of new individual drafts.
Lothar Braun presented his IPFIX DTLS Recommendations draft, he would
like to see this work appear as part of an update to the IPFIX
Implementation Guidelines.

Lothar also presented his draft on compressed IPFIX, describing his
recent work on smart electricity meters.  He would like to hear from
anyone else interested in IPFIX compression.

Shingo Kashima presented his draft on Data Link Monitoring; that
proposes to add a small set of new IPFIX IEs for various link layer
environments.  Juergen Quittek explained that - once consensus on the
proposed IEs has been reached on the list - there are three possible
ways to do this: simply submit a request for them to IANA, make an
Individual Submission to IETF, or wait until the WG can take this up
as a WG charter item.

Benoit presented the current version of the Mediation Protocol draft.
The draft's authors will meet later in IETF to work on it.

Sujay Gupta presented his Reliability Template draft, which proposes
a simplified version of the IPFIX Reliability reporting.  Discussion
of this idea will continue on the list.

Nevil will post a set of proposed new WG Milestones to the list.

The meeting finished at 1720.

-- 
---------------------------------------------------------------------
  Nevil Brownlee                    Computer Science Department | ITS
  Phone: +64 9 373 7599 x88941             The University of Auckland
  FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand


From paitken@cisco.com  Thu Mar 25 11:45:03 2010
Return-Path: <paitken@cisco.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id BD8943A6878 for <ipfix@core3.amsl.com>; Thu, 25 Mar 2010 11:45:03 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.869
X-Spam-Level: 
X-Spam-Status: No, score=-6.869 tagged_above=-999 required=5 tests=[BAYES_50=0.001, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 3dgdTpPSlCNr for <ipfix@core3.amsl.com>; Thu, 25 Mar 2010 11:45:00 -0700 (PDT)
Received: from ams-iport-1.cisco.com (ams-iport-1.cisco.com [144.254.224.140]) by core3.amsl.com (Postfix) with ESMTP id 61C353A6895 for <ipfix@ietf.org>; Thu, 25 Mar 2010 11:44:57 -0700 (PDT)
Authentication-Results: ams-iport-1.cisco.com; dkim=neutral (message not signed) header.i=none
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: Am4BAJNLq0uQ/uCWe2dsb2JhbACbKBUBAQsLJAYWBoghngmZEYJVgigE
X-IronPort-AV: E=Sophos;i="4.51,309,1267401600"; d="scan'208";a="58566815"
Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 25 Mar 2010 18:45:18 +0000
Received: from [10.55.81.180] (dhcp-10-55-81-180.cisco.com [10.55.81.180]) by ams-core-1.cisco.com (8.13.8/8.14.3) with ESMTP id o2PIjIJO020609 for <ipfix@ietf.org>; Thu, 25 Mar 2010 18:45:18 GMT
Message-ID: <4BABAF3E.7050404@cisco.com>
Date: Thu, 25 Mar 2010 18:45:18 +0000
From: Paul Aitken <paitken@cisco.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-GB; rv:1.8.1.22) Gecko/20090605 SeaMonkey/1.1.17 (Ubuntu-1.1.17+nobinonly-0ubuntu0.9.04.1)
MIME-Version: 1.0
To: "ipfix@ietf.org" <ipfix@ietf.org>
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 7bit
Subject: [IPFIX] [Fwd: [Technical Errata Reported] RFC3954 (2096)]
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 25 Mar 2010 18:45:03 -0000

FYI, I've opened a new errata on RFC 3954:


The following errata report has been submitted for RFC3954,
"Cisco Systems NetFlow Services Export Version 9".

--------------------------------------
You may review the report below and at:
http://www.rfc-editor.org/errata_search.php?rfc=3954&eid=2096

--------------------------------------
Type: Technical
Reported by: Paul Aitken <paitken@cisco.com>

Section: 5.3 and 6.2.

Original Text
-------------
    Padding

          The Exporter SHOULD insert some padding bytes so that the
          subsequent FlowSet starts at a 4-byte aligned boundary.  It is
          important to note that the Length field includes the padding
          bytes.  Padding SHOULD be using zeros.

Corrected Text
--------------
    Padding

          The Exporter SHOULD insert some padding bytes so that the
          subsequent FlowSet starts at a 4-byte aligned boundary.  It is
          important to note that the Length field includes the padding
          bytes.  The padding length MUST be shorter than any allowable
          record in the Set.  Padding SHOULD be using zeros.


Notes
-----
Addition of "The padding length MUST be shorter than any allowable
record in the Set."

With small field sizes, such that the record size <= 3, it's not
possible to distinguish padding from further data records (s 5.3) or
options data records (s 6.2).

eg, with a record length of 3, three records will consume 9 octets.
Three octets of padding will be added to this, giving a total length of
12 octets. The 12 octets now look like *four* records. In this case,
padding is NOT appropriate.

NB1 the same paragraph in section 6.1 is NOT affected, because the fixed
size of the other fields dictates that the only possibility is padding
of 2 octets.

NB2 this situation is anticipated in IPFIX (RFC 5101), from which the
additional text is taken.


Instructions:
-------------
This errata is currently posted as "Reported". If necessary, please
use "Reply All" to discuss whether it should be verified or
rejected. When a decision is reached, the verifying party (ISE &
Editorial Board)
can log in to change the status and edit the report, if necessary.

--------------------------------------
RFC3954 (draft-claise-netflow-9-08)
--------------------------------------
Title               : Cisco Systems NetFlow Services Export Version 9
Publication Date    : October 2004
Author(s)           : B. Claise, Ed.
Category            : INFORMATIONAL
Source              : INDEPENDENT
Area                : N/A
Stream              : INDEPENDENT
Verifying Party     : ISE & Editorial Board

-- 
Paul Aitken
Cisco Systems Ltd, Edinburgh, Scotland.
http://www.cisco.com/web/about/doing_business/legal/cri/index.html


From gfortaine@live.com  Fri Mar 26 10:48:55 2010
Return-Path: <gfortaine@live.com>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 150B13A6C0E for <ipfix@core3.amsl.com>; Fri, 26 Mar 2010 10:48:55 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -98.066
X-Spam-Level: 
X-Spam-Status: No, score=-98.066 tagged_above=-999 required=5 tests=[BAYES_50=0.001, DNS_FROM_OPENWHOIS=1.13, MSGID_FROM_MTA_HEADER=0.803, USER_IN_WHITELIST=-100]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Sjn5FMlVfPce for <ipfix@core3.amsl.com>; Fri, 26 Mar 2010 10:48:53 -0700 (PDT)
Received: from blu0-omc2-s35.blu0.hotmail.com (blu0-omc2-s35.blu0.hotmail.com [65.55.111.110]) by core3.amsl.com (Postfix) with ESMTP id 6D1913A6BF9 for <ipfix@ietf.org>; Fri, 26 Mar 2010 10:48:53 -0700 (PDT)
Received: from BLU0-SMTP52 ([65.55.111.73]) by blu0-omc2-s35.blu0.hotmail.com with Microsoft SMTPSVC(6.0.3790.3959);  Fri, 26 Mar 2010 10:49:16 -0700
X-Originating-IP: [83.198.249.206]
X-Originating-Email: [gfortaine@live.com]
Message-ID: <BLU0-SMTP52C57C68144C44FE184194C8230@phx.gbl>
Received: from localhost.localdomain ([83.198.249.206]) by BLU0-SMTP52.blu0.hotmail.com over TLS secured channel with Microsoft SMTPSVC(6.0.3790.3959); Fri, 26 Mar 2010 10:49:16 -0700
Date: Fri, 26 Mar 2010 18:49:11 +0100
From: Guillaume FORTAINE <gfortaine@live.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1.8) Gecko/20100301 Fedora/3.0.3-1.fc12 Thunderbird/3.0.3
MIME-Version: 1.0
To: ipfix@ietf.org
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-OriginalArrivalTime: 26 Mar 2010 17:49:16.0517 (UTC) FILETIME=[A7649950:01CACD0C]
Subject: [IPFIX] Problems with Flow Collection in large-scale networks - Flowmon Probe : Hardware-Accelerated IPFIX Autonomous 10 GbE probe
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 26 Mar 2010 17:48:55 -0000

Misses, Misters,

Let me introduce myself : Guillaume FORTAINE, Engineer in Computer
Science. Me and my partners, INVEA-TECH [0], are currently working on
High-Speed Networking Solutions.

Our flagship product is Flowmon probe that is a Hardware Accelerated
IPFIX Autonomous 10 GbE probe [1] :

http://www.cert.org/flocon/2009/presentations/Celeda_FlexibleFlow.pdf

The advantage of autonomous NetFlow probe is the possibility to deploy
NetFlow measurement where a dedicated source of NetFlow data is
missing. Autonomous NetFlow probes allows user to set any parameter
arbitrarily without fear that the bypassing traffic will be
influenced. On the other hand when enabling NetFlow on routers one
have to always be careful about the additional performance demands on
hardware resources by monitoring process (please see the following
article from Cisco entitled "NetFlow Performance Analysis" [2]).
Probes are usually more flexible and allow to test new features such
as different types of sampling, filtering or anonymization.

For you to have a better overview of our company :

We are young and progressive European IT company based on the results
of several EU R&D projects. We have world unique products for
wire-speed packet processing (100 Gigabit Ethernet networks Ready).

We look forward to your answer,

Best Regards,

[0] http://www.invea-tech.com
[1] http://www.invea-tech.com/products/flowmon-probes
[2] 
http://www.cisco.com/en/US/technologies/tk543/tk812/technologies_white_paper0900aecd802a0eb9.html

Guillaume FORTAINE
Tel : +33(0)631092519

From root@core3.amsl.com  Sat Mar 27 06:15:01 2010
Return-Path: <root@core3.amsl.com>
X-Original-To: ipfix@ietf.org
Delivered-To: ipfix@core3.amsl.com
Received: by core3.amsl.com (Postfix, from userid 0) id DF5723A69BB; Sat, 27 Mar 2010 06:15:01 -0700 (PDT)
From: Internet-Drafts@ietf.org
To: i-d-announce@ietf.org
Content-Type: Multipart/Mixed; Boundary="NextPart"
Mime-Version: 1.0
Message-Id: <20100327131501.DF5723A69BB@core3.amsl.com>
Date: Sat, 27 Mar 2010 06:15:01 -0700 (PDT)
Cc: ipfix@ietf.org
Subject: [IPFIX] I-D Action:draft-ietf-ipfix-mediators-problem-statement-09.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Sat, 27 Mar 2010 13:15:02 -0000

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts directories.
This draft is a work item of the IP Flow Information Export Working Group of the IETF.


	Title           : IPFIX Mediation: Problem Statement
	Author(s)       : A. Kobayashi, et al.
	Filename        : draft-ietf-ipfix-mediators-problem-statement-09.txt
	Pages           : 30
	Date            : 2010-03-27

Flow-based measurement is a popular method for various network
monitoring usages.  The sharing of flow-based information for
monitoring applications having different requirements raises some
open issues in terms of measurement system scalability, flow-based
measurement flexibility, and export reliability that IPFIX Mediation
may help resolve.  This document describes some problems related to
flow-based measurement that network administrators have been facing,
and then describes IPFIX Mediation applicability examples along with
the problems.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-ipfix-mediators-problem-statement-09.txt

Internet-Drafts are also available by anonymous FTP at:
ftp://ftp.ietf.org/internet-drafts/

Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

--NextPart
Content-Type: Message/External-body;
	name="draft-ietf-ipfix-mediators-problem-statement-09.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID: <2010-03-27060049.I-D@ietf.org>


--NextPart--

From n.brownlee@auckland.ac.nz  Tue Mar 30 13:01:10 2010
Return-Path: <n.brownlee@auckland.ac.nz>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 2C8F53A6877 for <ipfix@core3.amsl.com>; Tue, 30 Mar 2010 13:01:10 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -4.169
X-Spam-Level: 
X-Spam-Status: No, score=-4.169 tagged_above=-999 required=5 tests=[AWL=1.300,  BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id or8wrIYgty1P for <ipfix@core3.amsl.com>; Tue, 30 Mar 2010 13:01:09 -0700 (PDT)
Received: from mailhost.auckland.ac.nz (larry.its.auckland.ac.nz [130.216.12.34]) by core3.amsl.com (Postfix) with ESMTP id C6AC53A696D for <ipfix@ietf.org>; Tue, 30 Mar 2010 13:00:56 -0700 (PDT)
Received: from localhost (localhost.localdomain [127.0.0.1]) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 01A861B163 for <ipfix@ietf.org>; Wed, 31 Mar 2010 09:01:20 +1300 (NZDT)
X-Virus-Scanned: by amavisd-new at mailhost.auckland.ac.nz
Received: from mailhost.auckland.ac.nz ([127.0.0.1]) by localhost (larry.its.auckland.ac.nz [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 30h+-hdztaeQ for <ipfix@ietf.org>; Wed, 31 Mar 2010 09:01:19 +1300 (NZDT)
Received: from [192.172.226.54] (dyn54.caida.org [192.172.226.54]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mailhost.auckland.ac.nz (Postfix) with ESMTP id 794811B15C for <ipfix@ietf.org>; Wed, 31 Mar 2010 09:01:19 +1300 (NZDT)
Message-ID: <4BB2588C.2010201@auckland.ac.nz>
Date: Tue, 30 Mar 2010 13:01:16 -0700
From: Nevil Brownlee <n.brownlee@auckland.ac.nz>
User-Agent: Thunderbird 2.0.0.24 (X11/20100317)
MIME-Version: 1.0
To: IPFIX list <ipfix@ietf.org>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Subject: [IPFIX] Proposed Milestones update
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 30 Mar 2010 20:01:10 -0000

Hi all:

Here's my proposed update to the IPFIX Milestones:

Done		Submit Mediation Problem Statement I-D to IESG for 				publication as 
Informational RFC
Done		Submit initial draft on anonymization support
Done 		Submit initial draft on flow selection
Done 		Submit initial draft on structuring information elements

May 2010 	Submit Mediation Framework I-D to IESG for publication
		as Informational RFC
May 2010	Submit final version of PSAMP MIB module
May 2010 	Submit Configuration Data Model draft to IESG for
	 	publication as Standards track RFC
May 2010 	Submit anonymization support I-D to IESG for publication
		as Experimental RFC
Sep 2010 	Submit flow selection I-D to IESG for publication
		as Standards Track RFC
Sep 2010 	Submit structuring information elements I-D to IESG
		for publication as Standards Track RFC

Unless anyone objects, I'll get Dan to approve these so that the
Secretariat can update our Charter page.

Cheers, Nevil

-- 
---------------------------------------------------------------------
  Nevil Brownlee                    Computer Science Department | ITS
  Phone: +64 9 373 7599 x88941             The University of Auckland
  FAX: +64 9 373 7453   Private Bag 92019, Auckland 1142, New Zealand


From root@core3.amsl.com  Tue Mar 30 13:45:01 2010
Return-Path: <root@core3.amsl.com>
X-Original-To: ipfix@ietf.org
Delivered-To: ipfix@core3.amsl.com
Received: by core3.amsl.com (Postfix, from userid 0) id C82DC3A6A02; Tue, 30 Mar 2010 13:45:01 -0700 (PDT)
From: Internet-Drafts@ietf.org
To: i-d-announce@ietf.org
Content-Type: Multipart/Mixed; Boundary="NextPart"
Mime-Version: 1.0
Message-Id: <20100330204501.C82DC3A6A02@core3.amsl.com>
Date: Tue, 30 Mar 2010 13:45:01 -0700 (PDT)
Cc: ipfix@ietf.org
Subject: [IPFIX] I-D ACTION:draft-ietf-ipfix-export-per-sctp-stream-07.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 30 Mar 2010 20:45:01 -0000

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts 
directories.
This draft is a work item of the IP Flow Information Export Working Group of the IETF.

	Title		: IPFIX Export per SCTP Stream
	Author(s)	: B. Claise, P. Aitken, A. Johnson, G. Muenz
	Filename	: draft-ietf-ipfix-export-per-sctp-stream-07.txt
	Pages		: 23
	Date		: 2010-3-25
	
This document specifies an extension to the specifications 
in RFC5101, IP Flow Information Export (IPFIX), when using 
the Partial Reliability extension of SCTP (PR-SCTP, Partial 
Reliability Stream Control Transmission Protocol). 
When implemented at both the Exporting and Collecting Processes, 
this method offers several advantages such as the ability to 
calculate Data Record losses for PR-SCTP, immediate export of 
Template Withdrawal Messages, immediate reuse of Template IDs 
within an SCTP stream, reduced likelihood of Data Record loss, 
and reduced demands on the Collecting Process.  When implemented 
in only the Collecting or Exporting Process then normal IPFIX 
behavior will be seen without these additional benefits.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-ipfix-export-per-sctp-stream-07.txt

Internet-Drafts are also available by anonymous FTP at:
ftp://ftp.ietf.org/internet-drafts/

Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

--NextPart
Content-Type: Message/External-body;
	name="draft-ietf-ipfix-export-per-sctp-stream-07.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID: <2010-3-30134418.I-D@ietf.org>


--NextPart--


From akoba@nttv6.net  Tue Mar 30 18:22:03 2010
Return-Path: <akoba@nttv6.net>
X-Original-To: ipfix@core3.amsl.com
Delivered-To: ipfix@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 1885B3A68EC for <ipfix@core3.amsl.com>; Tue, 30 Mar 2010 18:22:03 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.058
X-Spam-Level: 
X-Spam-Status: No, score=-1.058 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, DNS_FROM_OPENWHOIS=1.13, HOST_MISMATCH_COM=0.311, RDNS_DYNAMIC=0.1]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id ZQtHv2uo4Ps7 for <ipfix@core3.amsl.com>; Tue, 30 Mar 2010 18:22:02 -0700 (PDT)
Received: from mail.nttv6.net (mail.nttv6.net [IPv6:2001:fa8::25]) by core3.amsl.com (Postfix) with ESMTP id 5AFBA3A6C58 for <ipfix@ietf.org>; Tue, 30 Mar 2010 18:21:27 -0700 (PDT)
Received: from [127.0.0.1] (dhcp-3-169.nttv6.com [192.47.163.169]) by mail.nttv6.net (8.14.3/8.14.3) with ESMTP id o2V1LpBF037970; Wed, 31 Mar 2010 10:21:52 +0900 (JST) (envelope-from akoba@nttv6.net)
Date: Wed, 31 Mar 2010 10:21:51 +0900
From: Atsushi Kobayashi <akoba@nttv6.net>
To: ipfix@ietf.org
In-Reply-To: <20100327131501.DF5723A69BB@core3.amsl.com>
References: <20100327131501.DF5723A69BB@core3.amsl.com>
Message-Id: <20100331102135.47B8.17391CF2@nttv6.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
X-Mailer: Becky! ver. 2.52.03 [ja] (Unregistered)
X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.2.3 (mail.nttv6.net [192.16.178.5]); Wed, 31 Mar 2010 10:21:52 +0900 (JST)
Cc: Yaron Sheffer <yaronf.ietf@gmail.com>
Subject: Re: [IPFIX] I-D Action:draft-ietf-ipfix-mediators-problem-statement-09.txt
X-BeenThere: ipfix@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: IPFIX WG discussion list <ipfix.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/ipfix>
List-Post: <mailto:ipfix@ietf.org>
List-Help: <mailto:ipfix-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/ipfix>, <mailto:ipfix-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 31 Mar 2010 01:22:03 -0000

Hi all,

As I mentioned in Anaheim meeting, this new version document covers all
issues raised in IETF Last Call. The security consideration was improved
thanks to valuable feedbacks from Yaron. Thank you, Yaron Sheffer.
And now, go to next step IESG review.

Regards,
Atsushi

On Sat, 27 Mar 2010 06:15:01 -0700 (PDT)
Internet-Drafts@ietf.org wrote:

> A New Internet-Draft is available from the on-line Internet-Drafts directories.
> This draft is a work item of the IP Flow Information Export Working Group of the IETF.
> 
> 
> 	Title           : IPFIX Mediation: Problem Statement
> 	Author(s)       : A. Kobayashi, et al.
> 	Filename        : draft-ietf-ipfix-mediators-problem-statement-09.txt
> 	Pages           : 30
> 	Date            : 2010-03-27
> 
> Flow-based measurement is a popular method for various network
> monitoring usages.  The sharing of flow-based information for
> monitoring applications having different requirements raises some
> open issues in terms of measurement system scalability, flow-based
> measurement flexibility, and export reliability that IPFIX Mediation
> may help resolve.  This document describes some problems related to
> flow-based measurement that network administrators have been facing,
> and then describes IPFIX Mediation applicability examples along with
> the problems.
> 
> A URL for this Internet-Draft is:
> http://www.ietf.org/internet-drafts/draft-ietf-ipfix-mediators-problem-statement-09.txt
> 
> Internet-Drafts are also available by anonymous FTP at:
> ftp://ftp.ietf.org/internet-drafts/
> 
> Below is the data which will enable a MIME compliant mail reader
> implementation to automatically retrieve the ASCII version of the
> Internet-Draft.

-- 
Atsushi Kobayashi <akoba@nttv6.net>


