
From sabine.randriamasy@alcatel-lucent.com  Wed Aug  1 02:13:15 2012
Return-Path: <sabine.randriamasy@alcatel-lucent.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 1CA3E21F8615 for <alto@ietfa.amsl.com>; Wed,  1 Aug 2012 02:13:15 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -9.756
X-Spam-Level: 
X-Spam-Status: No, score=-9.756 tagged_above=-999 required=5 tests=[AWL=0.493,  BAYES_00=-2.599, HELO_EQ_FR=0.35, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 2WXnx8sxwsoE for <alto@ietfa.amsl.com>; Wed,  1 Aug 2012 02:13:14 -0700 (PDT)
Received: from smail5.alcatel.fr (smail5.alcatel.fr [64.208.49.27]) by ietfa.amsl.com (Postfix) with ESMTP id 85F7421F85D5 for <alto@ietf.org>; Wed,  1 Aug 2012 02:13:12 -0700 (PDT)
Received: from FRMRSSXCHHUB01.dc-m.alcatel-lucent.com (FRMRSSXCHHUB01.dc-m.alcatel-lucent.com [135.120.45.61]) by smail5.alcatel.fr (8.14.3/8.14.3/ICT) with ESMTP id q719D3PF005390 (version=TLSv1/SSLv3 cipher=RC4-MD5 bits=128 verify=NOT); Wed, 1 Aug 2012 11:13:07 +0200
Received: from [172.27.204.117] (135.120.57.7) by FRMRSSXCHHUB01.dc-m.alcatel-lucent.com (135.120.45.61) with Microsoft SMTP Server (TLS) id 8.3.213.0; Wed, 1 Aug 2012 11:13:05 +0200
Message-ID: <5018F31C.60302@alcatel-lucent.com>
Date: Wed, 1 Aug 2012 11:13:00 +0200
From: Sabine Randriamasy <Sabine.Randriamasy@alcatel-lucent.com>
User-Agent: Thunderbird 2.0.0.19 (Windows/20081209)
MIME-Version: 1.0
To: Leeyoung <leeyoung@huawei.com>
References: <7AEB3D6833318045B4AE71C2C87E8E1720CC2D35@dfweml511-mbx.china.huawei.com> <5016C0DD.8050909@alcatel-lucent.com> <7AEB3D6833318045B4AE71C2C87E8E172905E4F5@dfweml511-mbs.china.huawei.com> <50181E51.9020102@alcatel-lucent.com> <7AEB3D6833318045B4AE71C2C87E8E172905E833@dfweml511-mbs.china.huawei.com>
In-Reply-To: <7AEB3D6833318045B4AE71C2C87E8E172905E833@dfweml511-mbs.china.huawei.com>
Content-Type: text/plain; charset="ISO-8859-1"; format=flowed
Content-Transfer-Encoding: 8bit
X-Scanned-By: MIMEDefang 2.69 on 155.132.188.13
Cc: "alto@ietf.org" <alto@ietf.org>
Subject: Re: [alto] [altoext] FW: New Version Notification	for	draft-lee-alto-app-net-info-exchange-00.txt
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 01 Aug 2012 09:13:15 -0000

Hi Young,

Thanks for your answers. As for the discussions on the cost types used 
in constraints, it will have to be per e-mail as I am not attending IETF 
this time.
Best regards
Sabine


Leeyoung a écrit :
> Hi Sabine,
>
> Thanks again for your further comment. Please see in-line for my response.
>
> Best Regards,
> Young
>
> -----Original Message-----
> From: Sabine Randriamasy [mailto:Sabine.Randriamasy@alcatel-lucent.com] 
> Sent: Tuesday, July 31, 2012 1:05 PM
> To: Leeyoung
> Cc: alto@ietf.org
> Subject: Re: [altoext] FW: New Version Notification for draft-lee-alto-app-net-info-exchange-00.txt
>
>
> Hi Young,
>
> Thanks a lot for your answers. Please see some more comments inline.
> Again with now the ALTO list in Cc
> Best regards
> Sabine
>
> Leeyoung a écrit :
>   
>> Hi Sabine,
>>
>> Here's my response to your comments to your detail questions. Please see in-line.
>> Sreekants or Greg may be able to provide more comments later. 
>>
>> Thanks,
>> Young
>>
>> -----Original Message-----
>> From: Sabine Randriamasy [mailto:Sabine.Randriamasy@alcatel-lucent.com] 
>> Sent: Monday, July 30, 2012 12:14 PM
>> To: Leeyoung
>> Cc: altoext@ietf.org
>> Subject: Re: [altoext] FW: New Version Notification for draft-lee-alto-app-net-info-exchange-00.txt
>>
>> Hi Young,
>>
>> I have read your draft and have a couple of comments and questions.
>>
>> In Section 2 Problem statement, I agree with the need to reduce the 
>> amount of conveyed ALTO information through topology filtering and 
>> constraint-based response filtering. The ALTO base protocol specifies 
>> these features for the Cost Map, Filtered cost map and Endpoint cost 
>> services. I also agree with the necessity to provide several types of 
>> cost information which stresses the necessity of extended filtering 
>> mechanisms. In the Constraints filtering extensions of Section 3, I 
>> didnt' really get how information confidentiality is preserved and the 
>> proposed specification raised the questions detailed below.
>>
>> I also agree that detailing end to end path costs at the link level at 
>> some places can be beneficial, especially for local bottlenecks 
>> sometimes with fast changing cost values. Figure 2 shows an example with 
>> Link capacity costs but section 2 uses them rather as capacity, where I 
>> figure the higher the better. Without a hint on how path costs are 
>> calculated it is hard to follow the rationale on the DC choice by ER1. 
>> See also the questions detailed below.
>>
>> Thanks,
>> Sabine
>>
>> ------------------------- Detailed questions -----------------------------
>>
>> ++ Section 3.1 ALTO Query from Application Stratum to Network Stratum
>> Names used in the list of requested information may be misleading:
>> - A 'Cost Type' see base protocol (§5.1.1) is expected to indicate an 
>> attribute that has a value such as number, boolean, string... and is is 
>> rather "semantic". 'summary' and 'graph' are cost attributes of 
>> different nature.
>>
>> YOUNG>> Thanks. Agreed. We may define a new cost attribute (TBD the name).
>>
>> - 'Constraints' such as 'min/max metric' look like objectives rather 
>> than expressions such as 'lt. value' as specified in the base protocol 
>> (§ 6.8.2.2.3)
>>
>> YOUNG>> In the encoding example we gave in the draft, we used the constraints
>> Convention such as 'lt, gt, eq, ...' Please see Section 3.5.
>>   
>>     
> Ok. Then it may help if the text in §3.1 is reflecting such an 
> expression of constraints.
>
> YOUNG>> Yes we will add explanation text in Section 3.1 on these constraints convention. 
>   
>> - what kind of information is member 'Parameters' is supposed to carry 
>> and for what use?
>>
>> YOUNG>> We put it as a place holder for indicating values of the resulting graph or link
>> representation of network abstraction, but in the actually encoding example, we simply used
>> what is defined in DstCostsConstraints Object (such as pktloss, etc.) to indicate the vaules
>> associated with the S-D path. We can delete it if this is not necessary. 
>>   
>>     
> Or maybe clarify the definition.
>
> YOUNG>> Yes. Will provide. 
>   
>> - Some members listed here do not appear in the ALTO Query Information 
>> Model of §3.3: 'parameters' 'objective-function'. Neither do they appear 
>> in the example of §3.5
>>
>> YOUNG>> Good observation. We put them as a place holder for future enhancement. 
>>
>> ++ Section 3.2 ALTO response from Network ....
>> - The list of S-D pairs should carry the Cost Type *values* as the 
>> supported Cost Types are expected to be listed and checked in the IRD.
>>
>> YOUNG>> We did not mean not leaving it out. See Section 3.5 example if that satisfies your question. 
>>   
>>     
> Ok, then adding the *value* to "cost type" would fully clarify
>
> YOUNG>> OK.
>
>   
>> - 'Constraint values': need to be clarified: is it the actual cost 
>> value? then the draft should specifiy that the constraits relate to Cost 
>> Types (as defined in §5.1.1).
>>
>> YOUNG>> Yes. Agree. 
>>
>> - how is the 'Administration Domain ID' of a S-D pair reported and why 
>> is it needed?  
>>
>> YOUNG>> Sorry for poor writing. We didn't mean this Admin Domain ID of a S-D pair. We actually meant here ALTO client may collect network resource information from different network domains in case of the multi-domain scenario. 
>>   
>>     
> ok
>   
>> ++ Section 3.3 Information Model of ALTO Query ...
>> Looking at the specification here, it is hard to figure out what is 
>> really extended here w.r.t. the Endpoint Cost Service query input 
>> specified in the base protocol.
>>
>> YOUNG>> Agreed. We will indicate what is new in the revision.
>>
>> - A new media type "CsoReqEndpointCostMap" and differs from 
>> "ReqEndpointCostMap" of the base protocol in that the set in member 
>> 'contraints' is encoded/interpreted differently, according to example in 
>> §3.5. Then a different member name than 'contraints' should be used.
>>
>> YOUNG>> Yes, we will make it clear. 
>>   
>>     
> Ok
>   
>> - As for member 'endpoints': unless you need to specify only 1 src EP, 
>> what is the need to define a new object "EndpointFilterExt" instead of 
>> keeping the "EndpointFilter"?
>>
>> YOUNG>> We may have a case in which to specify more than 1 src EP. The example is 
>> Simply an example with only 1 src EP. We have other examples where multiple sources may need to be specified. 
>>   
>>     
> Ok. Then why not keeping "EndpointFilter"?
>
> YOUNG>> 
>   
>> ++ Section 3.4 Information Model of ALTO Response ...
>> - object CsoInfoResourceEndpointCostMap has the same structure than 
>> InfoResourceEndpointCostMap of base protocol
>>
>> YOUNG>> Yes, we did not mean to invent new object.
>>   
>>     
> Ok. I guess a later version with updated information models and examples 
> will clarify this as well. 
>
> YOUNG>> Will clarify this in the revision.
>
>   
>> - what in the list of §3.2 does object 'DstCostsConstraints' refer to?
>> - this object seem to be encoded like a set of JSONNumbers that are the 
>> values of resp. hopcount, latency and packet loss. So I guess, they 
>> should be arranged in an array.
>>
>> YOUNG>> Actually it refers to constraint values. I tend to agree with you. 
>>
>> - why is this set of cost types already set at this level of 
>> specification? 
>>
>> YOUNG>> I don't understand this question. 
>>   
>>     
> I meant the list of cost types used in the constraints could be more 
> generic.
>
> YOUNG>> Let's discuss this point face-to-face. 
>
>   
>> ++ Section 3.5 ALTO protocol extension...
>> I agree with the idea of providing several Cost Type values at a time 
>> but the encoding could be lighter. The draft "Multi-Cost ALTO" proposes 
>> one example.
>> There is a real need that this section be harmonized with the previous 
>> specification sections, as the encoding provided in the examples are 
>> misleading.
>>
>> YOUNG>> Yes, we should have referred to your multi-cost draft properly. We did not mean to propose a new thing with respect to multi-cost. We will clean up what is new and what is referenced in the revision. 
>>   
>>     
> Ok.
>   
>> ++ Section 4.1 Representing....
>> - this section is rather about representing links and their attributes, 
>> so could be named accordingly
>>
>> YOUNG>> We discussed this aspect previously. Do you have a good suggestion of
>> Naming? 
>>   
>>     
> How about something like "Representing links and their attributes" ?
>
> YOUNG>> This sounds good to me. 
>
>   
>> - the graph specification is curretnly a list of links, do you intend to 
>> extend so that one can infer its structure?  
>>
>> YOUNG>> Please look at draft-bernstein-alto-large-bandwidth-cases-02.txt for
>> Further enhancement from this. We want to represent some path vector only with shared links and the link cost of those shared links so that some level of structure may be inferred by Application stratum.
>>   
>>     
> Ok.
>   
>> - is there a reason preventing member 'r-cap' to be specified as 
>> JSONNumber capacity?
>>
>> YOUNG>> I think we can. But I defer this question to Greg.
>>
>> - what is the usage of wt at this level of a spec ? (could also be 
>> 'JSONNumber weight')
>>
>> YOUNG>> Weight is similar to routing 'cost' notion of OSPF or other link cost that operators use for non-TE link cost. 
>>   
>>     
> I was mislead by the name 'wt' (sorry forgot the expression in the other 
> draft). 
>
> YOUNG>> OK
>
>   
>> Leeyoung a écrit :
>>   
>>     
>>> Hi All,
>>>
>>> We have just published ALTO extension to support application and network resource information exchange for high bandwidth applications.
>>> This is part of the i2aex initiative. 
>>>
>>> Thanks in advance for your comment and discussion. 
>>>
>>> Young
>>>
>>> -----Original Message-----
>>> From: internet-drafts@ietf.org [mailto:internet-drafts@ietf.org] 
>>> Sent: Monday, July 09, 2012 11:06 AM
>>> To: Leeyoung
>>> Cc: choits@etri.re.kr; Sreekanth madhavan; gregb@grotto-networking.com; Dhruv Dhody
>>> Subject: New Version Notification for draft-lee-alto-app-net-info-exchange-00.txt
>>>
>>>
>>> A new version of I-D, draft-lee-alto-app-net-info-exchange-00.txt
>>> has been successfully submitted by Young Lee and posted to the
>>> IETF repository.
>>>
>>> Filename:	 draft-lee-alto-app-net-info-exchange
>>> Revision:	 00
>>> Title:		 ALTO Extensions to Support Application and Network Resource Information Exchange for High Bandwidth Applications
>>> Creation date:	 2012-07-09
>>> WG ID:		 Individual Submission
>>> Number of pages: 14
>>> URL:             http://www.ietf.org/internet-drafts/draft-lee-alto-app-net-info-exchange-00.txt
>>> Status:          http://datatracker.ietf.org/doc/draft-lee-alto-app-net-info-exchange
>>> Htmlized:        http://tools.ietf.org/html/draft-lee-alto-app-net-info-exchange-00
>>>
>>>
>>> Abstract:
>>> This draft proposes ALTO information model and protocol extensions to
>>> support application and network resource information exchange for high
>>> bandwidth applications in partially controlled and controlled
>>> environments as part of the infrastructure to application information
>>> exposure (i2aex) initiative.
>>>
>>>
>>>
>>>                                                                                   
>>>
>>>
>>> The IETF Secretariat
>>> _______________________________________________
>>> altoext mailing list
>>> altoext@ietf.org
>>> https://www.ietf.org/mailman/listinfo/altoext
>>>   
>>>     
>>>       
>>   
>>     
>
>   

-- 

---------------------------------------------------------
Sabine RANDRIAMASY 
Alcatel-Lucent Bell Labs France 
Centre de Villarceau
Route de Villejust - 91620 NOZAY - FRANCE

E-MAIL : Sabine.Randriamasy@alcatel-lucent.com
TEL: +33 (0)1 30 77 27 45 
         (On Net) 2 103 27 45
---------------------------------------------------------


From sreekanth.madhavan@huawei.com  Wed Aug  1 03:28:41 2012
Return-Path: <sreekanth.madhavan@huawei.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 63DA321F8623 for <alto@ietfa.amsl.com>; Wed,  1 Aug 2012 03:28:41 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.149
X-Spam-Level: 
X-Spam-Status: No, score=-5.149 tagged_above=-999 required=5 tests=[AWL=-1.450, BAYES_00=-2.599, J_CHICKENPOX_56=0.6, MANGLED_NAIL=2.3, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id TX2ESo+rBZJs for <alto@ietfa.amsl.com>; Wed,  1 Aug 2012 03:28:40 -0700 (PDT)
Received: from dfwrgout.huawei.com (dfwrgout.huawei.com [206.16.17.72]) by ietfa.amsl.com (Postfix) with ESMTP id E0CF721F8628 for <alto@ietf.org>; Wed,  1 Aug 2012 03:28:39 -0700 (PDT)
Received: from 172.18.9.243 (EHLO dfweml202-edg.china.huawei.com) ([172.18.9.243]) by dfwrg01-dlp.huawei.com (MOS 4.2.3-GA FastPath) with ESMTP id AIP13917; Wed, 01 Aug 2012 02:28:39 -0800 (PST)
Received: from DFWEML404-HUB.china.huawei.com (10.193.5.203) by dfweml202-edg.china.huawei.com (172.18.9.108) with Microsoft SMTP Server (TLS) id 14.1.323.3; Wed, 1 Aug 2012 03:26:18 -0700
Received: from SZXEML412-HUB.china.huawei.com (10.82.67.91) by dfweml404-hub.china.huawei.com (10.193.5.203) with Microsoft SMTP Server (TLS) id 14.1.323.3; Wed, 1 Aug 2012 03:26:16 -0700
Received: from blrprnc03ns (10.18.96.92) by szxeml412-hub.china.huawei.com (10.82.67.91) with Microsoft SMTP Server id 14.1.323.3; Wed, 1 Aug 2012 18:26:05 +0800
From: sreekanth madhavan <sreekanthm@huawei.com>
To: <Sabine.Randriamasy@alcatel-lucent.com>
References: <7AEB3D6833318045B4AE71C2C87E8E1720CC2D35@dfweml511-mbx.china.huawei.com>	<5016C0DD.8050909@alcatel-lucent.com>	<7AEB3D6833318045B4AE71C2C87E8E172905E3A2@dfweml511-mbs.china.huawei.com> <501813EA.2040808@alcatel-lucent.com>
In-Reply-To: <501813EA.2040808@alcatel-lucent.com>
Date: Wed, 1 Aug 2012 15:56:05 +0530
Message-ID: <006001cd6fd0$0f14e240$2d3ea6c0$@com>
MIME-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
X-Mailer: Microsoft Office Outlook 12.0
Thread-Index: Ac1vQOjs8VGTn/3pSKmHTQQfrtX8IAAh0v2Q
Content-Language: en-us
X-Originating-IP: [10.18.96.92]
X-CFilter-Loop: Reflected
Cc: alto@ietf.org
Subject: Re: [alto] [altoext] FW: New Version Notification for draft-lee-alto-app-net-info-exchange-00.txt
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 01 Aug 2012 10:28:41 -0000

Hi Sabine,

Thanks for providing the comments.

My response for some of the comments , even though Young has already
clarified most of them.


++ Section 3.1 ALTO Query from Application Stratum to Network Stratum
Names used in the list of requested information may be misleading:

- A 'Cost Type' see base protocol (=A75.1.1) is expected to indicate an
attribute that has a value such as number, boolean, string... and is is
rather "semantic". 'summary' and 'graph' are cost attributes of =
different
nature.=20

----[sreekanth]: We will be extending the Cost-Type and would like to =
keep
the Cost-Mode as Summary and Gragh.


++ Section 3.4 Information Model of ALTO Response ...

- object CsoInfoResourceEndpointCostMap has the same structure than
InfoResourceEndpointCostMap of base protocol - what in the list of =
=A73.2 does
object 'DstCostsConstraints' refer to?

- this object seem to be encoded like a set of JSONNumbers that are the
values of resp. hopcount, latency and packet loss. So I guess, they =
should
be arranged in an array.=20

----[sreekanth]: Alto response:

   object {
               JSONNumber hopcount;
         JSONNumber latency;
         JSONNumber pktloss;
   } DstCostsConstraints;

   object EndpointDstCosts {
        DstCostsConstraints[TypedEndpointAddr];     ...
      };
      object {
        EndpointDstCosts [TypedEndpointAddr]<0..*>;
        ...
      } EndpointCostMapData;
      object {
        CostMode            cost-mode;
        CostType            cost-type;
        EndpointCostMapData map;
      } CsoInfoResourceEndpointCostMap;


For each Source Endpoint, a EndpointDstCosts object denotes the =
associated
"cost + constraints" to each
Destination Endpoint specified in the input parameters; the name for =
each
member in the object is the TypedEndpointAddr string identifying the
corresponding Destination Endpoint. This can be viewed as a two =
dimensional
array which holds cost + constraints value.

We extended the EndpointDstCosts by adding constraints value and hence
cannot reuse the EndpointDstCosts (which is wrong in the current
draft)object defined by base. The same is the case for
InfoResourceEndpointCostMap object of the base.


- why is this set of cost types already set at this level of =
specification?

----[sreekanth]: Could you please clarify this point ?

Regards,
Sreekanth
=20

-----Original Message-----
From: alto-bounces@ietf.org [mailto:alto-bounces@ietf.org] On Behalf Of
Sabine Randriamasy
Sent: Tuesday, July 31, 2012 10:51 PM
To: Leeyoung
Cc: alto@ietf.org
Subject: Re: [alto] [altoext] FW: New Version Notification for
draft-lee-alto-app-net-info-exchange-00.txt


Hi Young,
Thanks for your answers. Please see inline for further comments.
I have put the "alto" list in Cc rather than "altoext" which will be =
closed.
Best regards
Sabine

Leeyoung a =E9crit :
> Hi Sabine,
>
> Thanks for providing your comments to our drafts. Please see in-line =
for
my response.
>
> Best Regards,
> Young
>
> -----Original Message-----
> From: Sabine Randriamasy =
[mailto:Sabine.Randriamasy@alcatel-lucent.com]=20
> Sent: Monday, July 30, 2012 12:14 PM
> To: Leeyoung
> Cc: altoext@ietf.org
> Subject: Re: [altoext] FW: New Version Notification for
draft-lee-alto-app-net-info-exchange-00.txt
>
> Hi Young,
>
> I have read your draft and have a couple of comments and questions.
>
> In Section 2 Problem statement, I agree with the need to reduce the=20
> amount of conveyed ALTO information through topology filtering and=20
> constraint-based response filtering. The ALTO base protocol specifies=20
> these features for the Cost Map, Filtered cost map and Endpoint cost=20
> services. I also agree with the necessity to provide several types of=20
> cost information which stresses the necessity of extended filtering=20
> mechanisms. In the Constraints filtering extensions of Section 3, I=20
> didnt' really get how information confidentiality is preserved and the =

> proposed specification raised the questions detailed below.
>
> YOUNG>> Information confidentiality is a big issue which the current =
draft
has not addressed yet. The current draft was written in the "controlled" =
or
"partially controlled" environments which means there is a trusted
relationship amongst the entities involving the exchanges of resource =
data.=20
> In generic framework where there may be no such trusted relationship
amongst the entities, we need to address such issue down the road.=20
>  =20
Ok. Indeed, ALTO environment control is introduced in the Introduction.=20
I was mislead by seing "Such information may be considered sensitive to=20
the network provider just as..." just before "Section 3 provides ALTO=20
information model and protocol extensions to support... "
> I also agree that detailing end to end path costs at the link level at =

> some places can be beneficial, especially for local bottlenecks=20
> sometimes with fast changing cost values. Figure 2 shows an example =
with=20
> Link capacity costs but section 2 uses them rather as capacity, where =
I=20
> figure the higher the better. Without a hint on how path costs are=20
> calculated it is hard to follow the rationale on the DC choice by ER1. =

> See also the questions detailed below.
>
> YOUNG>> We actually enhanced this mechanism in section 6 of
draft-bernstein-alto-large-bandwidth-cases-02.txt
> where we include path vector along with link b/w. Path vector =
indicates
shared link information from which
> The bottle neck link can be identified.=20
>
> For instance, consider the following example from the draft.=20
>
>
>          +----+                  L0 Wt=3D10,BW=3D50         +----+
>          | N0 |-----------------------------------------| N3 |
>          +----+  `.                                     +----+
>            |       `. L4 Wt=3D7                             |
>            |         `-. BW=3D40                            |
>            |            `.  +----+                        |
>            |              `.| N4 |                        |
>            | L1          .' +----+                        |
>            | Wt=3D10      /                          L2     |
>            | BW=3D45     /                           Wt=3D12  |
>            |          /L5 Wt=3D10                    BW=3D30  |
>            |        .'    BW=3D45                           |
>            |       /                                      |
>            |      /                                       |
>          +----+ .'              L3 Wt=3D15 BW=3D42          +----+
>          | N1 |.........................................| N2 |
>          +----+                                         +----+
>
> Now suppose the network of Figure 1 is a TDM network controlled by=20
> GMPLS. Once again N0 representing a large data center and nodes N2
> and N3 as potential clients. However in this case the network provider =

> offers an additional path, P3, for getting from N0-N2.
>
>    Path  Src-Dest    Path Vector    Path Cost
>
>    P1    N0-N2       {L0, L2}       22
>    P2    N0-N3       {L0}           10
>    P3    N0-N2       {L1,L3}        25
>    ----------------------------------
>    Link        Bandwidth
>    L0             50
>    L1             45
>    L2             30
>    L3             42
>
> So the draft now propose that the network can provide abstract path=20
> list along with link bottlenecks. We also provide how this mechanism=20
> works for IP and other types of networks.=20
>  =20
I'm fine with the rationale in=20
draft-bernstein-alto-large-bandwidth-cases-02.txt. It's just that in=20
draft-lee-alto-app-net-info-exchange-00.txt I missed the relation=20
between the "LinkCapacity Cost" & ALTO Cost Map in Figure 2 and also the =

explanatory text text. Maybe a little clarifying edition can help. =20
> I will provide the responses to your detail questions later.=20
> Thanks.=20
>
> Young
>
>
>
>         =20
>
> Thanks,
> Sabine
>
> ------------------------- Detailed questions =
-----------------------------
>
> ++ Section 3.1 ALTO Query from Application Stratum to Network Stratum
> Names used in the list of requested information may be misleading:
> - A 'Cost Type' see base protocol (=A75.1.1) is expected to indicate =
an=20
> attribute that has a value such as number, boolean, string... and is =
is=20
> rather "semantic". 'summary' and 'graph' are cost attributes of=20
> different nature.
> - 'Constraints' such as 'min/max metric' look like objectives rather=20
> than expressions such as 'lt. value' as specified in the base protocol =

> (=A7 6.8.2.2.3)
> - what kind of information is member 'Parameters' is supposed to carry =

> and for what use?
> - Some members listed here do not appear in the ALTO Query Information =

> Model of =A73.3: 'parameters' 'objective-function'. Neither do they =
appear=20
> in the example of =A73.5
>
>
> ++ Section 3.2 ALTO response from Network ....
> - The list of S-D pairs should carry the Cost Type *values* as the=20
> supported Cost Types are expected to be listed and checked in the IRD.
> - 'Constraint values': need to be clarified: is it the actual cost=20
> value? then the draft should specifiy that the constraits relate to =
Cost=20
> Types (as defined in =A75.1.1).
> - how is the 'Administration Domain ID' of a S-D pair reported and why =

> is it needed? =20
>
>
> ++ Section 3.3 Information Model of ALTO Query ...
> Looking at the specification here, it is hard to figure out what is=20
> really extended here w.r.t. the Endpoint Cost Service query input=20
> specified in the base protocol.
> - A new media type "CsoReqEndpointCostMap" and differs from=20
> "ReqEndpointCostMap" of the base protocol in that the set in member=20
> 'contraints' is encoded/interpreted differently, according to example =
in=20
> =A73.5. Then a different member name than 'contraints' should be used.
> - As for member 'endpoints': unless you need to specify only 1 src EP, =

> what is the need to define a new object "EndpointFilterExt" instead of =

> keeping the "EndpointFilter"?
>
>
> ++ Section 3.4 Information Model of ALTO Response ...
> - object CsoInfoResourceEndpointCostMap has the same structure than=20
> InfoResourceEndpointCostMap of base protocol
> - what in the list of =A73.2 does object 'DstCostsConstraints' refer =
to?
> - this object seem to be encoded like a set of JSONNumbers that are =
the=20
> values of resp. hopcount, latency and packet loss. So I guess, they=20
> should be arranged in an array.
> - why is this set of cost types already set at this level of=20
> specification?=20
>
>
> ++ Section 3.5 ALTO protocol extension...
> I agree with the idea of providing several Cost Type values at a time=20
> but the encoding could be lighter. The draft "Multi-Cost ALTO" =
proposes=20
> one example.
> There is a real need that this section be harmonized with the previous =

> specification sections, as the encoding provided in the examples are=20
> misleading.
>
>
> ++ Section 4.1 Representing....
> - this section is rather about representing links and their =
attributes,=20
> so could be named accordingly
> - the graph specification is curretnly a list of links, do you intend =
to=20
> extend so that one can infer its structure? =20
> - is there a reason preventing member 'r-cap' to be specified as=20
> JSONNumber capacity?
> - what is the usage of wt at this level of a spec ? (could also be=20
> 'JSONNumber weight')
>
> Leeyoung a =E9crit :
>  =20
>> Hi All,
>>
>> We have just published ALTO extension to support application and =
network
resource information exchange for high bandwidth applications.
>> This is part of the i2aex initiative.=20
>>
>> Thanks in advance for your comment and discussion.=20
>>
>> Young
>>
>> -----Original Message-----
>> From: internet-drafts@ietf.org [mailto:internet-drafts@ietf.org]=20
>> Sent: Monday, July 09, 2012 11:06 AM
>> To: Leeyoung
>> Cc: choits@etri.re.kr; Sreekanth madhavan; =
gregb@grotto-networking.com;
Dhruv Dhody
>> Subject: New Version Notification for
draft-lee-alto-app-net-info-exchange-00.txt
>>
>>
>> A new version of I-D, draft-lee-alto-app-net-info-exchange-00.txt
>> has been successfully submitted by Young Lee and posted to the
>> IETF repository.
>>
>> Filename:	 draft-lee-alto-app-net-info-exchange
>> Revision:	 00
>> Title:		 ALTO Extensions to Support Application and Network
Resource Information Exchange for High Bandwidth Applications
>> Creation date:	 2012-07-09
>> WG ID:		 Individual Submission
>> Number of pages: 14
>> URL:
http://www.ietf.org/internet-drafts/draft-lee-alto-app-net-info-exchange-=
00.
txt
>> Status:
http://datatracker.ietf.org/doc/draft-lee-alto-app-net-info-exchange
>> Htmlized:
http://tools.ietf.org/html/draft-lee-alto-app-net-info-exchange-00
>>
>>
>> Abstract:
>> This draft proposes ALTO information model and protocol extensions to
>> support application and network resource information exchange for =
high
>> bandwidth applications in partially controlled and controlled
>> environments as part of the infrastructure to application information
>> exposure (i2aex) initiative.
>>
>>
>>
>>

>>
>>
>> The IETF Secretariat
>> _______________________________________________
>> altoext mailing list
>> altoext@ietf.org
>> https://www.ietf.org/mailman/listinfo/altoext
>>  =20
>>    =20
>
>  =20

_______________________________________________
alto mailing list
alto@ietf.org
https://www.ietf.org/mailman/listinfo/alto


From sreekanth.madhavan@huawei.com  Wed Aug  1 04:54:03 2012
Return-Path: <sreekanth.madhavan@huawei.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 0E1F521F86B6 for <alto@ietfa.amsl.com>; Wed,  1 Aug 2012 04:54:03 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.874
X-Spam-Level: 
X-Spam-Status: No, score=-5.874 tagged_above=-999 required=5 tests=[AWL=0.725,  BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Tc4gzd0JusWT for <alto@ietfa.amsl.com>; Wed,  1 Aug 2012 04:54:02 -0700 (PDT)
Received: from dfwrgout.huawei.com (dfwrgout.huawei.com [206.16.17.72]) by ietfa.amsl.com (Postfix) with ESMTP id F1E4021F86B9 for <alto@ietf.org>; Wed,  1 Aug 2012 04:54:01 -0700 (PDT)
Received: from 172.18.9.243 (EHLO dfweml202-edg.china.huawei.com) ([172.18.9.243]) by dfwrg01-dlp.huawei.com (MOS 4.2.3-GA FastPath) with ESMTP id AIP18989; Wed, 01 Aug 2012 03:54:01 -0800 (PST)
Received: from DFWEML406-HUB.china.huawei.com (10.193.5.131) by dfweml202-edg.china.huawei.com (172.18.9.108) with Microsoft SMTP Server (TLS) id 14.1.323.3; Wed, 1 Aug 2012 04:52:09 -0700
Received: from SZXEML409-HUB.china.huawei.com (10.82.67.136) by dfweml406-hub.china.huawei.com (10.193.5.131) with Microsoft SMTP Server (TLS) id 14.1.323.3; Wed, 1 Aug 2012 04:52:07 -0700
Received: from blrprnc03ns (10.18.96.92) by szxeml409-hub.china.huawei.com (10.82.67.136) with Microsoft SMTP Server id 14.1.323.3; Wed, 1 Aug 2012 19:51:57 +0800
From: sreekanth madhavan <sreekanthm@huawei.com>
To: <alto@ietf.org>
Date: Wed, 1 Aug 2012 17:21:56 +0530
Message-ID: <000001cd6fdc$0daa3840$28fea8c0$@com>
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: quoted-printable
X-Mailer: Microsoft Office Outlook 12.0
Thread-Index: Ac1v2uk5CciUkP/2Rh6VsbsfuSF2UQAACsjQ
Content-Language: en-us
X-Originating-IP: [10.18.96.92]
X-CFilter-Loop: Reflected
Subject: [alto] FW: New Version Notification for draft-madhavan-alto-caching-00.txt
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 01 Aug 2012 11:54:03 -0000

Hi All,

We have just published a draft on ALTO caching. We have separated =
caching and subscription details (draft-alto-caching-subscription-00) =
into two separate drafts. This draft discusses about the various cache =
invalidation mechanisms also.

Thanks in advance for your comment and discussion.=20

Regards,
Sreekanth

-----Original Message-----
From: internet-drafts@ietf.org [mailto:internet-drafts@ietf.org]=20
Sent: Wednesday, August 01, 2012 5:14 PM
To: sreekanth.madhavan@huawei.com
Cc: ravi.nandiraju@huawei.com
Subject: New Version Notification for draft-madhavan-alto-caching-00.txt


A new version of I-D, draft-madhavan-alto-caching-00.txt
has been successfully submitted by Sreekanth Madhavan and posted to the
IETF repository.

Filename:	 draft-madhavan-alto-caching
Revision:	 00
Title:		 ALTO Caching
Creation date:	 2012-08-01
WG ID:		 Individual Submission
Number of pages: 11
URL:             =
http://www.ietf.org/internet-drafts/draft-madhavan-alto-caching-00.txt
Status:          =
http://datatracker.ietf.org/doc/draft-madhavan-alto-caching
Htmlized:        =
http://tools.ietf.org/html/draft-madhavan-alto-caching-00


Abstract:
   The specification of the ALTO protocol uses map based approaches
   assuming that the information provided is static for a longer period
   of time.  The purpose of this memo is to provide a mechanism where
   ALTO clients cache the map information and the servers provide the
   expiration time for invalidation.

                                                                         =
        =20


The IETF Secretariat


From sreekanth.madhavan@huawei.com  Wed Aug  1 04:58:28 2012
Return-Path: <sreekanth.madhavan@huawei.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 35F8921F8679 for <alto@ietfa.amsl.com>; Wed,  1 Aug 2012 04:58:28 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.116
X-Spam-Level: 
X-Spam-Status: No, score=-6.116 tagged_above=-999 required=5 tests=[AWL=0.483,  BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id EqCoY-yOrZph for <alto@ietfa.amsl.com>; Wed,  1 Aug 2012 04:58:27 -0700 (PDT)
Received: from dfwrgout.huawei.com (dfwrgout.huawei.com [206.16.17.72]) by ietfa.amsl.com (Postfix) with ESMTP id 7F22221F867A for <alto@ietf.org>; Wed,  1 Aug 2012 04:58:27 -0700 (PDT)
Received: from 172.18.9.243 (EHLO dfweml202-edg.china.huawei.com) ([172.18.9.243]) by dfwrg02-dlp.huawei.com (MOS 4.2.3-GA FastPath) with ESMTP id AII19239; Wed, 01 Aug 2012 03:58:27 -0800 (PST)
Received: from DFWEML406-HUB.china.huawei.com (10.193.5.131) by dfweml202-edg.china.huawei.com (172.18.9.108) with Microsoft SMTP Server (TLS) id 14.1.323.3; Wed, 1 Aug 2012 04:56:33 -0700
Received: from SZXEML402-HUB.china.huawei.com (10.82.67.32) by dfweml406-hub.china.huawei.com (10.193.5.131) with Microsoft SMTP Server (TLS) id 14.1.323.3; Wed, 1 Aug 2012 04:56:32 -0700
Received: from blrprnc03ns (10.18.96.92) by szxeml402-hub.china.huawei.com (10.82.67.32) with Microsoft SMTP Server id 14.1.323.3; Wed, 1 Aug 2012 19:55:12 +0800
From: sreekanth madhavan <sreekanthm@huawei.com>
To: <alto@ietf.org>
Date: Wed, 1 Aug 2012 17:25:12 +0530
Message-ID: <000101cd6fdc$82228100$86678300$@com>
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: quoted-printable
X-Mailer: Microsoft Office Outlook 12.0
Thread-Index: Ac1v2vBgWDD1dPzWRUumVLy+PjHP/QAATN5g
Content-Language: en-us
X-Originating-IP: [10.18.96.92]
X-CFilter-Loop: Reflected
Subject: [alto] FW: New Version Notification for draft-madhavan-alto-subscription-00.txt
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 01 Aug 2012 11:58:28 -0000

Hi All,

We have just published a draft on ALTO subscription.=20

Please review and provide your comments.

Regards,
Sreekanth

-----Original Message-----
From: internet-drafts@ietf.org [mailto:internet-drafts@ietf.org]=20
Sent: Wednesday, August 01, 2012 5:14 PM
To: sreekanth.madhavan@huawei.com
Cc: ravi.nandiraju@huawei.com
Subject: New Version Notification for =
draft-madhavan-alto-subscription-00.txt


A new version of I-D, draft-madhavan-alto-subscription-00.txt
has been successfully submitted by Sreekanth Madhavan and posted to the
IETF repository.

Filename:	 draft-madhavan-alto-subscription
Revision:	 00
Title:		 ALTO Subscription
Creation date:	 2012-08-01
WG ID:		 Individual Submission
Number of pages: 15
URL:             =
http://www.ietf.org/internet-drafts/draft-madhavan-alto-subscription-00.t=
xt
Status:          =
http://datatracker.ietf.org/doc/draft-madhavan-alto-subscription
Htmlized:        =
http://tools.ietf.org/html/draft-madhavan-alto-subscription-00


Abstract:
   The specification of the ALTO protocol uses map based approaches
   assuming that the information provided is static for a longer period
   of time.  But in some cases network operators reallocate IP subnets
   from time to time which in turn changes the mapping partitions[I-
   D.ietf-alto-deployments].  Since the ALTO clients are unaware of the
   map information changes, clients need to query the servers for every
   service request and many such requests are redundant because the
   information was not changed.  The purpose of this memo is to propose
   a mechanism where ALTO clients can subscribe for event notifications
   from the server

                                                                         =
        =20


The IETF Secretariat


From richard.alimi@gmail.com  Wed Aug  1 14:58:24 2012
Return-Path: <richard.alimi@gmail.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 69B1311E835C for <alto@ietfa.amsl.com>; Wed,  1 Aug 2012 14:58:24 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.782
X-Spam-Level: 
X-Spam-Status: No, score=-2.782 tagged_above=-999 required=5 tests=[AWL=0.195,  BAYES_00=-2.599, FM_FORGED_GMAIL=0.622, RCVD_IN_DNSWL_LOW=-1]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 66k76O5khacl for <alto@ietfa.amsl.com>; Wed,  1 Aug 2012 14:58:23 -0700 (PDT)
Received: from mail-vc0-f172.google.com (mail-vc0-f172.google.com [209.85.220.172]) by ietfa.amsl.com (Postfix) with ESMTP id 1586111E831B for <alto@ietf.org>; Wed,  1 Aug 2012 14:58:22 -0700 (PDT)
Received: by vcbfo14 with SMTP id fo14so8116067vcb.31 for <alto@ietf.org>; Wed, 01 Aug 2012 14:58:22 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:sender:in-reply-to:references:from:date :x-google-sender-auth:message-id:subject:to:content-type; bh=jyUh9dr8MiRmDrsW0k/lO2PG+x1O3g2us0OVM1YXNQ4=; b=eVESKHPoQqVXDf9OGewrukXlzP0yDW6dTbcQBZ09JTdoaHVlgxNAkHjCOwnBzB7icN OvpFmE8d2E6NHdQJ105jTIJn+8G//dxwIhfah0/EXSgEp/9NddluBecxvw4WCoZ1zfHz s/R76r0XCkuuzcxMGwjjJOQifQOluvgsrg64ipmpN0MHSsNY3E8sOuEInM2z+iaZSm5l kAc8z2dZ8wm5WnzeJJaIS4nNbU0p4RAbK9OurnpPw6vNiMHePf7SQiYx2vVN0JWRpv0U 6GY5fsaEU5jhzv+OoaLekj25xe4Plf+p9pc8Eurh5/0t8D09qdLkHhe29ti3em+97wU/ 5CYg==
Received: by 10.58.133.48 with SMTP id oz16mr7213884veb.1.1343858302487; Wed, 01 Aug 2012 14:58:22 -0700 (PDT)
MIME-Version: 1.0
Sender: richard.alimi@gmail.com
Received: by 10.58.244.227 with HTTP; Wed, 1 Aug 2012 14:58:02 -0700 (PDT)
In-Reply-To: <CA+cvDaZZ7ow7QWMjmGkLs0E7c0PK2KjPApVtr_z4qQTOyBTwpQ@mail.gmail.com>
References: <CA+cvDaZZ7ow7QWMjmGkLs0E7c0PK2KjPApVtr_z4qQTOyBTwpQ@mail.gmail.com>
From: Richard Alimi <rich@velvetsea.net>
Date: Wed, 1 Aug 2012 14:58:02 -0700
X-Google-Sender-Auth: 9y-Cl2ZSTsHhEAa3zlDJtRT-pJU
Message-ID: <CA+cvDaYfYtyo7WLTUmbesMS2X89r4wL_5NUfWWrAzRcEG3Ve4g@mail.gmail.com>
To: alto@ietf.org
Content-Type: text/plain; charset=ISO-8859-1
Subject: Re: [alto] ALTO Operations and Management
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 01 Aug 2012 21:58:24 -0000

We discussed this during the meeting, and the suggestion was to get a
discussion going on the list for exactly which aspects of RFC5706
should be discussed in draft-ietf-alto-protocol.  I would very much
like to avoid revisiting what should and shouldn't be included at a
later date.  Thus, consider it started :)

I'll get the ball rolling:
- Logging
  - Use syslog
  - Minimum set of events that SHOULD be logged
    - Request logs (may already be provided by HTTP server implementation)
    - ALTO Map updates (version tag, cost mode, cost type, size in
both entries and encoded bytes)
    - Changes to runtime configuration (e.g., data source
configuration, security parameters)
    - Failure to update served ALTO information (e.g., failure to
contact a data source used to compute ALTO info)

Installation and Initial Setup
- By default an ALTO Server may be configured with a single PID
containing all IPv4/IPv6 addresses and a Cost Map with no entries.
Something more interesting should be configured explicitly (either
dynamically or statically).

Migration Path
- Applications making use of other data sources for network
information should consider ALTO an additional source of information,
but ALTO need not be the sole source of network information.
Therefore, it can be incorporated incrementally into applications.

Requirements on Other Protocols and Functional Components
- Protocol-wise, ALTO assumes standard HTTP and JSON implementations exist.
- Functional-wise: An ALTO Server assumes that it can gather
sufficient information to populate Network and Cost maps.  "Sufficient
information" is dependent on the information being exposed, but likely
includes information such as IGP and EGP RIBs (more details in Section
2.2 of draft-ietf-alto-protocol-12).  Extension documents may provide
specific mechanisms.

Impact on Network Operation
- Operators should consider impacts on (or integration with) Traffic
Engineering, since ALTO may shift traffic patterns.
- Privacy considerations for ISPs (see 10.1 of draft-ietf-alto-protocol-12)
- Impact of incorrect/faked guidance (see 10.3 of
draft-ietf-alto-deployments-04)


Anything else?

Thanks,
Rich


On Sun, Jul 29, 2012 at 11:43 PM, Richard Alimi <rich@velvetsea.net> wrote:
> Hi All,
>
> As you probably know, one of the new requirements on the protocol is:
>    An ALTO client protocol MUST provide adequate
>    mechanisms for operations and management support, as outlined in RFC
>    5706 [RFC5706].
>
> The points raised by RFC5706 are very good and should indeed be
> solved. What I'm missing currently is a clear picture of exactly what
> this requirement dictates for draft-ietf-alto-protocol.  In
> particular, the scope of RFC5706 is *huge* - to address every
> suggested point fully would take another couple of years (e.g.,
> specify a management API for ALTO Servers, a configuration language,
> techniques for verifying end-to-end operation, etc).
>
> I'll suggest a way forwards - please provide feedback if you have a
> chance.  I'd really like to get something mostly complete by the end
> of IETF week if possible.
>
> Proposal:
>
> If we separate Operations (RFC5706, Section 2) from Management
> (RFC5706, Section 3), I can make the following general observation:
> - Operations: the protocol draft seems like a fine place to document
> these, with the exception of 2.6 ("Verifying Correct Operation") which
> is covered in Section 6.3 of draft-ietf-alto-deployments-04
> - Management: draft-ietf-alto-deployments seems to already cover some
> of this, and could probably be expanded to cover the rest (well, the
> rest that we intend to solve) to keep the related information
> together.  That seems preferable than copying a bunch of content from
> draft-ietf-alto-deployments into draft-ietf-alto-protocol when we had
> intentionally kept them separate in the first place.  The exception to
> this might be dictating which protocols to use for some functions that
> we know would be needed (syslog, IPFIX), which seem fine to have in
> draft-ietf-alto-protocol.
>
> Thoughts on that?  The intent isn't to just punt this over to another
> draft, but rather to avoid duplicating text between different
> documents and finish the protocol draft in a reasonable amount of
> time.
>
> If that *does* sound reasonable, I can write up some more concrete
> text for the Operations section with a pointer indicating that
> Management considerations are being documented in
> draft-ietf-alto-deployments (seems like an Informative reference, so
> that should be fine?)
>
> If that *doesn't* sound reasonable, I'd really like to hear from
> people what should be documented in the ALTO Protocol draft itself to
> support Operations and Management.  Below my signature, I've included
> some quick notes below about what I think *could* be discussed for
> most of the points raised by RFC5706.  I don't think the authors of
> the new requirement envisioned all of that to be included in
> draft-ietf-alto-protocol, but I'd need some help in defining where to
> draw the line.
>
> Thanks,
> Rich
>
> -- Operations --
>
> (1) Installation and Initial Setup
> - By default an ALTO Server may be configured with a single PID
> containing all IPv4/IPv6 addresses and a Cost Map with no entries.
> Something more interesting should be configured explicitly (either
> dynamically or statically).
> - Specifications for underlying protocols (e.g., TCP, HTTP) should be
> consulted for their available settings/defaults.
>
> (2) Migration Path
> - No migration path for ALTO Servers since there is no previous
> protocol providing same functionality.
> - Client-side: Applications making use of other data sources for
> network information should consider ALTO an additional source of
> information, but ALTO need not be the sole source of network
> information.
>
> (3) Requirements on Other Protocols and Functional Components
> - Protocol-wise, ALTO assumes standard HTTP and JSON implementations exist.
> - Functional-wise: An ALTO Server assumes that it can gather
> sufficient information to populate Network and Cost maps.  "Sufficient
> information" is dependent on the information being exposed, but likely
> includes information such as IGP and EGP RIBs (more details in Section
> 2.2 of draft-ietf-alto-protocol-12).  Specific mechanisms have been
> suggested in the past as extensions (e.g.,
> draft-medved-alto-svr-apis-00).
>
> (4) Impact on Network Operation
> - Operators should consider impacts on (or integration with) Traffic
> Engineering, since ALTO may shift traffic patterns.
> - Privacy considerations for ISPs (see 10.1 of draft-ietf-alto-protocol-12)
> - Impact of incorrect/faked guidance (see 10.3 of
> draft-ietf-alto-deployments-04)
>
> (5) Verifying Correct Operation
> - For ALTO-specific monitoring and metrics, see 6.3 of
> draft-ietf-alto-deployments-04
>
> -- Management --
>
> (1) Interoperability
> - A common management API would be desirable given that ALTO Servers
> may typically be configured with dynamic data from various sources,
> and ALTO Servers are intended to scale horizontally for
> fault-tolerance and reliability
>   - management protocol should operate semantically at the level of
> ALTO maps or other exposed information
> - Entities that compute dynamic maps/guidance and entities that
> measure ALTO's effects should use IPFIX information elements
> - ALTO Servers should use syslog to log events.
>
> (2) Management Information
> - Informational Model
>   - Data Sources - the set of data sources made available to an ALTO Server
>   - Information Resource Directory - the set of ALTO Information
> exposed by an ALTO Server
>   - Section 4 and 5 of draft-ietf-alto-protocol-12 describe Network
> Map, PIDs and Cost Maps, which are central entities of the
> informational model.
>   - ALTO guidance: ALTO Server may use internal logic/algorithms to
> generate responses to ALTO Clients (map and endpoint-based queries);
> may be customized to a particular deployment environment, so not
> standardized (as long as responses follow protocol spec). May use
> Network Map and Cost Map to answer queries, or may use custom logic.
>
> (3) Fault Management
> - To be discussed in 6.3 of draft-ietf-alto-deployments
>
> (4) Configuration Management
> - ALTO Server requires at least the following (high-level) logical inputs:
>   - Data Sources (raw network , or explicitly-provided ALTO-level
> information such as a Network Map and Cost Map)
>   - Security policies
> - Configuration interface should use ALTO data structures (as
> specified in the ALTO Protocol itself) wherever possible to configure
> the ALTO information exposed by an ALTO Server.  Specific syntax may
> differ if porting to a language that is not JSON, but structures
> should remain consistent.
> - Specify standard configuration language and interface for interoperability
> - Multiple ALTO Servers my be deployed for horizontal scalability. A
> centralized configuration database may be used to ensure they are
> providing the desired ALTO information with appropriate security
> controls.  The ALTO information (e.g., Network Maps and Cost Maps)
> being served by each ALTO Server, as well as security policies (HTTP
> authentication, SSL/TLS client/server authentication, SSL/TLS
> encryption parameters) intended to serve the same information should
> be monitored for consistency.
> - Verifying Correct Operation: Discussed in 6.3 of
> draft-ietf-alto-deployments (and future revisions)
>
> (5) Accounting Management
> - Discussed in 6.3 of draft-ietf-alto-deployments (and future revisions)
>
> (6) Performance Management
> - List interesting counters
>   - e.g., Number of requests for each service, responses, bytes
> in/out, CPU utilization, ALTO map/guidance updates, number of PIDs,
> map sizes (bytes, number of entries), etc
> - Interesting data accessible by IPFIX
>   - e.g., bytes per src/dst IP
> - network devices
>   - link utilization
>
> (7) Security Management
> - Readers should refer to documentation on HTTP and SSL/TLS.
> - ALTO Servers will typically receive a large number of requests.
> Intrusion Detection Systems that can separate malicious attacks from
> noise are suggested when generating alerts.
>
>
> Thanks,
> Rich

From ietf@meetecho.com  Thu Aug  2 11:00:04 2012
Return-Path: <ietf@meetecho.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id CAF2A11E80E6 for <alto@ietfa.amsl.com>; Thu,  2 Aug 2012 11:00:03 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.514
X-Spam-Level: 
X-Spam-Status: No, score=-0.514 tagged_above=-999 required=5 tests=[AWL=0.205,  BAYES_00=-2.599, HELO_EQ_IT=0.635, HOST_EQ_IT=1.245]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id pKIwDHKByQLq for <alto@ietfa.amsl.com>; Thu,  2 Aug 2012 11:00:03 -0700 (PDT)
Received: from smtplq03.aruba.it (smtplq-out5.aruba.it [62.149.158.25]) by ietfa.amsl.com (Postfix) with SMTP id 9A01021F841B for <alto@ietf.org>; Thu,  2 Aug 2012 11:00:02 -0700 (PDT)
Received: (qmail 4248 invoked by uid 89); 2 Aug 2012 17:59:59 -0000
Received: from unknown (HELO smtp8.aruba.it) (62.149.158.228) by smtplq03.aruba.it with SMTP; 2 Aug 2012 17:59:59 -0000
Received: (qmail 32504 invoked by uid 89); 2 Aug 2012 17:59:59 -0000
Received: from unknown (HELO ?130.129.21.177?) (alex@meetecho.com@130.129.21.177) by smtp8.ad.aruba.it with ESMTPA; 2 Aug 2012 17:59:59 -0000
Message-ID: <501AC017.8010903@meetecho.com>
Date: Thu, 02 Aug 2012 19:59:51 +0200
From: Meetecho IETF support <ietf@meetecho.com>
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:14.0) Gecko/20120713 Thunderbird/14.0
MIME-Version: 1.0
To: alto@ietf.org
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 7bit
X-Spam-Rating: smtplq03.aruba.it 1.6.2 0/1000/N
Subject: [alto] Meetecho session recording available
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 02 Aug 2012 18:00:06 -0000

Dear all,

the full recording (synchronized video, audio, slides and jabber room)
of ALTO session at IETF-84 is available.

You can watch it by accessing the following URL:
http://ietf84.conf.meetecho.com/index.php/Recorded_Sessions#IETF84_ALTO

For the chair(s): please feel free to put the link to the recording in 
the minutes, if you think this might be useful.

In case of problems with the playout, just drop an e-mail to 
team@meetecho.com.

Cheers,
the Meetecho team

-- 
Meetecho s.r.l.
Web Conferencing and Collaboration Tools
www.meetecho.com

From Tina.Tsou.Zouting@huawei.com  Thu Aug  2 17:59:38 2012
Return-Path: <Tina.Tsou.Zouting@huawei.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D79A121F8BFE for <alto@ietfa.amsl.com>; Thu,  2 Aug 2012 17:59:38 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.958
X-Spam-Level: 
X-Spam-Status: No, score=-5.958 tagged_above=-999 required=5 tests=[AWL=0.640,  BAYES_00=-2.599, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id n1ZDd30DSdTI for <alto@ietfa.amsl.com>; Thu,  2 Aug 2012 17:59:38 -0700 (PDT)
Received: from dfwrgout.huawei.com (dfwrgout.huawei.com [206.16.17.72]) by ietfa.amsl.com (Postfix) with ESMTP id 2C12D21F8BFD for <alto@ietf.org>; Thu,  2 Aug 2012 17:59:38 -0700 (PDT)
Received: from 172.18.9.243 (EHLO dfweml201-edg.china.huawei.com) ([172.18.9.243]) by dfwrg02-dlp.huawei.com (MOS 4.2.3-GA FastPath) with ESMTP id AIJ45668; Thu, 02 Aug 2012 16:59:32 -0800 (PST)
Received: from DFWEML406-HUB.china.huawei.com (10.193.5.131) by dfweml201-edg.china.huawei.com (172.18.9.107) with Microsoft SMTP Server (TLS) id 14.1.323.3; Thu, 2 Aug 2012 17:57:43 -0700
Received: from DFWEML513-MBS.china.huawei.com ([169.254.4.5]) by dfweml406-hub.china.huawei.com ([10.193.5.131]) with mapi id 14.01.0323.003; Thu, 2 Aug 2012 17:57:46 -0700
From: Tina TSOU <Tina.Tsou.Zouting@huawei.com>
To: "alto@ietf.org" <alto@ietf.org>
Thread-Topic: PoC 7-9PM this evening: draft-xie-alto-sdn-extension-use-cases
Thread-Index: Ac1xEv8jRe15PagoTNq7sSBvwS1mxQ==
Date: Fri, 3 Aug 2012 00:57:46 +0000
Message-ID: <FBA95858-3D40-44F5-99B3-7690DFE6586D@huawei.com>
Accept-Language: en-US, zh-CN
Content-Language: en-US
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Content-Type: multipart/alternative; boundary="_000_FBA958583D4044F599B37690DFE6586Dhuaweicom_"
MIME-Version: 1.0
X-CFilter-Loop: Reflected
Subject: [alto] PoC 7-9PM this evening: draft-xie-alto-sdn-extension-use-cases
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 03 Aug 2012 00:59:39 -0000

--_000_FBA958583D4044F599B37690DFE6586Dhuaweicom_
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64

Qml0cy1OLUJpdGVzDQoNClRlY2hub2xvZ3kgJiBOZXR3b3JraW5nIFNvY2lhbA0KSUVURiA4NCBN
ZWV0aW5nIC0gVmFuY291dmVyLCBCQywgQ2FuYWRhDQpIeWF0dCBWYW5jb3V2ZXINCkRhdGU6IEF1
Z3VzdCAyLCAyMDEyDQpUaW1lOiAxOTowMC0yMTowMA0KTG9jYXRpb246IFBsYXphIEFCDQoNCg0K
VGluYQ0K

--_000_FBA958583D4044F599B37690DFE6586Dhuaweicom_
Content-Type: text/html; charset="utf-8"
Content-ID: <A0F3BCFB59AFBA4BA8E97DFA5A160466@huawei.com>
Content-Transfer-Encoding: base64
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--_000_FBA958583D4044F599B37690DFE6586Dhuaweicom_--

From yry@cs.yale.edu  Fri Aug  3 11:04:13 2012
Return-Path: <yry@cs.yale.edu>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 3575521F8D7B; Fri,  3 Aug 2012 11:04:13 -0700 (PDT)
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_33=0.6]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id qMIAISQrmDym; Fri,  3 Aug 2012 11:04:11 -0700 (PDT)
Received: from vm-emlprdomr-06.its.yale.edu (vm-emlprdomr-06.its.yale.edu [130.132.50.147]) by ietfa.amsl.com (Postfix) with ESMTP id A9FFC21F8D7D; Fri,  3 Aug 2012 11:04:04 -0700 (PDT)
Received: from Faculty-Supports-MacBook-Pro-2.local ([128.36.46.193]) (authenticated bits=0) by vm-emlprdomr-06.its.yale.edu (8.14.4/8.14.4) with ESMTP id q73I3vJD030038 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NOT); Fri, 3 Aug 2012 14:03:59 -0400
Message-ID: <501C128D.10809@cs.yale.edu>
Date: Fri, 03 Aug 2012 11:03:57 -0700
From: "Y. Richard Yang" <yry@cs.yale.edu>
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.7; rv:14.0) Gecko/20120713 Thunderbird/14.0
MIME-Version: 1.0
To: stefano previdi <sprevidi@cisco.com>
References: <CC404D94.2D5D%tnadeau@juniper.net> <501B0D75.4090009@raszuk.net> <7E89A05A-CE4E-4FCF-81AB-8F39B42FBF8E@cisco.com>
In-Reply-To: <7E89A05A-CE4E-4FCF-81AB-8F39B42FBF8E@cisco.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-Scanned-By: MIMEDefang 2.71 on 130.132.50.147
Cc: James Kempf <james.kempf@ericsson.com>, IETF ALTO <alto@ietf.org>, irs-discuss@ietf.org
Subject: Re: [alto] [irs-discuss]  About ALTO Vs. BGP-LS
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 03 Aug 2012 18:04:13 -0000

Hi Stefano,

Good post! I added the ALTO mailing list, given the relevance. I hope 
that this is OK cross posting.

First, a few comments on ALTO:

View granularity:

- ALTO currently defines two abstract network topology data structures: 
Network Map and Cost Map(s). More link-state oriented maps (graphs), 
with aggregations and transformations, can be added efficiently to ALTO. 
Some initial efforts are already on the way (e.g., see the graph 
representation proposal at page 9: 
http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf). Hence, 
I see a natural next-step is for ALTO to provide a link-state view to 
external entities.

- It is a good comment on the level of details that ALTO should 
delivery. This is a good question for the ALTO working group and the 
community to discuss. I feel that ALTO should provide multi-levels of 
granularity of views, and we should discuss in the working group.

Pull vs push delivery mechanism:

- There are more discussions on adding the incremental update in ALTO. 
Multiple mechanisms have been discussed. I feel that it is the right 
direction for ALTO.

Now let me understand the deployment model of BGP-LS. Your explanation 
on the issues of acquiring routing state is excellent. Let me start by 
understanding more details on the deployment model of BGP-LS inside a 
network:

- A set N_igp of BGP-LS instances are deployed to collect IGP info. You 
need multiple instances because IGP needs direct connectivity 
(adjacency). Each instance here receives (potentially multiple) IGP 
updates and streams (relays) to an another (remote) entity, which I 
assume is a master BGP-LS instance. So each of these N_igp instances is 
IGP-in and BGP-LS out. This appears to be shown in Figure 1 of 
draft-gredler-idr-ls-distribution.

- A set N_egp of BGP-LS instances are deployed to collect BGP feeds. You 
also may need multiple instances because the network does not want to 
see aggregated states but raw states. These instances will feed to the 
master BGP-LS as well.

- The master BGP-LS aggregates the multiple BGP-LS ins (and maybe some 
direct IGP/EGP ins) and pushes (after policy) to other BGP-LS peers to 
use: for example, an ALTO Server transforms/aggregates the feed to 
generate ALTO views (maps and graphs), and an PCE uses the feed to 
maintain its TED. One could even imagine that ALTO builds a detailed TED 
and feeds to PCE, but this beyond the scope of this discussion. The 
master BGP-LS is BGP-LS in and BGP-LS out. It is also possible that the 
master BGP-LS does not push to any other entities and simply maintains 
an internal DB for others to query.

Do I understand it correctly?

Now, we can take a look at more specifics of BGP-LS.

A first perspective is the semantics of the content. If the objective is 
to solve the aforementioned deployment issue, then an alternative 
solution is to introduce a simple LS update tunneling protocol, where a 
link-state proxy forwards LS messages to a collector. The current design 
of BGP-LS starts with such a feeling (i.e., an NLRI starts with the 
Protocol ID, which indicates it is from IS-IS level 1 IS-IS level 2, 
OSPF, etc).  However, the protocol appears to (try to) go beyond simple 
tunneling and introduces a common LS schema, by converting/filtering 
individual IGP LS messages to some common format. I feel that it can be 
helpful to first specify the schema (LS data model) instead of the 
specific encoding. For example, OSPF specifies LS Age, and this is 
filtered. (Please correct me if I missed it). On the other hand, one can 
think that some Age info can be helpful for one to understand the 
"freshness" of the LS. A problem studied in database is heterogeneous 
databases, to merge multiple data sources (IS-IS, OSPF, etc) to a single 
schema, and there can be many problems. If there is such a study, please 
do share a pointer.

A second perspective is using BGP as the transport. What key features 
from BGP do we really need (yes, weak-typed TLV encoding offers a lot of 
flexibility)? What features of BGP do we not need (e.g., BGP is a 
routing protocol and hence builds in features to handle convergence such 
as dampening)? What may be missing (e.g., a capability of pull or 
filtering of push). I feel that these issues should be discussed. If 
they have already been discussed, please do share a pointer, as I am 
definitely a new comer.

Thanks!

Richard

On 8/2/12 11:54 PM, stefano previdi wrote:
> All,
>
> as co-author of both BGP-LS and ALTO drafts, I'd try to clarify a few
> things:
>
> ALTO has been designed in order to deliver to applications (through
> http/json):
>
> 1. two maps representing the network topology in an abstracted view
>     (or set of views through multiple maps). The map does not include
>     the details of a link-state database and therefore have little use
>     for any element that would need to retrieve from the network the
>     detailed/complete network layer topology, for example: link
>     addresses or link BW resources, etc. IOW: ALTO maps do not have
>     the granularity of a link-state database and ALTO protocol is not
>     designed to deliver such details.
>
> and/or
>
> 2. Ranking services where a client sends a bunch of IP addresses in
>     a message and the ALTO server replies by ranking these addresses
>     based on their topological/network distance (or whatever criteria
>     the ALTO server has been configured for). This is called: Endpoint
>     Cost Service.
>
> When using ALTO maps, and the ALTO protocol being http/pull based,
> there's no such concept of unsolicited routing updates. An ALTO
> client is typically a browser that will pull the maps from an ALTO
> server using http. The ALTO server will make no effort to ensure the
> client has the latest view of the topology (i.e.: It's the role of the
> client to poll for new maps time to time).
>
> Now, in order for an ALTO server to deliver Maps or Ranking services,
> it needs to build some form of topology and in order to achieve this,
> it needs somehow to be fed by either the operator (configuration) or
> to receive dynamically topology information from the network (e.g.:
> ISIS/OSPF/BGP).
>
> Here we had two options:
> 1. ALTO server to implement ISIS/OSPF/BGP and establish IGP adjacencies
>     to ABRs or L1L2 routers in each area so to retrieve the LSDB from
>     each area. In practice I know no SP willing/accepting to open their
>     IGP to an ALTO server. Also, IGP requires direct connectivity
>     (adjacency) so from an operation point of view is complex and not
>     desired.
> 2. Use a database distribution protocol running on top of a reliable
>     transport layer that would allow an ALTO server to connect to a
>     _single_ and _remote_ Route Reflector (i.e.: no need to be directly
>     connected) and grab the whole network topology that will be updated
>     using standard routing protocol mechanisms (i.e.: routing updates)
>     and that would allow the operator to control (through policies and
>     filters) what to distribute and to which server.
>
>     Benefits: single (or dual at most for redundancy) connection for
>     the ALTO server (rather than having a direct adjacency with each
>     ABR) and, from the operator perspective, single point of
>     distribution of network topology (easier to apply policies and
>     control what you deliver). This is what BGP-LS is about.
>
> BGP-LS defines new AFI/SAFI for a new NLRI and attributes that convey
> link-state and, more generically, any type of topology information.
> The draft specifies NLRI and attributes that map whatever you can
> find in a link-state database.
>
> We currently have draft-gredler-idr-ls-distribution in the process
> (hopefully) to be adopted as WG item in IDR WG (so far we 're getting
> good support). We also have 3 implementations of BGP-LS.
>
> Deployment-wise: BGP-LS is not yet deployed, however, we already have
> deployments (I know at least two) where an ALTO server retrieves IP
> prefix information from remote BGP RRs (for ipv4). The same scheme
> will be used once BGP-LS will be deployed (so to say that it is not
> something that we invented after a couple of beers but corresponds to
> requirements for delivering network services to upper layers while
> still controlling efficiently what you distribute, to whom and in
> which form (note that, often, beers are anyway part of the process).
>
> More information here:
> http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02
> http://tools.ietf.org/html/draft-ietf-alto-protocol-12
>
> Hope this helps.
>
> s.
>
>
>
> On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:
>> Tom,
>>
>>> I agree that one of the top work items for this effort should be a
>>> standardized topology function, and one that is accessible via a
>>> non-routing protocol.
>> So if the requirement is to have topology export via non-routing protocol then I think we should seriously revisit or repackage the draft-gredler-idr-ls-distribution-01 which works for for both OSPF and ISIS.
>>
>> However before that let's really understand the requirement why it must be exported via non-routing protocol .... Keep in mind that just to parse BGP UPDATE messages and retrieve interesting pieces out it it requires very little code rather then full BGP implementation.
>>
>> The particular feature I like about draft-gredler-idr-ls-distribution-01 is that it is read-only ;)
>>
>> R.
>>
>>> 	I agree that one of the top work items for this effort should be a
>>> standardized topology function, and one that is accessible via a
>>> non-routing protocol.  While not exactly "low hanging fruit", it is
>>> something that (to me) is a clear work item with clear goals that should
>>> be tackled straight away.
>>>
>>> 	--Tom
>>>
>>>
>>>
>>> On 8/2/12 3:24 PM, "James Kempf" <james.kempf@ericsson.com> wrote:
>>>
>>>> So after seeing part of Alia's talk this morning (I had to leave in the
>>>> middle unfortunately), I'd like to make a couple suggestions. There were
>>>> a lot of ideas presented in the talk, enough for an entire IETF Area. I
>>>> think to make tangible progress, the work needs to be focussed on a small
>>>> subset that would be of immediate interest and usability.
>>>>
>>>> There are a couple areas that suggest themselves, but one that would be
>>>> useful in work that I've been involved in is a standardized format for
>>>> network topology representation and a protocol for exchanging it. The
>>>> Onix OpenFlow controller has a network information base with a
>>>> specialized format for network topology, and every OpenFlow controller
>>>> requires this. Having a standardized way to represent it might foster a
>>>> common topology database package. Another application is network
>>>> management. Every network management system needs some kind of topology
>>>> representation. Finally, though I am not an expert in PCE construction,
>>>> it would seem to me that a PCE would need some kind of topology
>>>> representation in order to perform path calculations. Having a way,for
>>>> example, for the OpenFlow controller and the PCE to exchange topology
>>>> information would be really useful.  I would say to start with physical
>>>> topology because that is fundamental, but make the format flexible enough
>>>> to support
>>>> virtual topology representation.
>>>>
>>>> 			jak
>>>> _______________________________________________
>>>> irs-discuss mailing list
>>>> irs-discuss@ietf.org
>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>> _______________________________________________
>>> irs-discuss mailing list
>>> irs-discuss@ietf.org
>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>
>>>
>> _______________________________________________
>> irs-discuss mailing list
>> irs-discuss@ietf.org
>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>
> _______________________________________________
> irs-discuss mailing list
> irs-discuss@ietf.org
> https://www.ietf.org/mailman/listinfo/irs-discuss


From yry@cs.yale.edu  Fri Aug  3 11:46:24 2012
Return-Path: <yry@cs.yale.edu>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 92B2F21F8E3C; Fri,  3 Aug 2012 11:46:24 -0700 (PDT)
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_33=0.6]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id j6zV8hZmuJQ0; Fri,  3 Aug 2012 11:46:23 -0700 (PDT)
Received: from vm-emlprdomr-02.its.yale.edu (vm-emlprdomr-02.its.yale.edu [130.132.50.143]) by ietfa.amsl.com (Postfix) with ESMTP id 0F14921F8E3A; Fri,  3 Aug 2012 11:46:22 -0700 (PDT)
Received: from Faculty-Supports-MacBook-Pro-2.local ([128.36.46.193]) (authenticated bits=0) by vm-emlprdomr-02.its.yale.edu (8.14.4/8.14.4) with ESMTP id q73IkDeo016975 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NOT); Fri, 3 Aug 2012 14:46:13 -0400
Message-ID: <501C1C74.6040006@cs.yale.edu>
Date: Fri, 03 Aug 2012 11:46:12 -0700
From: "Y. Richard Yang" <yry@cs.yale.edu>
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.7; rv:14.0) Gecko/20120713 Thunderbird/14.0
MIME-Version: 1.0
To: Susan Hares <susan.hares@huawei.com>, idr@ietf.org
References: <CC404D94.2D5D%tnadeau@juniper.net> <501B0D75.4090009@raszuk.net> <7E89A05A-CE4E-4FCF-81AB-8F39B42FBF8E@cisco.com> <501C128D.10809@cs.yale.edu> <728F9B956B2C48439CA9294B1723B14623756237@dfweml509-mbs.china.huawei.com>
In-Reply-To: <728F9B956B2C48439CA9294B1723B14623756237@dfweml509-mbs.china.huawei.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-Scanned-By: MIMEDefang 2.71 on 130.132.50.143
Cc: "irs-discuss@ietf.org" <irs-discuss@ietf.org>, James Kempf <james.kempf@ericsson.com>, IETF ALTO <alto@ietf.org>
Subject: Re: [alto] [irs-discuss] About ALTO Vs. BGP-LS
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 03 Aug 2012 18:46:24 -0000

Hi,

Sue suggested that I forward this email to idr for more discussions.

Thanks.

Richard

On 8/3/12 11:24 AM, Susan Hares wrote:
> Richard:
>
> Could you cross-post this note to the idr@ietf.org list. The BGP-LS draft is being considered for WG adoption. I think this perspective would help the idr WG.
>
> Thank you,
>
> Sue
>
> -----Original Message-----
> From: irs-discuss-bounces@ietf.org [mailto:irs-discuss-bounces@ietf.org] On Behalf Of Y. Richard Yang
> Sent: Friday, August 03, 2012 11:04 AM
> To: stefano previdi
> Cc: Thomas Nadeau; James Kempf; IETF ALTO; robert@raszuk.net; irs-discuss@ietf.org
> Subject: Re: [irs-discuss] About ALTO Vs. BGP-LS
>
> Hi Stefano,
>
> Good post! I added the ALTO mailing list, given the relevance. I hope
> that this is OK cross posting.
>
> First, a few comments on ALTO:
>
> View granularity:
>
> - ALTO currently defines two abstract network topology data structures:
> Network Map and Cost Map(s). More link-state oriented maps (graphs),
> with aggregations and transformations, can be added efficiently to ALTO.
> Some initial efforts are already on the way (e.g., see the graph
> representation proposal at page 9:
> http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf). Hence,
> I see a natural next-step is for ALTO to provide a link-state view to
> external entities.
>
> - It is a good comment on the level of details that ALTO should
> deliver. This is a good question for the ALTO working group and the
> community to discuss. I feel that ALTO should provide multi-levels of
> granularity of views, and we should discuss in the working group.
>
> Pull vs push delivery mechanism:
>
> - There are more discussions on adding the incremental update in ALTO.
> Multiple mechanisms have been discussed. I feel that it is the right
> direction for ALTO.
>
> Now let me understand the deployment model of BGP-LS. Your explanation
> on the issues of acquiring routing state is excellent. Let me start by
> understanding more details on the deployment model of BGP-LS inside a
> network:
>
> - A set N_igp of BGP-LS instances are deployed to collect IGP info. You
> need multiple instances because IGP needs direct connectivity
> (adjacency). Each instance here receives (potentially multiple) IGP
> updates and streams (relays) to an another (remote) entity, which I
> assume is a master BGP-LS instance. So each of these N_igp instances is
> IGP-in and BGP-LS out. This appears to be shown in Figure 1 of
> draft-gredler-idr-ls-distribution.
>
> - A set N_egp of BGP-LS instances are deployed to collect BGP feeds. You
> also may need multiple instances because the network does not want to
> see aggregated states but raw states. These instances will feed to the
> master BGP-LS as well.
>
> - The master BGP-LS aggregates the multiple BGP-LS ins (and maybe some
> direct IGP/EGP ins) and pushes (after policy) to other BGP-LS peers to
> use: for example, an ALTO Server transforms/aggregates the feed to
> generate ALTO views (maps and graphs), and an PCE uses the feed to
> maintain its TED. One could even imagine that ALTO builds a detailed TED
> and feeds to PCE, but this beyond the scope of this discussion. The
> master BGP-LS is BGP-LS in and BGP-LS out. It is also possible that the
> master BGP-LS does not push to any other entities and simply maintains
> an internal DB for others to query.
>
> Do I understand it correctly?
>
> Now, we can take a look at more specifics of BGP-LS.
>
> A first perspective is the semantics of the content. If the objective is
> to solve the aforementioned deployment issue, then an alternative
> solution is to introduce a simple LS update tunneling protocol, where a
> link-state proxy forwards LS messages to a collector. The current design
> of BGP-LS starts with such a feeling (i.e., an NLRI starts with the
> Protocol ID, which indicates it is from IS-IS level 1 IS-IS level 2,
> OSPF, etc).  However, the protocol appears to (try to) go beyond simple
> tunneling and introduces a common LS schema, by converting/filtering
> individual IGP LS messages to some common format. I feel that it can be
> helpful to first specify the schema (LS data model) instead of the
> specific encoding. For example, OSPF specifies LS Age, and this is
> filtered. (Please correct me if I missed it). On the other hand, one can
> think that some Age info can be helpful for one to understand the
> "freshness" of the LS. A problem studied in database is heterogeneous
> databases, to merge multiple data sources (IS-IS, OSPF, etc) to a single
> schema, and there can be many problems. If there is such a study, please
> do share a pointer.
>
> A second perspective is using BGP as the transport. What key features
> from BGP do we really need (yes, weak-typed TLV encoding offers a lot of
> flexibility)? What features of BGP do we not need (e.g., BGP is a
> routing protocol and hence builds in features to handle convergence such
> as dampening)? What may be missing (e.g., a capability of pull or
> filtering of push). I feel that these issues should be discussed. If
> they have already been discussed, please do share a pointer, as I am
> definitely a new comer.
>
> Thanks!
>
> Richard
>
> On 8/2/12 11:54 PM, stefano previdi wrote:
>> All,
>>
>> as co-author of both BGP-LS and ALTO drafts, I'd try to clarify a few
>> things:
>>
>> ALTO has been designed in order to deliver to applications (through
>> http/json):
>>
>> 1. two maps representing the network topology in an abstracted view
>>      (or set of views through multiple maps). The map does not include
>>      the details of a link-state database and therefore have little use
>>      for any element that would need to retrieve from the network the
>>      detailed/complete network layer topology, for example: link
>>      addresses or link BW resources, etc. IOW: ALTO maps do not have
>>      the granularity of a link-state database and ALTO protocol is not
>>      designed to deliver such details.
>>
>> and/or
>>
>> 2. Ranking services where a client sends a bunch of IP addresses in
>>      a message and the ALTO server replies by ranking these addresses
>>      based on their topological/network distance (or whatever criteria
>>      the ALTO server has been configured for). This is called: Endpoint
>>      Cost Service.
>>
>> When using ALTO maps, and the ALTO protocol being http/pull based,
>> there's no such concept of unsolicited routing updates. An ALTO
>> client is typically a browser that will pull the maps from an ALTO
>> server using http. The ALTO server will make no effort to ensure the
>> client has the latest view of the topology (i.e.: It's the role of the
>> client to poll for new maps time to time).
>>
>> Now, in order for an ALTO server to deliver Maps or Ranking services,
>> it needs to build some form of topology and in order to achieve this,
>> it needs somehow to be fed by either the operator (configuration) or
>> to receive dynamically topology information from the network (e.g.:
>> ISIS/OSPF/BGP).
>>
>> Here we had two options:
>> 1. ALTO server to implement ISIS/OSPF/BGP and establish IGP adjacencies
>>      to ABRs or L1L2 routers in each area so to retrieve the LSDB from
>>      each area. In practice I know no SP willing/accepting to open their
>>      IGP to an ALTO server. Also, IGP requires direct connectivity
>>      (adjacency) so from an operation point of view is complex and not
>>      desired.
>> 2. Use a database distribution protocol running on top of a reliable
>>      transport layer that would allow an ALTO server to connect to a
>>      _single_ and _remote_ Route Reflector (i.e.: no need to be directly
>>      connected) and grab the whole network topology that will be updated
>>      using standard routing protocol mechanisms (i.e.: routing updates)
>>      and that would allow the operator to control (through policies and
>>      filters) what to distribute and to which server.
>>
>>      Benefits: single (or dual at most for redundancy) connection for
>>      the ALTO server (rather than having a direct adjacency with each
>>      ABR) and, from the operator perspective, single point of
>>      distribution of network topology (easier to apply policies and
>>      control what you deliver). This is what BGP-LS is about.
>>
>> BGP-LS defines new AFI/SAFI for a new NLRI and attributes that convey
>> link-state and, more generically, any type of topology information.
>> The draft specifies NLRI and attributes that map whatever you can
>> find in a link-state database.
>>
>> We currently have draft-gredler-idr-ls-distribution in the process
>> (hopefully) to be adopted as WG item in IDR WG (so far we 're getting
>> good support). We also have 3 implementations of BGP-LS.
>>
>> Deployment-wise: BGP-LS is not yet deployed, however, we already have
>> deployments (I know at least two) where an ALTO server retrieves IP
>> prefix information from remote BGP RRs (for ipv4). The same scheme
>> will be used once BGP-LS will be deployed (so to say that it is not
>> something that we invented after a couple of beers but corresponds to
>> requirements for delivering network services to upper layers while
>> still controlling efficiently what you distribute, to whom and in
>> which form (note that, often, beers are anyway part of the process).
>>
>> More information here:
>> http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02
>> http://tools.ietf.org/html/draft-ietf-alto-protocol-12
>>
>> Hope this helps.
>>
>> s.
>>
>>
>>
>> On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:
>>> Tom,
>>>
>>>> I agree that one of the top work items for this effort should be a
>>>> standardized topology function, and one that is accessible via a
>>>> non-routing protocol.
>>> So if the requirement is to have topology export via non-routing protocol then I think we should seriously revisit or repackage the draft-gredler-idr-ls-distribution-01 which works for for both OSPF and ISIS.
>>>
>>> However before that let's really understand the requirement why it must be exported via non-routing protocol .... Keep in mind that just to parse BGP UPDATE messages and retrieve interesting pieces out it it requires very little code rather then full BGP implementation.
>>>
>>> The particular feature I like about draft-gredler-idr-ls-distribution-01 is that it is read-only ;)
>>>
>>> R.
>>>
>>>> 	I agree that one of the top work items for this effort should be a
>>>> standardized topology function, and one that is accessible via a
>>>> non-routing protocol.  While not exactly "low hanging fruit", it is
>>>> something that (to me) is a clear work item with clear goals that should
>>>> be tackled straight away.
>>>>
>>>> 	--Tom
>>>>
>>>>
>>>>
>>>> On 8/2/12 3:24 PM, "James Kempf" <james.kempf@ericsson.com> wrote:
>>>>
>>>>> So after seeing part of Alia's talk this morning (I had to leave in the
>>>>> middle unfortunately), I'd like to make a couple suggestions. There were
>>>>> a lot of ideas presented in the talk, enough for an entire IETF Area. I
>>>>> think to make tangible progress, the work needs to be focussed on a small
>>>>> subset that would be of immediate interest and usability.
>>>>>
>>>>> There are a couple areas that suggest themselves, but one that would be
>>>>> useful in work that I've been involved in is a standardized format for
>>>>> network topology representation and a protocol for exchanging it. The
>>>>> Onix OpenFlow controller has a network information base with a
>>>>> specialized format for network topology, and every OpenFlow controller
>>>>> requires this. Having a standardized way to represent it might foster a
>>>>> common topology database package. Another application is network
>>>>> management. Every network management system needs some kind of topology
>>>>> representation. Finally, though I am not an expert in PCE construction,
>>>>> it would seem to me that a PCE would need some kind of topology
>>>>> representation in order to perform path calculations. Having a way,for
>>>>> example, for the OpenFlow controller and the PCE to exchange topology
>>>>> information would be really useful.  I would say to start with physical
>>>>> topology because that is fundamental, but make the format flexible enough
>>>>> to support
>>>>> virtual topology representation.
>>>>>
>>>>> 			jak
>>>>> _______________________________________________
>>>>> irs-discuss mailing list
>>>>> irs-discuss@ietf.org
>>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>> _______________________________________________
>>>> irs-discuss mailing list
>>>> irs-discuss@ietf.org
>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>>
>>>>
>>> _______________________________________________
>>> irs-discuss mailing list
>>> irs-discuss@ietf.org
>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>
>> _______________________________________________
>> irs-discuss mailing list
>> irs-discuss@ietf.org
>> https://www.ietf.org/mailman/listinfo/irs-discuss
> _______________________________________________
> irs-discuss mailing list
> irs-discuss@ietf.org
> https://www.ietf.org/mailman/listinfo/irs-discuss


From yry@cs.yale.edu  Fri Aug  3 13:50:42 2012
Return-Path: <yry@cs.yale.edu>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 0738D21E808F; Fri,  3 Aug 2012 13:50:42 -0700 (PDT)
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=[AWL=0.000,  BAYES_00=-2.599, J_CHICKENPOX_33=0.6]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 30E0G8cduim9; Fri,  3 Aug 2012 13:50:40 -0700 (PDT)
Received: from vm-emlprdomr-02.its.yale.edu (vm-emlprdomr-02.its.yale.edu [130.132.50.143]) by ietfa.amsl.com (Postfix) with ESMTP id E464821E808E; Fri,  3 Aug 2012 13:50:39 -0700 (PDT)
Received: from Faculty-Supports-MacBook-Pro-2.local ([128.36.46.193]) (authenticated bits=0) by vm-emlprdomr-02.its.yale.edu (8.14.4/8.14.4) with ESMTP id q73KoXle019706 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NOT); Fri, 3 Aug 2012 16:50:33 -0400
Message-ID: <501C3999.3080804@cs.yale.edu>
Date: Fri, 03 Aug 2012 13:50:33 -0700
From: "Y. Richard Yang" <yry@cs.yale.edu>
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.7; rv:14.0) Gecko/20120713 Thunderbird/14.0
MIME-Version: 1.0
To: Volker Hilt <volker.hilt@bell-labs.com>
References: <CC404D94.2D5D%tnadeau@juniper.net> <501B0D75.4090009@raszuk.net> <7E89A05A-CE4E-4FCF-81AB-8F39B42FBF8E@cisco.com> <501C128D.10809@cs.yale.edu> <501C2047.1000100@bell-labs.com>
In-Reply-To: <501C2047.1000100@bell-labs.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-Scanned-By: MIMEDefang 2.71 on 130.132.50.143
Cc: idr@ietf.org, irs-discuss@ietf.org, James Kempf <james.kempf@ericsson.com>, IETF ALTO <alto@ietf.org>
Subject: Re: [alto] [irs-discuss] About ALTO Vs. BGP-LS
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 03 Aug 2012 20:50:42 -0000

Hi Volker,

On 8/3/12 12:02 PM, Volker Hilt wrote:
>
> I believe that the capability to provide a simplified view on the 
> network topology to an application is a key feature rather than a bug. 
> Applications that want to have a view on network topology don't need 
> need a fine grained view on the topology in most casts and benefit 
> from having an abstracted view. This will simplify processing for the 
> application and enables providers to control the exposure of details. 
Agreed.

> We've seen some numbers for topology data sizes in the incremental 
> updates presentation in the ALTO meeting, which provide some insights 
> into the amounts of data needed for different topology sizes.
>
> I like the idea of enabling an ALTO server to offer different levels 
> of details. This will enable a server to tailor responses to the 
> specific client. It will add complexity as the ALTO server itself will 
> have to maintain the most complex topology it is offering and will 
> have to keep this topology up to date. But this is an interesting 
> question for discussion in the WG.
>
Glad that you also like the idea of different levels of details of the 
network topology. If the ALTO Server is given a detailed topology 
(constructed from multiple sources, such as routing feed, LSP feed, 
configuration files), we can offer multiple topology 
operators/aggregators to explore the detailed topology, according to 
need and policy. A simple example operator is level (i.e., hierarchy 
such as at the area level), but one might specify other operators (e.g., 
fisheye focus). There are studies on representation of multi-level 
graphs that we can try to take advantage of. This can be a subject for 
the group to explore.

We may need to study/collect use cases for multiple level topology. One 
interesting example immediately coming to mind is the Abstract Node 
concept (specify IP prefix/ASN) used in RSVP-TE.

Cheers,

Richard

> Cheers,
>
> Volker
>
>
>
>
>
>
>
> On 03.08.2012 11:03, Y. Richard Yang wrote:
>> Hi Stefano,
>>
>> Good post! I added the ALTO mailing list, given the relevance. I hope
>> that this is OK cross posting.
>>
>> First, a few comments on ALTO:
>>
>> View granularity:
>>
>> - ALTO currently defines two abstract network topology data structures:
>> Network Map and Cost Map(s). More link-state oriented maps (graphs),
>> with aggregations and transformations, can be added efficiently to ALTO.
>> Some initial efforts are already on the way (e.g., see the graph
>> representation proposal at page 9:
>> http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf). Hence,
>> I see a natural next-step is for ALTO to provide a link-state view to
>> external entities.
>>
>> - It is a good comment on the level of details that ALTO should
>> delivery. This is a good question for the ALTO working group and the
>> community to discuss. I feel that ALTO should provide multi-levels of
>> granularity of views, and we should discuss in the working group.
>>
>> Pull vs push delivery mechanism:
>>
>> - There are more discussions on adding the incremental update in ALTO.
>> Multiple mechanisms have been discussed. I feel that it is the right
>> direction for ALTO.
>>
>> Now let me understand the deployment model of BGP-LS. Your explanation
>> on the issues of acquiring routing state is excellent. Let me start by
>> understanding more details on the deployment model of BGP-LS inside a
>> network:
>>
>> - A set N_igp of BGP-LS instances are deployed to collect IGP info. You
>> need multiple instances because IGP needs direct connectivity
>> (adjacency). Each instance here receives (potentially multiple) IGP
>> updates and streams (relays) to an another (remote) entity, which I
>> assume is a master BGP-LS instance. So each of these N_igp instances is
>> IGP-in and BGP-LS out. This appears to be shown in Figure 1 of
>> draft-gredler-idr-ls-distribution.
>>
>> - A set N_egp of BGP-LS instances are deployed to collect BGP feeds. You
>> also may need multiple instances because the network does not want to
>> see aggregated states but raw states. These instances will feed to the
>> master BGP-LS as well.
>>
>> - The master BGP-LS aggregates the multiple BGP-LS ins (and maybe some
>> direct IGP/EGP ins) and pushes (after policy) to other BGP-LS peers to
>> use: for example, an ALTO Server transforms/aggregates the feed to
>> generate ALTO views (maps and graphs), and an PCE uses the feed to
>> maintain its TED. One could even imagine that ALTO builds a detailed TED
>> and feeds to PCE, but this beyond the scope of this discussion. The
>> master BGP-LS is BGP-LS in and BGP-LS out. It is also possible that the
>> master BGP-LS does not push to any other entities and simply maintains
>> an internal DB for others to query.
>>
>> Do I understand it correctly?
>>
>> Now, we can take a look at more specifics of BGP-LS.
>>
>> A first perspective is the semantics of the content. If the objective is
>> to solve the aforementioned deployment issue, then an alternative
>> solution is to introduce a simple LS update tunneling protocol, where a
>> link-state proxy forwards LS messages to a collector. The current design
>> of BGP-LS starts with such a feeling (i.e., an NLRI starts with the
>> Protocol ID, which indicates it is from IS-IS level 1 IS-IS level 2,
>> OSPF, etc). However, the protocol appears to (try to) go beyond simple
>> tunneling and introduces a common LS schema, by converting/filtering
>> individual IGP LS messages to some common format. I feel that it can be
>> helpful to first specify the schema (LS data model) instead of the
>> specific encoding. For example, OSPF specifies LS Age, and this is
>> filtered. (Please correct me if I missed it). On the other hand, one can
>> think that some Age info can be helpful for one to understand the
>> "freshness" of the LS. A problem studied in database is heterogeneous
>> databases, to merge multiple data sources (IS-IS, OSPF, etc) to a single
>> schema, and there can be many problems. If there is such a study, please
>> do share a pointer.
>>
>> A second perspective is using BGP as the transport. What key features
>> from BGP do we really need (yes, weak-typed TLV encoding offers a lot of
>> flexibility)? What features of BGP do we not need (e.g., BGP is a
>> routing protocol and hence builds in features to handle convergence such
>> as dampening)? What may be missing (e.g., a capability of pull or
>> filtering of push). I feel that these issues should be discussed. If
>> they have already been discussed, please do share a pointer, as I am
>> definitely a new comer.
>>
>> Thanks!
>>
>> Richard
>>
>> On 8/2/12 11:54 PM, stefano previdi wrote:
>>> All,
>>>
>>> as co-author of both BGP-LS and ALTO drafts, I'd try to clarify a few
>>> things:
>>>
>>> ALTO has been designed in order to deliver to applications (through
>>> http/json):
>>>
>>> 1. two maps representing the network topology in an abstracted view
>>> (or set of views through multiple maps). The map does not include
>>> the details of a link-state database and therefore have little use
>>> for any element that would need to retrieve from the network the
>>> detailed/complete network layer topology, for example: link
>>> addresses or link BW resources, etc. IOW: ALTO maps do not have
>>> the granularity of a link-state database and ALTO protocol is not
>>> designed to deliver such details.
>>>
>>> and/or
>>>
>>> 2. Ranking services where a client sends a bunch of IP addresses in
>>> a message and the ALTO server replies by ranking these addresses
>>> based on their topological/network distance (or whatever criteria
>>> the ALTO server has been configured for). This is called: Endpoint
>>> Cost Service.
>>>
>>> When using ALTO maps, and the ALTO protocol being http/pull based,
>>> there's no such concept of unsolicited routing updates. An ALTO
>>> client is typically a browser that will pull the maps from an ALTO
>>> server using http. The ALTO server will make no effort to ensure the
>>> client has the latest view of the topology (i.e.: It's the role of the
>>> client to poll for new maps time to time).
>>>
>>> Now, in order for an ALTO server to deliver Maps or Ranking services,
>>> it needs to build some form of topology and in order to achieve this,
>>> it needs somehow to be fed by either the operator (configuration) or
>>> to receive dynamically topology information from the network (e.g.:
>>> ISIS/OSPF/BGP).
>>>
>>> Here we had two options:
>>> 1. ALTO server to implement ISIS/OSPF/BGP and establish IGP adjacencies
>>> to ABRs or L1L2 routers in each area so to retrieve the LSDB from
>>> each area. In practice I know no SP willing/accepting to open their
>>> IGP to an ALTO server. Also, IGP requires direct connectivity
>>> (adjacency) so from an operation point of view is complex and not
>>> desired.
>>> 2. Use a database distribution protocol running on top of a reliable
>>> transport layer that would allow an ALTO server to connect to a
>>> _single_ and _remote_ Route Reflector (i.e.: no need to be directly
>>> connected) and grab the whole network topology that will be updated
>>> using standard routing protocol mechanisms (i.e.: routing updates)
>>> and that would allow the operator to control (through policies and
>>> filters) what to distribute and to which server.
>>>
>>> Benefits: single (or dual at most for redundancy) connection for
>>> the ALTO server (rather than having a direct adjacency with each
>>> ABR) and, from the operator perspective, single point of
>>> distribution of network topology (easier to apply policies and
>>> control what you deliver). This is what BGP-LS is about.
>>>
>>> BGP-LS defines new AFI/SAFI for a new NLRI and attributes that convey
>>> link-state and, more generically, any type of topology information.
>>> The draft specifies NLRI and attributes that map whatever you can
>>> find in a link-state database.
>>>
>>> We currently have draft-gredler-idr-ls-distribution in the process
>>> (hopefully) to be adopted as WG item in IDR WG (so far we 're getting
>>> good support). We also have 3 implementations of BGP-LS.
>>>
>>> Deployment-wise: BGP-LS is not yet deployed, however, we already have
>>> deployments (I know at least two) where an ALTO server retrieves IP
>>> prefix information from remote BGP RRs (for ipv4). The same scheme
>>> will be used once BGP-LS will be deployed (so to say that it is not
>>> something that we invented after a couple of beers but corresponds to
>>> requirements for delivering network services to upper layers while
>>> still controlling efficiently what you distribute, to whom and in
>>> which form (note that, often, beers are anyway part of the process).
>>>
>>> More information here:
>>> http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02
>>> http://tools.ietf.org/html/draft-ietf-alto-protocol-12
>>>
>>> Hope this helps.
>>>
>>> s.
>>>
>>>
>>>
>>> On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:
>>>> Tom,
>>>>
>>>>> I agree that one of the top work items for this effort should be a
>>>>> standardized topology function, and one that is accessible via a
>>>>> non-routing protocol.
>>>> So if the requirement is to have topology export via non-routing
>>>> protocol then I think we should seriously revisit or repackage the
>>>> draft-gredler-idr-ls-distribution-01 which works for for both OSPF
>>>> and ISIS.
>>>>
>>>> However before that let's really understand the requirement why it
>>>> must be exported via non-routing protocol .... Keep in mind that just
>>>> to parse BGP UPDATE messages and retrieve interesting pieces out it
>>>> it requires very little code rather then full BGP implementation.
>>>>
>>>> The particular feature I like about
>>>> draft-gredler-idr-ls-distribution-01 is that it is read-only ;)
>>>>
>>>> R.
>>>>
>>>>> I agree that one of the top work items for this effort should be a
>>>>> standardized topology function, and one that is accessible via a
>>>>> non-routing protocol. While not exactly "low hanging fruit", it is
>>>>> something that (to me) is a clear work item with clear goals that
>>>>> should
>>>>> be tackled straight away.
>>>>>
>>>>> --Tom
>>>>>
>>>>>
>>>>>
>>>>> On 8/2/12 3:24 PM, "James Kempf" <james.kempf@ericsson.com> wrote:
>>>>>
>>>>>> So after seeing part of Alia's talk this morning (I had to leave in
>>>>>> the
>>>>>> middle unfortunately), I'd like to make a couple suggestions. There
>>>>>> were
>>>>>> a lot of ideas presented in the talk, enough for an entire IETF
>>>>>> Area. I
>>>>>> think to make tangible progress, the work needs to be focussed on a
>>>>>> small
>>>>>> subset that would be of immediate interest and usability.
>>>>>>
>>>>>> There are a couple areas that suggest themselves, but one that
>>>>>> would be
>>>>>> useful in work that I've been involved in is a standardized 
>>>>>> format for
>>>>>> network topology representation and a protocol for exchanging it. 
>>>>>> The
>>>>>> Onix OpenFlow controller has a network information base with a
>>>>>> specialized format for network topology, and every OpenFlow 
>>>>>> controller
>>>>>> requires this. Having a standardized way to represent it might
>>>>>> foster a
>>>>>> common topology database package. Another application is network
>>>>>> management. Every network management system needs some kind of
>>>>>> topology
>>>>>> representation. Finally, though I am not an expert in PCE
>>>>>> construction,
>>>>>> it would seem to me that a PCE would need some kind of topology
>>>>>> representation in order to perform path calculations. Having a 
>>>>>> way,for
>>>>>> example, for the OpenFlow controller and the PCE to exchange 
>>>>>> topology
>>>>>> information would be really useful. I would say to start with 
>>>>>> physical
>>>>>> topology because that is fundamental, but make the format flexible
>>>>>> enough
>>>>>> to support
>>>>>> virtual topology representation.
>>>>>>
>>>>>> jak
>>>>>> _______________________________________________
>>>>>> irs-discuss mailing list
>>>>>> irs-discuss@ietf.org
>>>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>>> _______________________________________________
>>>>> irs-discuss mailing list
>>>>> irs-discuss@ietf.org
>>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>>>
>>>>>
>>>> _______________________________________________
>>>> irs-discuss mailing list
>>>> irs-discuss@ietf.org
>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>>
>>> _______________________________________________
>>> irs-discuss mailing list
>>> irs-discuss@ietf.org
>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>
>> _______________________________________________
>> irs-discuss mailing list
>> irs-discuss@ietf.org
>> https://www.ietf.org/mailman/listinfo/irs-discuss

From susan.hares@huawei.com  Fri Aug  3 11:26:16 2012
Return-Path: <susan.hares@huawei.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D9A7F21F8E4B; Fri,  3 Aug 2012 11:26:16 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -4.871
X-Spam-Level: 
X-Spam-Status: No, score=-4.871 tagged_above=-999 required=5 tests=[AWL=1.128,  BAYES_00=-2.599, J_CHICKENPOX_33=0.6, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id YTwEnOo-vavb; Fri,  3 Aug 2012 11:26:15 -0700 (PDT)
Received: from dfwrgout.huawei.com (dfwrgout.huawei.com [206.16.17.72]) by ietfa.amsl.com (Postfix) with ESMTP id 68DF721F8DEB; Fri,  3 Aug 2012 11:26:15 -0700 (PDT)
Received: from 172.18.9.243 (EHLO dfweml201-edg.china.huawei.com) ([172.18.9.243]) by dfwrg02-dlp.huawei.com (MOS 4.2.3-GA FastPath) with ESMTP id AIM52309; Fri, 03 Aug 2012 10:26:15 -0800 (PST)
Received: from DFWEML404-HUB.china.huawei.com (10.193.5.203) by dfweml201-edg.china.huawei.com (172.18.9.107) with Microsoft SMTP Server (TLS) id 14.1.323.3; Fri, 3 Aug 2012 11:24:27 -0700
Received: from dfweml509-mbs.china.huawei.com ([169.254.12.123]) by dfweml404-hub.china.huawei.com ([10.193.5.203]) with mapi id 14.01.0323.003; Fri, 3 Aug 2012 11:24:31 -0700
From: Susan Hares <susan.hares@huawei.com>
To: "Y. Richard Yang" <yry@cs.yale.edu>, stefano previdi <sprevidi@cisco.com>
Thread-Topic: [irs-discuss] About ALTO Vs. BGP-LS
Thread-Index: AQHNcaJp5nxIKpJMfkeuWhP2sbTgaZdIY11w
Date: Fri, 3 Aug 2012 18:24:30 +0000
Message-ID: <728F9B956B2C48439CA9294B1723B14623756237@dfweml509-mbs.china.huawei.com>
References: <CC404D94.2D5D%tnadeau@juniper.net> <501B0D75.4090009@raszuk.net> <7E89A05A-CE4E-4FCF-81AB-8F39B42FBF8E@cisco.com> <501C128D.10809@cs.yale.edu>
In-Reply-To: <501C128D.10809@cs.yale.edu>
Accept-Language: en-US, zh-CN
Content-Language: en-US
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [10.212.245.217]
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-CFilter-Loop: Reflected
X-Mailman-Approved-At: Mon, 06 Aug 2012 10:41:07 -0700
Cc: James Kempf <james.kempf@ericsson.com>, IETF ALTO <alto@ietf.org>, "irs-discuss@ietf.org" <irs-discuss@ietf.org>
Subject: Re: [alto] [irs-discuss] About ALTO Vs. BGP-LS
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 03 Aug 2012 18:26:17 -0000

Richard:

Could you cross-post this note to the idr@ietf.org list. The BGP-LS draft i=
s being considered for WG adoption. I think this perspective would help the=
 idr WG.=20

Thank you,=20

Sue=20

-----Original Message-----
From: irs-discuss-bounces@ietf.org [mailto:irs-discuss-bounces@ietf.org] On=
 Behalf Of Y. Richard Yang
Sent: Friday, August 03, 2012 11:04 AM
To: stefano previdi
Cc: Thomas Nadeau; James Kempf; IETF ALTO; robert@raszuk.net; irs-discuss@i=
etf.org
Subject: Re: [irs-discuss] About ALTO Vs. BGP-LS

Hi Stefano,

Good post! I added the ALTO mailing list, given the relevance. I hope=20
that this is OK cross posting.

First, a few comments on ALTO:

View granularity:

- ALTO currently defines two abstract network topology data structures:=20
Network Map and Cost Map(s). More link-state oriented maps (graphs),=20
with aggregations and transformations, can be added efficiently to ALTO.=20
Some initial efforts are already on the way (e.g., see the graph=20
representation proposal at page 9:=20
http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf). Hence,=20
I see a natural next-step is for ALTO to provide a link-state view to=20
external entities.

- It is a good comment on the level of details that ALTO should=20
delivery. This is a good question for the ALTO working group and the=20
community to discuss. I feel that ALTO should provide multi-levels of=20
granularity of views, and we should discuss in the working group.

Pull vs push delivery mechanism:

- There are more discussions on adding the incremental update in ALTO.=20
Multiple mechanisms have been discussed. I feel that it is the right=20
direction for ALTO.

Now let me understand the deployment model of BGP-LS. Your explanation=20
on the issues of acquiring routing state is excellent. Let me start by=20
understanding more details on the deployment model of BGP-LS inside a=20
network:

- A set N_igp of BGP-LS instances are deployed to collect IGP info. You=20
need multiple instances because IGP needs direct connectivity=20
(adjacency). Each instance here receives (potentially multiple) IGP=20
updates and streams (relays) to an another (remote) entity, which I=20
assume is a master BGP-LS instance. So each of these N_igp instances is=20
IGP-in and BGP-LS out. This appears to be shown in Figure 1 of=20
draft-gredler-idr-ls-distribution.

- A set N_egp of BGP-LS instances are deployed to collect BGP feeds. You=20
also may need multiple instances because the network does not want to=20
see aggregated states but raw states. These instances will feed to the=20
master BGP-LS as well.

- The master BGP-LS aggregates the multiple BGP-LS ins (and maybe some=20
direct IGP/EGP ins) and pushes (after policy) to other BGP-LS peers to=20
use: for example, an ALTO Server transforms/aggregates the feed to=20
generate ALTO views (maps and graphs), and an PCE uses the feed to=20
maintain its TED. One could even imagine that ALTO builds a detailed TED=20
and feeds to PCE, but this beyond the scope of this discussion. The=20
master BGP-LS is BGP-LS in and BGP-LS out. It is also possible that the=20
master BGP-LS does not push to any other entities and simply maintains=20
an internal DB for others to query.

Do I understand it correctly?

Now, we can take a look at more specifics of BGP-LS.

A first perspective is the semantics of the content. If the objective is=20
to solve the aforementioned deployment issue, then an alternative=20
solution is to introduce a simple LS update tunneling protocol, where a=20
link-state proxy forwards LS messages to a collector. The current design=20
of BGP-LS starts with such a feeling (i.e., an NLRI starts with the=20
Protocol ID, which indicates it is from IS-IS level 1 IS-IS level 2,=20
OSPF, etc).  However, the protocol appears to (try to) go beyond simple=20
tunneling and introduces a common LS schema, by converting/filtering=20
individual IGP LS messages to some common format. I feel that it can be=20
helpful to first specify the schema (LS data model) instead of the=20
specific encoding. For example, OSPF specifies LS Age, and this is=20
filtered. (Please correct me if I missed it). On the other hand, one can=20
think that some Age info can be helpful for one to understand the=20
"freshness" of the LS. A problem studied in database is heterogeneous=20
databases, to merge multiple data sources (IS-IS, OSPF, etc) to a single=20
schema, and there can be many problems. If there is such a study, please=20
do share a pointer.

A second perspective is using BGP as the transport. What key features=20
from BGP do we really need (yes, weak-typed TLV encoding offers a lot of=20
flexibility)? What features of BGP do we not need (e.g., BGP is a=20
routing protocol and hence builds in features to handle convergence such=20
as dampening)? What may be missing (e.g., a capability of pull or=20
filtering of push). I feel that these issues should be discussed. If=20
they have already been discussed, please do share a pointer, as I am=20
definitely a new comer.

Thanks!

Richard

On 8/2/12 11:54 PM, stefano previdi wrote:
> All,
>
> as co-author of both BGP-LS and ALTO drafts, I'd try to clarify a few
> things:
>
> ALTO has been designed in order to deliver to applications (through
> http/json):
>
> 1. two maps representing the network topology in an abstracted view
>     (or set of views through multiple maps). The map does not include
>     the details of a link-state database and therefore have little use
>     for any element that would need to retrieve from the network the
>     detailed/complete network layer topology, for example: link
>     addresses or link BW resources, etc. IOW: ALTO maps do not have
>     the granularity of a link-state database and ALTO protocol is not
>     designed to deliver such details.
>
> and/or
>
> 2. Ranking services where a client sends a bunch of IP addresses in
>     a message and the ALTO server replies by ranking these addresses
>     based on their topological/network distance (or whatever criteria
>     the ALTO server has been configured for). This is called: Endpoint
>     Cost Service.
>
> When using ALTO maps, and the ALTO protocol being http/pull based,
> there's no such concept of unsolicited routing updates. An ALTO
> client is typically a browser that will pull the maps from an ALTO
> server using http. The ALTO server will make no effort to ensure the
> client has the latest view of the topology (i.e.: It's the role of the
> client to poll for new maps time to time).
>
> Now, in order for an ALTO server to deliver Maps or Ranking services,
> it needs to build some form of topology and in order to achieve this,
> it needs somehow to be fed by either the operator (configuration) or
> to receive dynamically topology information from the network (e.g.:
> ISIS/OSPF/BGP).
>
> Here we had two options:
> 1. ALTO server to implement ISIS/OSPF/BGP and establish IGP adjacencies
>     to ABRs or L1L2 routers in each area so to retrieve the LSDB from
>     each area. In practice I know no SP willing/accepting to open their
>     IGP to an ALTO server. Also, IGP requires direct connectivity
>     (adjacency) so from an operation point of view is complex and not
>     desired.
> 2. Use a database distribution protocol running on top of a reliable
>     transport layer that would allow an ALTO server to connect to a
>     _single_ and _remote_ Route Reflector (i.e.: no need to be directly
>     connected) and grab the whole network topology that will be updated
>     using standard routing protocol mechanisms (i.e.: routing updates)
>     and that would allow the operator to control (through policies and
>     filters) what to distribute and to which server.
>
>     Benefits: single (or dual at most for redundancy) connection for
>     the ALTO server (rather than having a direct adjacency with each
>     ABR) and, from the operator perspective, single point of
>     distribution of network topology (easier to apply policies and
>     control what you deliver). This is what BGP-LS is about.
>
> BGP-LS defines new AFI/SAFI for a new NLRI and attributes that convey
> link-state and, more generically, any type of topology information.
> The draft specifies NLRI and attributes that map whatever you can
> find in a link-state database.
>
> We currently have draft-gredler-idr-ls-distribution in the process
> (hopefully) to be adopted as WG item in IDR WG (so far we 're getting
> good support). We also have 3 implementations of BGP-LS.
>
> Deployment-wise: BGP-LS is not yet deployed, however, we already have
> deployments (I know at least two) where an ALTO server retrieves IP
> prefix information from remote BGP RRs (for ipv4). The same scheme
> will be used once BGP-LS will be deployed (so to say that it is not
> something that we invented after a couple of beers but corresponds to
> requirements for delivering network services to upper layers while
> still controlling efficiently what you distribute, to whom and in
> which form (note that, often, beers are anyway part of the process).
>
> More information here:
> http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02
> http://tools.ietf.org/html/draft-ietf-alto-protocol-12
>
> Hope this helps.
>
> s.
>
>
>
> On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:
>> Tom,
>>
>>> I agree that one of the top work items for this effort should be a
>>> standardized topology function, and one that is accessible via a
>>> non-routing protocol.
>> So if the requirement is to have topology export via non-routing protoco=
l then I think we should seriously revisit or repackage the draft-gredler-i=
dr-ls-distribution-01 which works for for both OSPF and ISIS.
>>
>> However before that let's really understand the requirement why it must =
be exported via non-routing protocol .... Keep in mind that just to parse B=
GP UPDATE messages and retrieve interesting pieces out it it requires very =
little code rather then full BGP implementation.
>>
>> The particular feature I like about draft-gredler-idr-ls-distribution-01=
 is that it is read-only ;)
>>
>> R.
>>
>>> 	I agree that one of the top work items for this effort should be a
>>> standardized topology function, and one that is accessible via a
>>> non-routing protocol.  While not exactly "low hanging fruit", it is
>>> something that (to me) is a clear work item with clear goals that shoul=
d
>>> be tackled straight away.
>>>
>>> 	--Tom
>>>
>>>
>>>
>>> On 8/2/12 3:24 PM, "James Kempf" <james.kempf@ericsson.com> wrote:
>>>
>>>> So after seeing part of Alia's talk this morning (I had to leave in th=
e
>>>> middle unfortunately), I'd like to make a couple suggestions. There we=
re
>>>> a lot of ideas presented in the talk, enough for an entire IETF Area. =
I
>>>> think to make tangible progress, the work needs to be focussed on a sm=
all
>>>> subset that would be of immediate interest and usability.
>>>>
>>>> There are a couple areas that suggest themselves, but one that would b=
e
>>>> useful in work that I've been involved in is a standardized format for
>>>> network topology representation and a protocol for exchanging it. The
>>>> Onix OpenFlow controller has a network information base with a
>>>> specialized format for network topology, and every OpenFlow controller
>>>> requires this. Having a standardized way to represent it might foster =
a
>>>> common topology database package. Another application is network
>>>> management. Every network management system needs some kind of topolog=
y
>>>> representation. Finally, though I am not an expert in PCE construction=
,
>>>> it would seem to me that a PCE would need some kind of topology
>>>> representation in order to perform path calculations. Having a way,for
>>>> example, for the OpenFlow controller and the PCE to exchange topology
>>>> information would be really useful.  I would say to start with physica=
l
>>>> topology because that is fundamental, but make the format flexible eno=
ugh
>>>> to support
>>>> virtual topology representation.
>>>>
>>>> 			jak
>>>> _______________________________________________
>>>> irs-discuss mailing list
>>>> irs-discuss@ietf.org
>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>> _______________________________________________
>>> irs-discuss mailing list
>>> irs-discuss@ietf.org
>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>
>>>
>> _______________________________________________
>> irs-discuss mailing list
>> irs-discuss@ietf.org
>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>
> _______________________________________________
> irs-discuss mailing list
> irs-discuss@ietf.org
> https://www.ietf.org/mailman/listinfo/irs-discuss

_______________________________________________
irs-discuss mailing list
irs-discuss@ietf.org
https://www.ietf.org/mailman/listinfo/irs-discuss

From volker.hilt@bell-labs.com  Fri Aug  3 12:02:50 2012
Return-Path: <volker.hilt@bell-labs.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 2FD3A21F8D92; Fri,  3 Aug 2012 12:02:50 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -9.149
X-Spam-Level: 
X-Spam-Status: No, score=-9.149 tagged_above=-999 required=5 tests=[AWL=0.500,  BAYES_00=-2.599, HELO_EQ_FR=0.35, J_CHICKENPOX_33=0.6, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id UTvxvKrgdBt2; Fri,  3 Aug 2012 12:02:48 -0700 (PDT)
Received: from smail2.alcatel.fr (smail2.alcatel.fr [64.208.49.57]) by ietfa.amsl.com (Postfix) with ESMTP id 9AB0721F8D69; Fri,  3 Aug 2012 12:02:48 -0700 (PDT)
Received: from FRMRSSXCHHUB03.dc-m.alcatel-lucent.com (FRMRSSXCHHUB03.dc-m.alcatel-lucent.com [135.120.45.63]) by smail2.alcatel.fr (8.14.3/8.14.3/ICT) with ESMTP id q73J2dWc022752 (version=TLSv1/SSLv3 cipher=RC4-MD5 bits=128 verify=NOT); Fri, 3 Aug 2012 21:02:41 +0200
Received: from US70UWXCHHUB01.zam.alcatel-lucent.com (135.5.2.48) by FRMRSSXCHHUB03.dc-m.alcatel-lucent.com (135.120.45.63) with Microsoft SMTP Server (TLS) id 8.3.213.0; Fri, 3 Aug 2012 21:02:39 +0200
Received: from [135.244.144.209] (135.5.27.12) by US70UWXCHHUB01.zam.alcatel-lucent.com (135.5.2.48) with Microsoft SMTP Server (TLS) id 14.2.247.3; Fri, 3 Aug 2012 15:02:38 -0400
Message-ID: <501C2047.1000100@bell-labs.com>
Date: Fri, 3 Aug 2012 12:02:31 -0700
From: Volker Hilt <volker.hilt@bell-labs.com>
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:11.0) Gecko/20120327 Thunderbird/11.0.1
MIME-Version: 1.0
To: "Y. Richard Yang" <yry@cs.yale.edu>
References: <CC404D94.2D5D%tnadeau@juniper.net> <501B0D75.4090009@raszuk.net> <7E89A05A-CE4E-4FCF-81AB-8F39B42FBF8E@cisco.com> <501C128D.10809@cs.yale.edu>
In-Reply-To: <501C128D.10809@cs.yale.edu>
Content-Type: text/plain; charset="ISO-8859-1"; format=flowed
Content-Transfer-Encoding: 7bit
X-Originating-IP: [135.5.27.12]
X-Scanned-By: MIMEDefang 2.69 on 155.132.188.80
X-Mailman-Approved-At: Mon, 06 Aug 2012 10:41:07 -0700
Cc: idr@ietf.org, irs-discuss@ietf.org, James Kempf <james.kempf@ericsson.com>, IETF ALTO <alto@ietf.org>
Subject: Re: [alto] [irs-discuss] About ALTO Vs. BGP-LS
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 03 Aug 2012 19:02:50 -0000

Stefano, Richard,

great discussion.

I believe that the capability to provide a simplified view on the 
network topology to an application is a key feature rather than a bug. 
Applications that want to have a view on network topology don't need 
need a fine grained view on the topology in most casts and benefit from 
having an abstracted view. This will simplify processing for the 
application and enables providers to control the exposure of details. 
We've seen some numbers for topology data sizes in the incremental 
updates presentation in the ALTO meeting, which provide some insights 
into the amounts of data needed for different topology sizes.

I like the idea of enabling an ALTO server to offer different levels of 
details. This will enable a server to tailor responses to the specific 
client. It will add complexity as the ALTO server itself will have to 
maintain the most complex topology it is offering and will have to keep 
this topology up to date. But this is an interesting question for 
discussion in the WG.

Cheers,

Volker







On 03.08.2012 11:03, Y. Richard Yang wrote:
> Hi Stefano,
>
> Good post! I added the ALTO mailing list, given the relevance. I hope
> that this is OK cross posting.
>
> First, a few comments on ALTO:
>
> View granularity:
>
> - ALTO currently defines two abstract network topology data structures:
> Network Map and Cost Map(s). More link-state oriented maps (graphs),
> with aggregations and transformations, can be added efficiently to ALTO.
> Some initial efforts are already on the way (e.g., see the graph
> representation proposal at page 9:
> http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf). Hence,
> I see a natural next-step is for ALTO to provide a link-state view to
> external entities.
>
> - It is a good comment on the level of details that ALTO should
> delivery. This is a good question for the ALTO working group and the
> community to discuss. I feel that ALTO should provide multi-levels of
> granularity of views, and we should discuss in the working group.
>
> Pull vs push delivery mechanism:
>
> - There are more discussions on adding the incremental update in ALTO.
> Multiple mechanisms have been discussed. I feel that it is the right
> direction for ALTO.
>
> Now let me understand the deployment model of BGP-LS. Your explanation
> on the issues of acquiring routing state is excellent. Let me start by
> understanding more details on the deployment model of BGP-LS inside a
> network:
>
> - A set N_igp of BGP-LS instances are deployed to collect IGP info. You
> need multiple instances because IGP needs direct connectivity
> (adjacency). Each instance here receives (potentially multiple) IGP
> updates and streams (relays) to an another (remote) entity, which I
> assume is a master BGP-LS instance. So each of these N_igp instances is
> IGP-in and BGP-LS out. This appears to be shown in Figure 1 of
> draft-gredler-idr-ls-distribution.
>
> - A set N_egp of BGP-LS instances are deployed to collect BGP feeds. You
> also may need multiple instances because the network does not want to
> see aggregated states but raw states. These instances will feed to the
> master BGP-LS as well.
>
> - The master BGP-LS aggregates the multiple BGP-LS ins (and maybe some
> direct IGP/EGP ins) and pushes (after policy) to other BGP-LS peers to
> use: for example, an ALTO Server transforms/aggregates the feed to
> generate ALTO views (maps and graphs), and an PCE uses the feed to
> maintain its TED. One could even imagine that ALTO builds a detailed TED
> and feeds to PCE, but this beyond the scope of this discussion. The
> master BGP-LS is BGP-LS in and BGP-LS out. It is also possible that the
> master BGP-LS does not push to any other entities and simply maintains
> an internal DB for others to query.
>
> Do I understand it correctly?
>
> Now, we can take a look at more specifics of BGP-LS.
>
> A first perspective is the semantics of the content. If the objective is
> to solve the aforementioned deployment issue, then an alternative
> solution is to introduce a simple LS update tunneling protocol, where a
> link-state proxy forwards LS messages to a collector. The current design
> of BGP-LS starts with such a feeling (i.e., an NLRI starts with the
> Protocol ID, which indicates it is from IS-IS level 1 IS-IS level 2,
> OSPF, etc). However, the protocol appears to (try to) go beyond simple
> tunneling and introduces a common LS schema, by converting/filtering
> individual IGP LS messages to some common format. I feel that it can be
> helpful to first specify the schema (LS data model) instead of the
> specific encoding. For example, OSPF specifies LS Age, and this is
> filtered. (Please correct me if I missed it). On the other hand, one can
> think that some Age info can be helpful for one to understand the
> "freshness" of the LS. A problem studied in database is heterogeneous
> databases, to merge multiple data sources (IS-IS, OSPF, etc) to a single
> schema, and there can be many problems. If there is such a study, please
> do share a pointer.
>
> A second perspective is using BGP as the transport. What key features
> from BGP do we really need (yes, weak-typed TLV encoding offers a lot of
> flexibility)? What features of BGP do we not need (e.g., BGP is a
> routing protocol and hence builds in features to handle convergence such
> as dampening)? What may be missing (e.g., a capability of pull or
> filtering of push). I feel that these issues should be discussed. If
> they have already been discussed, please do share a pointer, as I am
> definitely a new comer.
>
> Thanks!
>
> Richard
>
> On 8/2/12 11:54 PM, stefano previdi wrote:
>> All,
>>
>> as co-author of both BGP-LS and ALTO drafts, I'd try to clarify a few
>> things:
>>
>> ALTO has been designed in order to deliver to applications (through
>> http/json):
>>
>> 1. two maps representing the network topology in an abstracted view
>> (or set of views through multiple maps). The map does not include
>> the details of a link-state database and therefore have little use
>> for any element that would need to retrieve from the network the
>> detailed/complete network layer topology, for example: link
>> addresses or link BW resources, etc. IOW: ALTO maps do not have
>> the granularity of a link-state database and ALTO protocol is not
>> designed to deliver such details.
>>
>> and/or
>>
>> 2. Ranking services where a client sends a bunch of IP addresses in
>> a message and the ALTO server replies by ranking these addresses
>> based on their topological/network distance (or whatever criteria
>> the ALTO server has been configured for). This is called: Endpoint
>> Cost Service.
>>
>> When using ALTO maps, and the ALTO protocol being http/pull based,
>> there's no such concept of unsolicited routing updates. An ALTO
>> client is typically a browser that will pull the maps from an ALTO
>> server using http. The ALTO server will make no effort to ensure the
>> client has the latest view of the topology (i.e.: It's the role of the
>> client to poll for new maps time to time).
>>
>> Now, in order for an ALTO server to deliver Maps or Ranking services,
>> it needs to build some form of topology and in order to achieve this,
>> it needs somehow to be fed by either the operator (configuration) or
>> to receive dynamically topology information from the network (e.g.:
>> ISIS/OSPF/BGP).
>>
>> Here we had two options:
>> 1. ALTO server to implement ISIS/OSPF/BGP and establish IGP adjacencies
>> to ABRs or L1L2 routers in each area so to retrieve the LSDB from
>> each area. In practice I know no SP willing/accepting to open their
>> IGP to an ALTO server. Also, IGP requires direct connectivity
>> (adjacency) so from an operation point of view is complex and not
>> desired.
>> 2. Use a database distribution protocol running on top of a reliable
>> transport layer that would allow an ALTO server to connect to a
>> _single_ and _remote_ Route Reflector (i.e.: no need to be directly
>> connected) and grab the whole network topology that will be updated
>> using standard routing protocol mechanisms (i.e.: routing updates)
>> and that would allow the operator to control (through policies and
>> filters) what to distribute and to which server.
>>
>> Benefits: single (or dual at most for redundancy) connection for
>> the ALTO server (rather than having a direct adjacency with each
>> ABR) and, from the operator perspective, single point of
>> distribution of network topology (easier to apply policies and
>> control what you deliver). This is what BGP-LS is about.
>>
>> BGP-LS defines new AFI/SAFI for a new NLRI and attributes that convey
>> link-state and, more generically, any type of topology information.
>> The draft specifies NLRI and attributes that map whatever you can
>> find in a link-state database.
>>
>> We currently have draft-gredler-idr-ls-distribution in the process
>> (hopefully) to be adopted as WG item in IDR WG (so far we 're getting
>> good support). We also have 3 implementations of BGP-LS.
>>
>> Deployment-wise: BGP-LS is not yet deployed, however, we already have
>> deployments (I know at least two) where an ALTO server retrieves IP
>> prefix information from remote BGP RRs (for ipv4). The same scheme
>> will be used once BGP-LS will be deployed (so to say that it is not
>> something that we invented after a couple of beers but corresponds to
>> requirements for delivering network services to upper layers while
>> still controlling efficiently what you distribute, to whom and in
>> which form (note that, often, beers are anyway part of the process).
>>
>> More information here:
>> http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02
>> http://tools.ietf.org/html/draft-ietf-alto-protocol-12
>>
>> Hope this helps.
>>
>> s.
>>
>>
>>
>> On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:
>>> Tom,
>>>
>>>> I agree that one of the top work items for this effort should be a
>>>> standardized topology function, and one that is accessible via a
>>>> non-routing protocol.
>>> So if the requirement is to have topology export via non-routing
>>> protocol then I think we should seriously revisit or repackage the
>>> draft-gredler-idr-ls-distribution-01 which works for for both OSPF
>>> and ISIS.
>>>
>>> However before that let's really understand the requirement why it
>>> must be exported via non-routing protocol .... Keep in mind that just
>>> to parse BGP UPDATE messages and retrieve interesting pieces out it
>>> it requires very little code rather then full BGP implementation.
>>>
>>> The particular feature I like about
>>> draft-gredler-idr-ls-distribution-01 is that it is read-only ;)
>>>
>>> R.
>>>
>>>> I agree that one of the top work items for this effort should be a
>>>> standardized topology function, and one that is accessible via a
>>>> non-routing protocol. While not exactly "low hanging fruit", it is
>>>> something that (to me) is a clear work item with clear goals that
>>>> should
>>>> be tackled straight away.
>>>>
>>>> --Tom
>>>>
>>>>
>>>>
>>>> On 8/2/12 3:24 PM, "James Kempf" <james.kempf@ericsson.com> wrote:
>>>>
>>>>> So after seeing part of Alia's talk this morning (I had to leave in
>>>>> the
>>>>> middle unfortunately), I'd like to make a couple suggestions. There
>>>>> were
>>>>> a lot of ideas presented in the talk, enough for an entire IETF
>>>>> Area. I
>>>>> think to make tangible progress, the work needs to be focussed on a
>>>>> small
>>>>> subset that would be of immediate interest and usability.
>>>>>
>>>>> There are a couple areas that suggest themselves, but one that
>>>>> would be
>>>>> useful in work that I've been involved in is a standardized format for
>>>>> network topology representation and a protocol for exchanging it. The
>>>>> Onix OpenFlow controller has a network information base with a
>>>>> specialized format for network topology, and every OpenFlow controller
>>>>> requires this. Having a standardized way to represent it might
>>>>> foster a
>>>>> common topology database package. Another application is network
>>>>> management. Every network management system needs some kind of
>>>>> topology
>>>>> representation. Finally, though I am not an expert in PCE
>>>>> construction,
>>>>> it would seem to me that a PCE would need some kind of topology
>>>>> representation in order to perform path calculations. Having a way,for
>>>>> example, for the OpenFlow controller and the PCE to exchange topology
>>>>> information would be really useful. I would say to start with physical
>>>>> topology because that is fundamental, but make the format flexible
>>>>> enough
>>>>> to support
>>>>> virtual topology representation.
>>>>>
>>>>> jak
>>>>> _______________________________________________
>>>>> irs-discuss mailing list
>>>>> irs-discuss@ietf.org
>>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>> _______________________________________________
>>>> irs-discuss mailing list
>>>> irs-discuss@ietf.org
>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>>
>>>>
>>> _______________________________________________
>>> irs-discuss mailing list
>>> irs-discuss@ietf.org
>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>
>> _______________________________________________
>> irs-discuss mailing list
>> irs-discuss@ietf.org
>> https://www.ietf.org/mailman/listinfo/irs-discuss
>
> _______________________________________________
> irs-discuss mailing list
> irs-discuss@ietf.org
> https://www.ietf.org/mailman/listinfo/irs-discuss

From akatlas@gmail.com  Fri Aug  3 16:57:10 2012
Return-Path: <akatlas@gmail.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id B4FB021F8DAB; Fri,  3 Aug 2012 16:57:10 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.45
X-Spam-Level: 
X-Spam-Status: No, score=-2.45 tagged_above=-999 required=5 tests=[AWL=-1.118,  BAYES_00=-2.599, HTML_MESSAGE=0.001, J_CHICKENPOX_33=0.6, RCVD_IN_DNSWL_LOW=-1, SARE_HTML_USL_OBFU=1.666]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id S3Qx6q1RsQr0; Fri,  3 Aug 2012 16:57:08 -0700 (PDT)
Received: from mail-gg0-f172.google.com (mail-gg0-f172.google.com [209.85.161.172]) by ietfa.amsl.com (Postfix) with ESMTP id 6759B21F8D20; Fri,  3 Aug 2012 16:57:08 -0700 (PDT)
Received: by ggnc4 with SMTP id c4so1563428ggn.31 for <multiple recipients>; Fri, 03 Aug 2012 16:57:08 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=I8UdNKB+tzrwSGda7KkdpJXpW88WSKVjDkdJcVcvY2w=; b=YIjSNYBBgAf7zAQaSuGho7VbNYaYYqXio2Qv7NGq+RpBFxR4Ah03J/ldmw6Wy3vfnS vnmDbtn/JPxX7M+1owgPTurquZ8WbF3cciogRJbCf+NxY7YbfRj0KNpLS+wVOQELJdyl 3Y8lx6i5hC3R3YkRp5E8rAmbC/R+ijGSY4Iqdur8FRQT342+clWYZQyG8G+XRFQMjmtD 21G//YNMqlTn+SMO84vo0Pg7+IHWgiwkK2ibGKvgFXEbm3k7raDLKZUp0NtqdZ3wxPCU ifbQCTvLeL+Dz2qdkRhxPP4cVuUU8HIXdqdeHG0XpCVbdK4JX+ok2B1nmMLaxAE4Ll8o cogg==
MIME-Version: 1.0
Received: by 10.42.61.16 with SMTP id s16mr4770807ich.7.1344038227496; Fri, 03 Aug 2012 16:57:07 -0700 (PDT)
Received: by 10.50.34.169 with HTTP; Fri, 3 Aug 2012 16:57:07 -0700 (PDT)
Received: by 10.50.34.169 with HTTP; Fri, 3 Aug 2012 16:57:07 -0700 (PDT)
In-Reply-To: <501C3999.3080804@cs.yale.edu>
References: <CC404D94.2D5D%tnadeau@juniper.net> <501B0D75.4090009@raszuk.net> <7E89A05A-CE4E-4FCF-81AB-8F39B42FBF8E@cisco.com> <501C128D.10809@cs.yale.edu> <501C2047.1000100@bell-labs.com> <501C3999.3080804@cs.yale.edu>
Date: Fri, 3 Aug 2012 19:57:07 -0400
Message-ID: <CAG4d1rdo3BuEukb9F0aR=wOkKFEM9_esaJVJvn+e_UndQ3jELg@mail.gmail.com>
From: Alia Atlas <akatlas@gmail.com>
To: "Y. Richard Yang" <yry@cs.yale.edu>
Content-Type: multipart/alternative; boundary=20cf301cc2e21d4fbf04c6654988
X-Mailman-Approved-At: Mon, 06 Aug 2012 10:41:07 -0700
Cc: idr@ietf.org, Volker Hilt <volker.hilt@bell-labs.com>, irs-discuss@ietf.org, James Kempf <james.kempf@ericsson.com>, IETF ALTO <alto@ietf.org>
Subject: Re: [alto] [irs-discuss] About ALTO Vs. BGP-LS
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 03 Aug 2012 23:57:10 -0000

--20cf301cc2e21d4fbf04c6654988
Content-Type: text/plain; charset=ISO-8859-1

I strongly agree that we need to start collecting the use-cases for
multiple abstraction levels as well as physical/virtual/private.

Alia
On Aug 3, 2012 4:51 PM, "Y. Richard Yang" <yry@cs.yale.edu> wrote:

> Hi Volker,
>
> On 8/3/12 12:02 PM, Volker Hilt wrote:
>
>>
>> I believe that the capability to provide a simplified view on the network
>> topology to an application is a key feature rather than a bug. Applications
>> that want to have a view on network topology don't need need a fine grained
>> view on the topology in most casts and benefit from having an abstracted
>> view. This will simplify processing for the application and enables
>> providers to control the exposure of details.
>>
> Agreed.
>
>  We've seen some numbers for topology data sizes in the incremental
>> updates presentation in the ALTO meeting, which provide some insights into
>> the amounts of data needed for different topology sizes.
>>
>> I like the idea of enabling an ALTO server to offer different levels of
>> details. This will enable a server to tailor responses to the specific
>> client. It will add complexity as the ALTO server itself will have to
>> maintain the most complex topology it is offering and will have to keep
>> this topology up to date. But this is an interesting question for
>> discussion in the WG.
>>
>>  Glad that you also like the idea of different levels of details of the
> network topology. If the ALTO Server is given a detailed topology
> (constructed from multiple sources, such as routing feed, LSP feed,
> configuration files), we can offer multiple topology operators/aggregators
> to explore the detailed topology, according to need and policy. A simple
> example operator is level (i.e., hierarchy such as at the area level), but
> one might specify other operators (e.g., fisheye focus). There are studies
> on representation of multi-level graphs that we can try to take advantage
> of. This can be a subject for the group to explore.
>
> We may need to study/collect use cases for multiple level topology. One
> interesting example immediately coming to mind is the Abstract Node concept
> (specify IP prefix/ASN) used in RSVP-TE.
>
> Cheers,
>
> Richard
>
>  Cheers,
>>
>> Volker
>>
>>
>>
>>
>>
>>
>>
>> On 03.08.2012 11:03, Y. Richard Yang wrote:
>>
>>> Hi Stefano,
>>>
>>> Good post! I added the ALTO mailing list, given the relevance. I hope
>>> that this is OK cross posting.
>>>
>>> First, a few comments on ALTO:
>>>
>>> View granularity:
>>>
>>> - ALTO currently defines two abstract network topology data structures:
>>> Network Map and Cost Map(s). More link-state oriented maps (graphs),
>>> with aggregations and transformations, can be added efficiently to ALTO.
>>> Some initial efforts are already on the way (e.g., see the graph
>>> representation proposal at page 9:
>>> http://www.ietf.org/**proceedings/84/slides/slides-**84-alto-1.pdf<http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf>).
>>> Hence,
>>> I see a natural next-step is for ALTO to provide a link-state view to
>>> external entities.
>>>
>>> - It is a good comment on the level of details that ALTO should
>>> delivery. This is a good question for the ALTO working group and the
>>> community to discuss. I feel that ALTO should provide multi-levels of
>>> granularity of views, and we should discuss in the working group.
>>>
>>> Pull vs push delivery mechanism:
>>>
>>> - There are more discussions on adding the incremental update in ALTO.
>>> Multiple mechanisms have been discussed. I feel that it is the right
>>> direction for ALTO.
>>>
>>> Now let me understand the deployment model of BGP-LS. Your explanation
>>> on the issues of acquiring routing state is excellent. Let me start by
>>> understanding more details on the deployment model of BGP-LS inside a
>>> network:
>>>
>>> - A set N_igp of BGP-LS instances are deployed to collect IGP info. You
>>> need multiple instances because IGP needs direct connectivity
>>> (adjacency). Each instance here receives (potentially multiple) IGP
>>> updates and streams (relays) to an another (remote) entity, which I
>>> assume is a master BGP-LS instance. So each of these N_igp instances is
>>> IGP-in and BGP-LS out. This appears to be shown in Figure 1 of
>>> draft-gredler-idr-ls-**distribution.
>>>
>>> - A set N_egp of BGP-LS instances are deployed to collect BGP feeds. You
>>> also may need multiple instances because the network does not want to
>>> see aggregated states but raw states. These instances will feed to the
>>> master BGP-LS as well.
>>>
>>> - The master BGP-LS aggregates the multiple BGP-LS ins (and maybe some
>>> direct IGP/EGP ins) and pushes (after policy) to other BGP-LS peers to
>>> use: for example, an ALTO Server transforms/aggregates the feed to
>>> generate ALTO views (maps and graphs), and an PCE uses the feed to
>>> maintain its TED. One could even imagine that ALTO builds a detailed TED
>>> and feeds to PCE, but this beyond the scope of this discussion. The
>>> master BGP-LS is BGP-LS in and BGP-LS out. It is also possible that the
>>> master BGP-LS does not push to any other entities and simply maintains
>>> an internal DB for others to query.
>>>
>>> Do I understand it correctly?
>>>
>>> Now, we can take a look at more specifics of BGP-LS.
>>>
>>> A first perspective is the semantics of the content. If the objective is
>>> to solve the aforementioned deployment issue, then an alternative
>>> solution is to introduce a simple LS update tunneling protocol, where a
>>> link-state proxy forwards LS messages to a collector. The current design
>>> of BGP-LS starts with such a feeling (i.e., an NLRI starts with the
>>> Protocol ID, which indicates it is from IS-IS level 1 IS-IS level 2,
>>> OSPF, etc). However, the protocol appears to (try to) go beyond simple
>>> tunneling and introduces a common LS schema, by converting/filtering
>>> individual IGP LS messages to some common format. I feel that it can be
>>> helpful to first specify the schema (LS data model) instead of the
>>> specific encoding. For example, OSPF specifies LS Age, and this is
>>> filtered. (Please correct me if I missed it). On the other hand, one can
>>> think that some Age info can be helpful for one to understand the
>>> "freshness" of the LS. A problem studied in database is heterogeneous
>>> databases, to merge multiple data sources (IS-IS, OSPF, etc) to a single
>>> schema, and there can be many problems. If there is such a study, please
>>> do share a pointer.
>>>
>>> A second perspective is using BGP as the transport. What key features
>>> from BGP do we really need (yes, weak-typed TLV encoding offers a lot of
>>> flexibility)? What features of BGP do we not need (e.g., BGP is a
>>> routing protocol and hence builds in features to handle convergence such
>>> as dampening)? What may be missing (e.g., a capability of pull or
>>> filtering of push). I feel that these issues should be discussed. If
>>> they have already been discussed, please do share a pointer, as I am
>>> definitely a new comer.
>>>
>>> Thanks!
>>>
>>> Richard
>>>
>>> On 8/2/12 11:54 PM, stefano previdi wrote:
>>>
>>>> All,
>>>>
>>>> as co-author of both BGP-LS and ALTO drafts, I'd try to clarify a few
>>>> things:
>>>>
>>>> ALTO has been designed in order to deliver to applications (through
>>>> http/json):
>>>>
>>>> 1. two maps representing the network topology in an abstracted view
>>>> (or set of views through multiple maps). The map does not include
>>>> the details of a link-state database and therefore have little use
>>>> for any element that would need to retrieve from the network the
>>>> detailed/complete network layer topology, for example: link
>>>> addresses or link BW resources, etc. IOW: ALTO maps do not have
>>>> the granularity of a link-state database and ALTO protocol is not
>>>> designed to deliver such details.
>>>>
>>>> and/or
>>>>
>>>> 2. Ranking services where a client sends a bunch of IP addresses in
>>>> a message and the ALTO server replies by ranking these addresses
>>>> based on their topological/network distance (or whatever criteria
>>>> the ALTO server has been configured for). This is called: Endpoint
>>>> Cost Service.
>>>>
>>>> When using ALTO maps, and the ALTO protocol being http/pull based,
>>>> there's no such concept of unsolicited routing updates. An ALTO
>>>> client is typically a browser that will pull the maps from an ALTO
>>>> server using http. The ALTO server will make no effort to ensure the
>>>> client has the latest view of the topology (i.e.: It's the role of the
>>>> client to poll for new maps time to time).
>>>>
>>>> Now, in order for an ALTO server to deliver Maps or Ranking services,
>>>> it needs to build some form of topology and in order to achieve this,
>>>> it needs somehow to be fed by either the operator (configuration) or
>>>> to receive dynamically topology information from the network (e.g.:
>>>> ISIS/OSPF/BGP).
>>>>
>>>> Here we had two options:
>>>> 1. ALTO server to implement ISIS/OSPF/BGP and establish IGP adjacencies
>>>> to ABRs or L1L2 routers in each area so to retrieve the LSDB from
>>>> each area. In practice I know no SP willing/accepting to open their
>>>> IGP to an ALTO server. Also, IGP requires direct connectivity
>>>> (adjacency) so from an operation point of view is complex and not
>>>> desired.
>>>> 2. Use a database distribution protocol running on top of a reliable
>>>> transport layer that would allow an ALTO server to connect to a
>>>> _single_ and _remote_ Route Reflector (i.e.: no need to be directly
>>>> connected) and grab the whole network topology that will be updated
>>>> using standard routing protocol mechanisms (i.e.: routing updates)
>>>> and that would allow the operator to control (through policies and
>>>> filters) what to distribute and to which server.
>>>>
>>>> Benefits: single (or dual at most for redundancy) connection for
>>>> the ALTO server (rather than having a direct adjacency with each
>>>> ABR) and, from the operator perspective, single point of
>>>> distribution of network topology (easier to apply policies and
>>>> control what you deliver). This is what BGP-LS is about.
>>>>
>>>> BGP-LS defines new AFI/SAFI for a new NLRI and attributes that convey
>>>> link-state and, more generically, any type of topology information.
>>>> The draft specifies NLRI and attributes that map whatever you can
>>>> find in a link-state database.
>>>>
>>>> We currently have draft-gredler-idr-ls-**distribution in the process
>>>> (hopefully) to be adopted as WG item in IDR WG (so far we 're getting
>>>> good support). We also have 3 implementations of BGP-LS.
>>>>
>>>> Deployment-wise: BGP-LS is not yet deployed, however, we already have
>>>> deployments (I know at least two) where an ALTO server retrieves IP
>>>> prefix information from remote BGP RRs (for ipv4). The same scheme
>>>> will be used once BGP-LS will be deployed (so to say that it is not
>>>> something that we invented after a couple of beers but corresponds to
>>>> requirements for delivering network services to upper layers while
>>>> still controlling efficiently what you distribute, to whom and in
>>>> which form (note that, often, beers are anyway part of the process).
>>>>
>>>> More information here:
>>>> http://tools.ietf.org/html/**draft-gredler-idr-ls-**distribution-02<http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02>
>>>> http://tools.ietf.org/html/**draft-ietf-alto-protocol-12<http://tools.ietf.org/html/draft-ietf-alto-protocol-12>
>>>>
>>>> Hope this helps.
>>>>
>>>> s.
>>>>
>>>>
>>>>
>>>> On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:
>>>>
>>>>> Tom,
>>>>>
>>>>>  I agree that one of the top work items for this effort should be a
>>>>>> standardized topology function, and one that is accessible via a
>>>>>> non-routing protocol.
>>>>>>
>>>>> So if the requirement is to have topology export via non-routing
>>>>> protocol then I think we should seriously revisit or repackage the
>>>>> draft-gredler-idr-ls-**distribution-01 which works for for both OSPF
>>>>> and ISIS.
>>>>>
>>>>> However before that let's really understand the requirement why it
>>>>> must be exported via non-routing protocol .... Keep in mind that just
>>>>> to parse BGP UPDATE messages and retrieve interesting pieces out it
>>>>> it requires very little code rather then full BGP implementation.
>>>>>
>>>>> The particular feature I like about
>>>>> draft-gredler-idr-ls-**distribution-01 is that it is read-only ;)
>>>>>
>>>>> R.
>>>>>
>>>>>  I agree that one of the top work items for this effort should be a
>>>>>> standardized topology function, and one that is accessible via a
>>>>>> non-routing protocol. While not exactly "low hanging fruit", it is
>>>>>> something that (to me) is a clear work item with clear goals that
>>>>>> should
>>>>>> be tackled straight away.
>>>>>>
>>>>>> --Tom
>>>>>>
>>>>>>
>>>>>>
>>>>>> On 8/2/12 3:24 PM, "James Kempf" <james.kempf@ericsson.com> wrote:
>>>>>>
>>>>>>  So after seeing part of Alia's talk this morning (I had to leave in
>>>>>>> the
>>>>>>> middle unfortunately), I'd like to make a couple suggestions. There
>>>>>>> were
>>>>>>> a lot of ideas presented in the talk, enough for an entire IETF
>>>>>>> Area. I
>>>>>>> think to make tangible progress, the work needs to be focussed on a
>>>>>>> small
>>>>>>> subset that would be of immediate interest and usability.
>>>>>>>
>>>>>>> There are a couple areas that suggest themselves, but one that
>>>>>>> would be
>>>>>>> useful in work that I've been involved in is a standardized format
>>>>>>> for
>>>>>>> network topology representation and a protocol for exchanging it. The
>>>>>>> Onix OpenFlow controller has a network information base with a
>>>>>>> specialized format for network topology, and every OpenFlow
>>>>>>> controller
>>>>>>> requires this. Having a standardized way to represent it might
>>>>>>> foster a
>>>>>>> common topology database package. Another application is network
>>>>>>> management. Every network management system needs some kind of
>>>>>>> topology
>>>>>>> representation. Finally, though I am not an expert in PCE
>>>>>>> construction,
>>>>>>> it would seem to me that a PCE would need some kind of topology
>>>>>>> representation in order to perform path calculations. Having a
>>>>>>> way,for
>>>>>>> example, for the OpenFlow controller and the PCE to exchange topology
>>>>>>> information would be really useful. I would say to start with
>>>>>>> physical
>>>>>>> topology because that is fundamental, but make the format flexible
>>>>>>> enough
>>>>>>> to support
>>>>>>> virtual topology representation.
>>>>>>>
>>>>>>> jak
>>>>>>> ______________________________**_________________
>>>>>>> irs-discuss mailing list
>>>>>>> irs-discuss@ietf.org
>>>>>>> https://www.ietf.org/mailman/**listinfo/irs-discuss<https://www.ietf.org/mailman/listinfo/irs-discuss>
>>>>>>>
>>>>>> ______________________________**_________________
>>>>>> irs-discuss mailing list
>>>>>> irs-discuss@ietf.org
>>>>>> https://www.ietf.org/mailman/**listinfo/irs-discuss<https://www.ietf.org/mailman/listinfo/irs-discuss>
>>>>>>
>>>>>>
>>>>>>  ______________________________**_________________
>>>>> irs-discuss mailing list
>>>>> irs-discuss@ietf.org
>>>>> https://www.ietf.org/mailman/**listinfo/irs-discuss<https://www.ietf.org/mailman/listinfo/irs-discuss>
>>>>>
>>>>>  ______________________________**_________________
>>>> irs-discuss mailing list
>>>> irs-discuss@ietf.org
>>>> https://www.ietf.org/mailman/**listinfo/irs-discuss<https://www.ietf.org/mailman/listinfo/irs-discuss>
>>>>
>>>
>>> ______________________________**_________________
>>> irs-discuss mailing list
>>> irs-discuss@ietf.org
>>> https://www.ietf.org/mailman/**listinfo/irs-discuss<https://www.ietf.org/mailman/listinfo/irs-discuss>
>>>
>> ______________________________**_________________
> irs-discuss mailing list
> irs-discuss@ietf.org
> https://www.ietf.org/mailman/**listinfo/irs-discuss<https://www.ietf.org/mailman/listinfo/irs-discuss>
>

--20cf301cc2e21d4fbf04c6654988
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<p>I strongly agree that we need to start collecting the use-cases for mult=
iple abstraction levels as well as physical/virtual/private.</p>
<p>Alia</p>
<div class=3D"gmail_quote">On Aug 3, 2012 4:51 PM, &quot;Y. Richard Yang&qu=
ot; &lt;<a href=3D"mailto:yry@cs.yale.edu">yry@cs.yale.edu</a>&gt; wrote:<b=
r type=3D"attribution"><blockquote class=3D"gmail_quote" style=3D"margin:0 =
0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Hi Volker,<br>
<br>
On 8/3/12 12:02 PM, Volker Hilt wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<br>
I believe that the capability to provide a simplified view on the network t=
opology to an application is a key feature rather than a bug. Applications =
that want to have a view on network topology don&#39;t need need a fine gra=
ined view on the topology in most casts and benefit from having an abstract=
ed view. This will simplify processing for the application and enables prov=
iders to control the exposure of details. <br>

</blockquote>
Agreed.<br>
<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
We&#39;ve seen some numbers for topology data sizes in the incremental upda=
tes presentation in the ALTO meeting, which provide some insights into the =
amounts of data needed for different topology sizes.<br>
<br>
I like the idea of enabling an ALTO server to offer different levels of det=
ails. This will enable a server to tailor responses to the specific client.=
 It will add complexity as the ALTO server itself will have to maintain the=
 most complex topology it is offering and will have to keep this topology u=
p to date. But this is an interesting question for discussion in the WG.<br=
>

<br>
</blockquote>
Glad that you also like the idea of different levels of details of the netw=
ork topology. If the ALTO Server is given a detailed topology (constructed =
from multiple sources, such as routing feed, LSP feed, configuration files)=
, we can offer multiple topology operators/aggregators to explore the detai=
led topology, according to need and policy. A simple example operator is le=
vel (i.e., hierarchy such as at the area level), but one might specify othe=
r operators (e.g., fisheye focus). There are studies on representation of m=
ulti-level graphs that we can try to take advantage of. This can be a subje=
ct for the group to explore.<br>

<br>
We may need to study/collect use cases for multiple level topology. One int=
eresting example immediately coming to mind is the Abstract Node concept (s=
pecify IP prefix/ASN) used in RSVP-TE.<br>
<br>
Cheers,<br>
<br>
Richard<br>
<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
Cheers,<br>
<br>
Volker<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
On 03.08.2012 11:03, Y. Richard Yang wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
Hi Stefano,<br>
<br>
Good post! I added the ALTO mailing list, given the relevance. I hope<br>
that this is OK cross posting.<br>
<br>
First, a few comments on ALTO:<br>
<br>
View granularity:<br>
<br>
- ALTO currently defines two abstract network topology data structures:<br>
Network Map and Cost Map(s). More link-state oriented maps (graphs),<br>
with aggregations and transformations, can be added efficiently to ALTO.<br=
>
Some initial efforts are already on the way (e.g., see the graph<br>
representation proposal at page 9:<br>
<a href=3D"http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf" =
target=3D"_blank">http://www.ietf.org/<u></u>proceedings/84/slides/slides-<=
u></u>84-alto-1.pdf</a>). Hence,<br>
I see a natural next-step is for ALTO to provide a link-state view to<br>
external entities.<br>
<br>
- It is a good comment on the level of details that ALTO should<br>
delivery. This is a good question for the ALTO working group and the<br>
community to discuss. I feel that ALTO should provide multi-levels of<br>
granularity of views, and we should discuss in the working group.<br>
<br>
Pull vs push delivery mechanism:<br>
<br>
- There are more discussions on adding the incremental update in ALTO.<br>
Multiple mechanisms have been discussed. I feel that it is the right<br>
direction for ALTO.<br>
<br>
Now let me understand the deployment model of BGP-LS. Your explanation<br>
on the issues of acquiring routing state is excellent. Let me start by<br>
understanding more details on the deployment model of BGP-LS inside a<br>
network:<br>
<br>
- A set N_igp of BGP-LS instances are deployed to collect IGP info. You<br>
need multiple instances because IGP needs direct connectivity<br>
(adjacency). Each instance here receives (potentially multiple) IGP<br>
updates and streams (relays) to an another (remote) entity, which I<br>
assume is a master BGP-LS instance. So each of these N_igp instances is<br>
IGP-in and BGP-LS out. This appears to be shown in Figure 1 of<br>
draft-gredler-idr-ls-<u></u>distribution.<br>
<br>
- A set N_egp of BGP-LS instances are deployed to collect BGP feeds. You<br=
>
also may need multiple instances because the network does not want to<br>
see aggregated states but raw states. These instances will feed to the<br>
master BGP-LS as well.<br>
<br>
- The master BGP-LS aggregates the multiple BGP-LS ins (and maybe some<br>
direct IGP/EGP ins) and pushes (after policy) to other BGP-LS peers to<br>
use: for example, an ALTO Server transforms/aggregates the feed to<br>
generate ALTO views (maps and graphs), and an PCE uses the feed to<br>
maintain its TED. One could even imagine that ALTO builds a detailed TED<br=
>
and feeds to PCE, but this beyond the scope of this discussion. The<br>
master BGP-LS is BGP-LS in and BGP-LS out. It is also possible that the<br>
master BGP-LS does not push to any other entities and simply maintains<br>
an internal DB for others to query.<br>
<br>
Do I understand it correctly?<br>
<br>
Now, we can take a look at more specifics of BGP-LS.<br>
<br>
A first perspective is the semantics of the content. If the objective is<br=
>
to solve the aforementioned deployment issue, then an alternative<br>
solution is to introduce a simple LS update tunneling protocol, where a<br>
link-state proxy forwards LS messages to a collector. The current design<br=
>
of BGP-LS starts with such a feeling (i.e., an NLRI starts with the<br>
Protocol ID, which indicates it is from IS-IS level 1 IS-IS level 2,<br>
OSPF, etc). However, the protocol appears to (try to) go beyond simple<br>
tunneling and introduces a common LS schema, by converting/filtering<br>
individual IGP LS messages to some common format. I feel that it can be<br>
helpful to first specify the schema (LS data model) instead of the<br>
specific encoding. For example, OSPF specifies LS Age, and this is<br>
filtered. (Please correct me if I missed it). On the other hand, one can<br=
>
think that some Age info can be helpful for one to understand the<br>
&quot;freshness&quot; of the LS. A problem studied in database is heterogen=
eous<br>
databases, to merge multiple data sources (IS-IS, OSPF, etc) to a single<br=
>
schema, and there can be many problems. If there is such a study, please<br=
>
do share a pointer.<br>
<br>
A second perspective is using BGP as the transport. What key features<br>
from BGP do we really need (yes, weak-typed TLV encoding offers a lot of<br=
>
flexibility)? What features of BGP do we not need (e.g., BGP is a<br>
routing protocol and hence builds in features to handle convergence such<br=
>
as dampening)? What may be missing (e.g., a capability of pull or<br>
filtering of push). I feel that these issues should be discussed. If<br>
they have already been discussed, please do share a pointer, as I am<br>
definitely a new comer.<br>
<br>
Thanks!<br>
<br>
Richard<br>
<br>
On 8/2/12 11:54 PM, stefano previdi wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
All,<br>
<br>
as co-author of both BGP-LS and ALTO drafts, I&#39;d try to clarify a few<b=
r>
things:<br>
<br>
ALTO has been designed in order to deliver to applications (through<br>
http/json):<br>
<br>
1. two maps representing the network topology in an abstracted view<br>
(or set of views through multiple maps). The map does not include<br>
the details of a link-state database and therefore have little use<br>
for any element that would need to retrieve from the network the<br>
detailed/complete network layer topology, for example: link<br>
addresses or link BW resources, etc. IOW: ALTO maps do not have<br>
the granularity of a link-state database and ALTO protocol is not<br>
designed to deliver such details.<br>
<br>
and/or<br>
<br>
2. Ranking services where a client sends a bunch of IP addresses in<br>
a message and the ALTO server replies by ranking these addresses<br>
based on their topological/network distance (or whatever criteria<br>
the ALTO server has been configured for). This is called: Endpoint<br>
Cost Service.<br>
<br>
When using ALTO maps, and the ALTO protocol being http/pull based,<br>
there&#39;s no such concept of unsolicited routing updates. An ALTO<br>
client is typically a browser that will pull the maps from an ALTO<br>
server using http. The ALTO server will make no effort to ensure the<br>
client has the latest view of the topology (i.e.: It&#39;s the role of the<=
br>
client to poll for new maps time to time).<br>
<br>
Now, in order for an ALTO server to deliver Maps or Ranking services,<br>
it needs to build some form of topology and in order to achieve this,<br>
it needs somehow to be fed by either the operator (configuration) or<br>
to receive dynamically topology information from the network (e.g.:<br>
ISIS/OSPF/BGP).<br>
<br>
Here we had two options:<br>
1. ALTO server to implement ISIS/OSPF/BGP and establish IGP adjacencies<br>
to ABRs or L1L2 routers in each area so to retrieve the LSDB from<br>
each area. In practice I know no SP willing/accepting to open their<br>
IGP to an ALTO server. Also, IGP requires direct connectivity<br>
(adjacency) so from an operation point of view is complex and not<br>
desired.<br>
2. Use a database distribution protocol running on top of a reliable<br>
transport layer that would allow an ALTO server to connect to a<br>
_single_ and _remote_ Route Reflector (i.e.: no need to be directly<br>
connected) and grab the whole network topology that will be updated<br>
using standard routing protocol mechanisms (i.e.: routing updates)<br>
and that would allow the operator to control (through policies and<br>
filters) what to distribute and to which server.<br>
<br>
Benefits: single (or dual at most for redundancy) connection for<br>
the ALTO server (rather than having a direct adjacency with each<br>
ABR) and, from the operator perspective, single point of<br>
distribution of network topology (easier to apply policies and<br>
control what you deliver). This is what BGP-LS is about.<br>
<br>
BGP-LS defines new AFI/SAFI for a new NLRI and attributes that convey<br>
link-state and, more generically, any type of topology information.<br>
The draft specifies NLRI and attributes that map whatever you can<br>
find in a link-state database.<br>
<br>
We currently have draft-gredler-idr-ls-<u></u>distribution in the process<b=
r>
(hopefully) to be adopted as WG item in IDR WG (so far we &#39;re getting<b=
r>
good support). We also have 3 implementations of BGP-LS.<br>
<br>
Deployment-wise: BGP-LS is not yet deployed, however, we already have<br>
deployments (I know at least two) where an ALTO server retrieves IP<br>
prefix information from remote BGP RRs (for ipv4). The same scheme<br>
will be used once BGP-LS will be deployed (so to say that it is not<br>
something that we invented after a couple of beers but corresponds to<br>
requirements for delivering network services to upper layers while<br>
still controlling efficiently what you distribute, to whom and in<br>
which form (note that, often, beers are anyway part of the process).<br>
<br>
More information here:<br>
<a href=3D"http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02"=
 target=3D"_blank">http://tools.ietf.org/html/<u></u>draft-gredler-idr-ls-<=
u></u>distribution-02</a><br>
<a href=3D"http://tools.ietf.org/html/draft-ietf-alto-protocol-12" target=
=3D"_blank">http://tools.ietf.org/html/<u></u>draft-ietf-alto-protocol-12</=
a><br>
<br>
Hope this helps.<br>
<br>
s.<br>
<br>
<br>
<br>
On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
Tom,<br>
<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
I agree that one of the top work items for this effort should be a<br>
standardized topology function, and one that is accessible via a<br>
non-routing protocol.<br>
</blockquote>
So if the requirement is to have topology export via non-routing<br>
protocol then I think we should seriously revisit or repackage the<br>
draft-gredler-idr-ls-<u></u>distribution-01 which works for for both OSPF<b=
r>
and ISIS.<br>
<br>
However before that let&#39;s really understand the requirement why it<br>
must be exported via non-routing protocol .... Keep in mind that just<br>
to parse BGP UPDATE messages and retrieve interesting pieces out it<br>
it requires very little code rather then full BGP implementation.<br>
<br>
The particular feature I like about<br>
draft-gredler-idr-ls-<u></u>distribution-01 is that it is read-only ;)<br>
<br>
R.<br>
<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
I agree that one of the top work items for this effort should be a<br>
standardized topology function, and one that is accessible via a<br>
non-routing protocol. While not exactly &quot;low hanging fruit&quot;, it i=
s<br>
something that (to me) is a clear work item with clear goals that<br>
should<br>
be tackled straight away.<br>
<br>
--Tom<br>
<br>
<br>
<br>
On 8/2/12 3:24 PM, &quot;James Kempf&quot; &lt;<a href=3D"mailto:james.kemp=
f@ericsson.com" target=3D"_blank">james.kempf@ericsson.com</a>&gt; wrote:<b=
r>
<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
So after seeing part of Alia&#39;s talk this morning (I had to leave in<br>
the<br>
middle unfortunately), I&#39;d like to make a couple suggestions. There<br>
were<br>
a lot of ideas presented in the talk, enough for an entire IETF<br>
Area. I<br>
think to make tangible progress, the work needs to be focussed on a<br>
small<br>
subset that would be of immediate interest and usability.<br>
<br>
There are a couple areas that suggest themselves, but one that<br>
would be<br>
useful in work that I&#39;ve been involved in is a standardized format for<=
br>
network topology representation and a protocol for exchanging it. The<br>
Onix OpenFlow controller has a network information base with a<br>
specialized format for network topology, and every OpenFlow controller<br>
requires this. Having a standardized way to represent it might<br>
foster a<br>
common topology database package. Another application is network<br>
management. Every network management system needs some kind of<br>
topology<br>
representation. Finally, though I am not an expert in PCE<br>
construction,<br>
it would seem to me that a PCE would need some kind of topology<br>
representation in order to perform path calculations. Having a way,for<br>
example, for the OpenFlow controller and the PCE to exchange topology<br>
information would be really useful. I would say to start with physical<br>
topology because that is fundamental, but make the format flexible<br>
enough<br>
to support<br>
virtual topology representation.<br>
<br>
jak<br>
______________________________<u></u>_________________<br>
irs-discuss mailing list<br>
<a href=3D"mailto:irs-discuss@ietf.org" target=3D"_blank">irs-discuss@ietf.=
org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/irs-discuss" target=3D"_bl=
ank">https://www.ietf.org/mailman/<u></u>listinfo/irs-discuss</a><br>
</blockquote>
______________________________<u></u>_________________<br>
irs-discuss mailing list<br>
<a href=3D"mailto:irs-discuss@ietf.org" target=3D"_blank">irs-discuss@ietf.=
org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/irs-discuss" target=3D"_bl=
ank">https://www.ietf.org/mailman/<u></u>listinfo/irs-discuss</a><br>
<br>
<br>
</blockquote>
______________________________<u></u>_________________<br>
irs-discuss mailing list<br>
<a href=3D"mailto:irs-discuss@ietf.org" target=3D"_blank">irs-discuss@ietf.=
org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/irs-discuss" target=3D"_bl=
ank">https://www.ietf.org/mailman/<u></u>listinfo/irs-discuss</a><br>
<br>
</blockquote>
______________________________<u></u>_________________<br>
irs-discuss mailing list<br>
<a href=3D"mailto:irs-discuss@ietf.org" target=3D"_blank">irs-discuss@ietf.=
org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/irs-discuss" target=3D"_bl=
ank">https://www.ietf.org/mailman/<u></u>listinfo/irs-discuss</a><br>
</blockquote>
<br>
______________________________<u></u>_________________<br>
irs-discuss mailing list<br>
<a href=3D"mailto:irs-discuss@ietf.org" target=3D"_blank">irs-discuss@ietf.=
org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/irs-discuss" target=3D"_bl=
ank">https://www.ietf.org/mailman/<u></u>listinfo/irs-discuss</a><br>
</blockquote></blockquote>
______________________________<u></u>_________________<br>
irs-discuss mailing list<br>
<a href=3D"mailto:irs-discuss@ietf.org" target=3D"_blank">irs-discuss@ietf.=
org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/irs-discuss" target=3D"_bl=
ank">https://www.ietf.org/mailman/<u></u>listinfo/irs-discuss</a><br>
</blockquote></div>

--20cf301cc2e21d4fbf04c6654988--

From gregb@grotto-networking.com  Mon Aug  6 10:49:46 2012
Return-Path: <gregb@grotto-networking.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id DE07411E80A2; Mon,  6 Aug 2012 10:49:46 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.092
X-Spam-Level: 
X-Spam-Status: No, score=-2.092 tagged_above=-999 required=5 tests=[AWL=-0.094, BAYES_00=-2.599, HTML_MESSAGE=0.001, J_CHICKENPOX_33=0.6]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id aMXjSk6SWPJS; Mon,  6 Aug 2012 10:49:44 -0700 (PDT)
Received: from smtp.webfaction.com (mail6.webfaction.com [74.55.86.74]) by ietfa.amsl.com (Postfix) with ESMTP id 4336111E808A; Mon,  6 Aug 2012 10:49:44 -0700 (PDT)
Received: from [192.168.0.124] (c-67-170-243-110.hsd1.ca.comcast.net [67.170.243.110]) by smtp.webfaction.com (Postfix) with ESMTP id 030F22105505; Mon,  6 Aug 2012 12:49:42 -0500 (CDT)
Message-ID: <502003B2.5010605@grotto-networking.com>
Date: Mon, 06 Aug 2012 10:49:38 -0700
From: Greg Bernstein <gregb@grotto-networking.com>
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:14.0) Gecko/20120713 Thunderbird/14.0
MIME-Version: 1.0
To: "Y. Richard Yang" <yry@cs.yale.edu>
References: <CC404D94.2D5D%tnadeau@juniper.net> <501B0D75.4090009@raszuk.net> <7E89A05A-CE4E-4FCF-81AB-8F39B42FBF8E@cisco.com> <501C128D.10809@cs.yale.edu> <728F9B956B2C48439CA9294B1723B14623756237@dfweml509-mbs.china.huawei.com> <501C1C74.6040006@cs.yale.edu>
In-Reply-To: <501C1C74.6040006@cs.yale.edu>
Content-Type: multipart/alternative; boundary="------------090209040004060208090600"
Cc: idr@ietf.org, James Kempf <james.kempf@ericsson.com>, IETF ALTO <alto@ietf.org>, "irs-discuss@ietf.org" <irs-discuss@ietf.org>
Subject: Re: [alto] [irs-discuss] About ALTO Vs. BGP-LS
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 06 Aug 2012 17:49:47 -0000

This is a multi-part message in MIME format.
--------------090209040004060208090600
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Hi Richard, Stefano, et. al. thanks for the explanations and pointers to 
the BGP-LS draft. This helps avoid any confusion that BGP-LS is in 
competition with ALTO.

BGP-LS (http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02) 
could be used to feed topology/path information to ALTO servers or path 
computation elements (PCE).  As part of that process it would collect 
link state information from IGPs and possibly abstract it.

One partial point of confusion for me as a non-BGP speaker ;-)   was in 
the BGP-LS abstract which stated "In a number of environments, */a 
component external to a network/* is called upon to perform computations 
based on the network topology and current state of the connections 
within the network, including traffic engineering information."  or  
"This document describes a mechanism by which links state and traffic 
engineering information can be collected from networks and *shared with 
external components* using the BGP routing protocol."
I always considered PCE clients and servers as being within the network 
(working with privileged or sensitive network information) and an ALTO 
server as abstracting that information for use by network clients.  I 
guess by "external" to the network meaning was external to the "data plane".

I also noticed some parallels with our ALTO large bandwidth 
use-case/modeling work (section 6 of 
http://tools.ietf.org/html/draft-bernstein-alto-large-bandwidth-cases-02) and 
BGP-LS. In particular we both seem to want graph and path abstractions.  
For path abstractions (aggregation) we needed (due to bandwidth 
constraints) model a path as one or more abstract links with appropriate 
properties.  It seems a similar situation would hold when trying to 
aggregate paths with SRLG properties. These similarities make sense 
since both drafts are concerned with sharing information about the 
network in an aggregated/abstract form.

Best Regards
Greg B.


On 8/3/2012 11:46 AM, Y. Richard Yang wrote:
> Hi,
>
> Sue suggested that I forward this email to idr for more discussions.
>
> Thanks.
>
> Richard
>
> On 8/3/12 11:24 AM, Susan Hares wrote:
>> Richard:
>>
>> Could you cross-post this note to the idr@ietf.org list. The BGP-LS 
>> draft is being considered for WG adoption. I think this perspective 
>> would help the idr WG.
>>
>> Thank you,
>>
>> Sue
>>
>> -----Original Message-----
>> From: irs-discuss-bounces@ietf.org 
>> [mailto:irs-discuss-bounces@ietf.org] On Behalf Of Y. Richard Yang
>> Sent: Friday, August 03, 2012 11:04 AM
>> To: stefano previdi
>> Cc: Thomas Nadeau; James Kempf; IETF ALTO; robert@raszuk.net; 
>> irs-discuss@ietf.org
>> Subject: Re: [irs-discuss] About ALTO Vs. BGP-LS
>>
>> Hi Stefano,
>>
>> Good post! I added the ALTO mailing list, given the relevance. I hope
>> that this is OK cross posting.
>>
>> First, a few comments on ALTO:
>>
>> View granularity:
>>
>> - ALTO currently defines two abstract network topology data structures:
>> Network Map and Cost Map(s). More link-state oriented maps (graphs),
>> with aggregations and transformations, can be added efficiently to ALTO.
>> Some initial efforts are already on the way (e.g., see the graph
>> representation proposal at page 9:
>> http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf). Hence,
>> I see a natural next-step is for ALTO to provide a link-state view to
>> external entities.
>>
>> - It is a good comment on the level of details that ALTO should
>> deliver. This is a good question for the ALTO working group and the
>> community to discuss. I feel that ALTO should provide multi-levels of
>> granularity of views, and we should discuss in the working group.
>>
>> Pull vs push delivery mechanism:
>>
>> - There are more discussions on adding the incremental update in ALTO.
>> Multiple mechanisms have been discussed. I feel that it is the right
>> direction for ALTO.
>>
>> Now let me understand the deployment model of BGP-LS. Your explanation
>> on the issues of acquiring routing state is excellent. Let me start by
>> understanding more details on the deployment model of BGP-LS inside a
>> network:
>>
>> - A set N_igp of BGP-LS instances are deployed to collect IGP info. You
>> need multiple instances because IGP needs direct connectivity
>> (adjacency). Each instance here receives (potentially multiple) IGP
>> updates and streams (relays) to an another (remote) entity, which I
>> assume is a master BGP-LS instance. So each of these N_igp instances is
>> IGP-in and BGP-LS out. This appears to be shown in Figure 1 of
>> draft-gredler-idr-ls-distribution.
>>
>> - A set N_egp of BGP-LS instances are deployed to collect BGP feeds. You
>> also may need multiple instances because the network does not want to
>> see aggregated states but raw states. These instances will feed to the
>> master BGP-LS as well.
>>
>> - The master BGP-LS aggregates the multiple BGP-LS ins (and maybe some
>> direct IGP/EGP ins) and pushes (after policy) to other BGP-LS peers to
>> use: for example, an ALTO Server transforms/aggregates the feed to
>> generate ALTO views (maps and graphs), and an PCE uses the feed to
>> maintain its TED. One could even imagine that ALTO builds a detailed TED
>> and feeds to PCE, but this beyond the scope of this discussion. The
>> master BGP-LS is BGP-LS in and BGP-LS out. It is also possible that the
>> master BGP-LS does not push to any other entities and simply maintains
>> an internal DB for others to query.
>>
>> Do I understand it correctly?
>>
>> Now, we can take a look at more specifics of BGP-LS.
>>
>> A first perspective is the semantics of the content. If the objective is
>> to solve the aforementioned deployment issue, then an alternative
>> solution is to introduce a simple LS update tunneling protocol, where a
>> link-state proxy forwards LS messages to a collector. The current design
>> of BGP-LS starts with such a feeling (i.e., an NLRI starts with the
>> Protocol ID, which indicates it is from IS-IS level 1 IS-IS level 2,
>> OSPF, etc).  However, the protocol appears to (try to) go beyond simple
>> tunneling and introduces a common LS schema, by converting/filtering
>> individual IGP LS messages to some common format. I feel that it can be
>> helpful to first specify the schema (LS data model) instead of the
>> specific encoding. For example, OSPF specifies LS Age, and this is
>> filtered. (Please correct me if I missed it). On the other hand, one can
>> think that some Age info can be helpful for one to understand the
>> "freshness" of the LS. A problem studied in database is heterogeneous
>> databases, to merge multiple data sources (IS-IS, OSPF, etc) to a single
>> schema, and there can be many problems. If there is such a study, please
>> do share a pointer.
>>
>> A second perspective is using BGP as the transport. What key features
>> from BGP do we really need (yes, weak-typed TLV encoding offers a lot of
>> flexibility)? What features of BGP do we not need (e.g., BGP is a
>> routing protocol and hence builds in features to handle convergence such
>> as dampening)? What may be missing (e.g., a capability of pull or
>> filtering of push). I feel that these issues should be discussed. If
>> they have already been discussed, please do share a pointer, as I am
>> definitely a new comer.
>>
>> Thanks!
>>
>> Richard
>>
>> On 8/2/12 11:54 PM, stefano previdi wrote:
>>> All,
>>>
>>> as co-author of both BGP-LS and ALTO drafts, I'd try to clarify a few
>>> things:
>>>
>>> ALTO has been designed in order to deliver to applications (through
>>> http/json):
>>>
>>> 1. two maps representing the network topology in an abstracted view
>>>      (or set of views through multiple maps). The map does not include
>>>      the details of a link-state database and therefore have little use
>>>      for any element that would need to retrieve from the network the
>>>      detailed/complete network layer topology, for example: link
>>>      addresses or link BW resources, etc. IOW: ALTO maps do not have
>>>      the granularity of a link-state database and ALTO protocol is not
>>>      designed to deliver such details.
>>>
>>> and/or
>>>
>>> 2. Ranking services where a client sends a bunch of IP addresses in
>>>      a message and the ALTO server replies by ranking these addresses
>>>      based on their topological/network distance (or whatever criteria
>>>      the ALTO server has been configured for). This is called: Endpoint
>>>      Cost Service.
>>>
>>> When using ALTO maps, and the ALTO protocol being http/pull based,
>>> there's no such concept of unsolicited routing updates. An ALTO
>>> client is typically a browser that will pull the maps from an ALTO
>>> server using http. The ALTO server will make no effort to ensure the
>>> client has the latest view of the topology (i.e.: It's the role of the
>>> client to poll for new maps time to time).
>>>
>>> Now, in order for an ALTO server to deliver Maps or Ranking services,
>>> it needs to build some form of topology and in order to achieve this,
>>> it needs somehow to be fed by either the operator (configuration) or
>>> to receive dynamically topology information from the network (e.g.:
>>> ISIS/OSPF/BGP).
>>>
>>> Here we had two options:
>>> 1. ALTO server to implement ISIS/OSPF/BGP and establish IGP adjacencies
>>>      to ABRs or L1L2 routers in each area so to retrieve the LSDB from
>>>      each area. In practice I know no SP willing/accepting to open 
>>> their
>>>      IGP to an ALTO server. Also, IGP requires direct connectivity
>>>      (adjacency) so from an operation point of view is complex and not
>>>      desired.
>>> 2. Use a database distribution protocol running on top of a reliable
>>>      transport layer that would allow an ALTO server to connect to a
>>>      _single_ and _remote_ Route Reflector (i.e.: no need to be 
>>> directly
>>>      connected) and grab the whole network topology that will be 
>>> updated
>>>      using standard routing protocol mechanisms (i.e.: routing updates)
>>>      and that would allow the operator to control (through policies and
>>>      filters) what to distribute and to which server.
>>>
>>>      Benefits: single (or dual at most for redundancy) connection for
>>>      the ALTO server (rather than having a direct adjacency with each
>>>      ABR) and, from the operator perspective, single point of
>>>      distribution of network topology (easier to apply policies and
>>>      control what you deliver). This is what BGP-LS is about.
>>>
>>> BGP-LS defines new AFI/SAFI for a new NLRI and attributes that convey
>>> link-state and, more generically, any type of topology information.
>>> The draft specifies NLRI and attributes that map whatever you can
>>> find in a link-state database.
>>>
>>> We currently have draft-gredler-idr-ls-distribution in the process
>>> (hopefully) to be adopted as WG item in IDR WG (so far we 're getting
>>> good support). We also have 3 implementations of BGP-LS.
>>>
>>> Deployment-wise: BGP-LS is not yet deployed, however, we already have
>>> deployments (I know at least two) where an ALTO server retrieves IP
>>> prefix information from remote BGP RRs (for ipv4). The same scheme
>>> will be used once BGP-LS will be deployed (so to say that it is not
>>> something that we invented after a couple of beers but corresponds to
>>> requirements for delivering network services to upper layers while
>>> still controlling efficiently what you distribute, to whom and in
>>> which form (note that, often, beers are anyway part of the process).
>>>
>>> More information here:
>>> http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02
>>> http://tools.ietf.org/html/draft-ietf-alto-protocol-12
>>>
>>> Hope this helps.
>>>
>>> s.
>>>
>>>
>>>
>>> On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:
>>>> Tom,
>>>>
>>>>> I agree that one of the top work items for this effort should be a
>>>>> standardized topology function, and one that is accessible via a
>>>>> non-routing protocol.
>>>> So if the requirement is to have topology export via non-routing 
>>>> protocol then I think we should seriously revisit or repackage the 
>>>> draft-gredler-idr-ls-distribution-01 which works for for both OSPF 
>>>> and ISIS.
>>>>
>>>> However before that let's really understand the requirement why it 
>>>> must be exported via non-routing protocol .... Keep in mind that 
>>>> just to parse BGP UPDATE messages and retrieve interesting pieces 
>>>> out it it requires very little code rather then full BGP 
>>>> implementation.
>>>>
>>>> The particular feature I like about 
>>>> draft-gredler-idr-ls-distribution-01 is that it is read-only ;)
>>>>
>>>> R.
>>>>
>>>>>     I agree that one of the top work items for this effort should 
>>>>> be a
>>>>> standardized topology function, and one that is accessible via a
>>>>> non-routing protocol.  While not exactly "low hanging fruit", it is
>>>>> something that (to me) is a clear work item with clear goals that 
>>>>> should
>>>>> be tackled straight away.
>>>>>
>>>>>     --Tom
>>>>>
>>>>>
>>>>>
>>>>> On 8/2/12 3:24 PM, "James Kempf" <james.kempf@ericsson.com> wrote:
>>>>>
>>>>>> So after seeing part of Alia's talk this morning (I had to leave 
>>>>>> in the
>>>>>> middle unfortunately), I'd like to make a couple suggestions. 
>>>>>> There were
>>>>>> a lot of ideas presented in the talk, enough for an entire IETF 
>>>>>> Area. I
>>>>>> think to make tangible progress, the work needs to be focussed on 
>>>>>> a small
>>>>>> subset that would be of immediate interest and usability.
>>>>>>
>>>>>> There are a couple areas that suggest themselves, but one that 
>>>>>> would be
>>>>>> useful in work that I've been involved in is a standardized 
>>>>>> format for
>>>>>> network topology representation and a protocol for exchanging it. 
>>>>>> The
>>>>>> Onix OpenFlow controller has a network information base with a
>>>>>> specialized format for network topology, and every OpenFlow 
>>>>>> controller
>>>>>> requires this. Having a standardized way to represent it might 
>>>>>> foster a
>>>>>> common topology database package. Another application is network
>>>>>> management. Every network management system needs some kind of 
>>>>>> topology
>>>>>> representation. Finally, though I am not an expert in PCE 
>>>>>> construction,
>>>>>> it would seem to me that a PCE would need some kind of topology
>>>>>> representation in order to perform path calculations. Having a 
>>>>>> way,for
>>>>>> example, for the OpenFlow controller and the PCE to exchange 
>>>>>> topology
>>>>>> information would be really useful.  I would say to start with 
>>>>>> physical
>>>>>> topology because that is fundamental, but make the format 
>>>>>> flexible enough
>>>>>> to support
>>>>>> virtual topology representation.
>>>>>>
>>>>>>             jak
>>>>>> _______________________________________________
>>>>>> irs-discuss mailing list
>>>>>> irs-discuss@ietf.org
>>>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>>> _______________________________________________
>>>>> irs-discuss mailing list
>>>>> irs-discuss@ietf.org
>>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>>>
>>>>>
>>>> _______________________________________________
>>>> irs-discuss mailing list
>>>> irs-discuss@ietf.org
>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>>
>>> _______________________________________________
>>> irs-discuss mailing list
>>> irs-discuss@ietf.org
>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>> _______________________________________________
>> irs-discuss mailing list
>> irs-discuss@ietf.org
>> https://www.ietf.org/mailman/listinfo/irs-discuss
>
> _______________________________________________
> alto mailing list
> alto@ietf.org
> https://www.ietf.org/mailman/listinfo/alto
>
>


-- 
===================================================
Dr Greg Bernstein, Grotto Networking (510) 573-2237


--------------090209040004060208090600
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <div class="moz-cite-prefix">Hi Richard, Stefano, et. al. thanks for
      the explanations and pointers to the BGP-LS draft. This helps
      avoid any confusion that BGP-LS is in competition with ALTO. <br>
      <br>
      BGP-LS (<a
        href="http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02">http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02</a>)
      could be used to feed topology/path information to ALTO servers or
      path computation elements (PCE).&nbsp; As part of that process it would
      collect link state information from IGPs and possibly abstract it.<br>
      <br>
      One partial point of confusion for me as a non-BGP speaker <span
        class="moz-smiley-s3"><span> ;-) </span></span>&nbsp; was in the
      BGP-LS abstract which stated "In a number of environments, <b><i>a
          component external to a network</i></b> is called upon to
      perform computations based on the network topology and current
      state of the connections within the network, including traffic
      engineering information."&nbsp; or&nbsp; "This document describes a
      mechanism by which links state and traffic engineering information
      can be collected from networks and <b>shared with external
        components</b> using the BGP routing protocol."<br>
      I always considered PCE clients and servers as being within the
      network (working with privileged or sensitive network information)
      and an ALTO server as abstracting that information for use by
      network clients.&nbsp; I guess by "external" to the network meaning was
      external to the "data plane".<br>
      <br>
      I also noticed some parallels with our ALTO large bandwidth
      use-case/modeling work (section 6 of <a
href="http://tools.ietf.org/html/draft-bernstein-alto-large-bandwidth-cases-02">http://tools.ietf.org/html/draft-bernstein-alto-large-bandwidth-cases-02</a>)
      and BGP-LS. In particular we both seem to want graph and path
      abstractions.&nbsp; For path abstractions (aggregation) we needed (due
      to bandwidth constraints) model a path as one or more abstract
      links with appropriate properties.&nbsp; It seems a similar situation
      would hold when trying to aggregate paths with SRLG properties.&nbsp;
      These similarities make sense since both drafts are concerned with
      sharing information about the network in an aggregated/abstract
      form.<br>
      <br>
      Best Regards<br>
      Greg B.<br>
      <br>
      &nbsp;<br>
      On 8/3/2012 11:46 AM, Y. Richard Yang wrote:<br>
    </div>
    <blockquote cite="mid:501C1C74.6040006@cs.yale.edu" type="cite">Hi,
      <br>
      <br>
      Sue suggested that I forward this email to idr for more
      discussions.
      <br>
      <br>
      Thanks.
      <br>
      <br>
      Richard
      <br>
      <br>
      On 8/3/12 11:24 AM, Susan Hares wrote:
      <br>
      <blockquote type="cite">Richard:
        <br>
        <br>
        Could you cross-post this note to the <a class="moz-txt-link-abbreviated" href="mailto:idr@ietf.org">idr@ietf.org</a> list. The
        BGP-LS draft is being considered for WG adoption. I think this
        perspective would help the idr WG.
        <br>
        <br>
        Thank you,
        <br>
        <br>
        Sue
        <br>
        <br>
        -----Original Message-----
        <br>
        From: <a class="moz-txt-link-abbreviated" href="mailto:irs-discuss-bounces@ietf.org">irs-discuss-bounces@ietf.org</a>
        [<a class="moz-txt-link-freetext" href="mailto:irs-discuss-bounces@ietf.org">mailto:irs-discuss-bounces@ietf.org</a>] On Behalf Of Y. Richard
        Yang
        <br>
        Sent: Friday, August 03, 2012 11:04 AM
        <br>
        To: stefano previdi
        <br>
        Cc: Thomas Nadeau; James Kempf; IETF ALTO; <a class="moz-txt-link-abbreviated" href="mailto:robert@raszuk.net">robert@raszuk.net</a>;
        <a class="moz-txt-link-abbreviated" href="mailto:irs-discuss@ietf.org">irs-discuss@ietf.org</a>
        <br>
        Subject: Re: [irs-discuss] About ALTO Vs. BGP-LS
        <br>
        <br>
        Hi Stefano,
        <br>
        <br>
        Good post! I added the ALTO mailing list, given the relevance. I
        hope
        <br>
        that this is OK cross posting.
        <br>
        <br>
        First, a few comments on ALTO:
        <br>
        <br>
        View granularity:
        <br>
        <br>
        - ALTO currently defines two abstract network topology data
        structures:
        <br>
        Network Map and Cost Map(s). More link-state oriented maps
        (graphs),
        <br>
        with aggregations and transformations, can be added efficiently
        to ALTO.
        <br>
        Some initial efforts are already on the way (e.g., see the graph
        <br>
        representation proposal at page 9:
        <br>
        <a class="moz-txt-link-freetext" href="http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf">http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf</a>).
        Hence,
        <br>
        I see a natural next-step is for ALTO to provide a link-state
        view to
        <br>
        external entities.
        <br>
        <br>
        - It is a good comment on the level of details that ALTO should
        <br>
        deliver. This is a good question for the ALTO working group and
        the
        <br>
        community to discuss. I feel that ALTO should provide
        multi-levels of
        <br>
        granularity of views, and we should discuss in the working
        group.
        <br>
        <br>
        Pull vs push delivery mechanism:
        <br>
        <br>
        - There are more discussions on adding the incremental update in
        ALTO.
        <br>
        Multiple mechanisms have been discussed. I feel that it is the
        right
        <br>
        direction for ALTO.
        <br>
        <br>
        Now let me understand the deployment model of BGP-LS. Your
        explanation
        <br>
        on the issues of acquiring routing state is excellent. Let me
        start by
        <br>
        understanding more details on the deployment model of BGP-LS
        inside a
        <br>
        network:
        <br>
        <br>
        - A set N_igp of BGP-LS instances are deployed to collect IGP
        info. You
        <br>
        need multiple instances because IGP needs direct connectivity
        <br>
        (adjacency). Each instance here receives (potentially multiple)
        IGP
        <br>
        updates and streams (relays) to an another (remote) entity,
        which I
        <br>
        assume is a master BGP-LS instance. So each of these N_igp
        instances is
        <br>
        IGP-in and BGP-LS out. This appears to be shown in Figure 1 of
        <br>
        draft-gredler-idr-ls-distribution.
        <br>
        <br>
        - A set N_egp of BGP-LS instances are deployed to collect BGP
        feeds. You
        <br>
        also may need multiple instances because the network does not
        want to
        <br>
        see aggregated states but raw states. These instances will feed
        to the
        <br>
        master BGP-LS as well.
        <br>
        <br>
        - The master BGP-LS aggregates the multiple BGP-LS ins (and
        maybe some
        <br>
        direct IGP/EGP ins) and pushes (after policy) to other BGP-LS
        peers to
        <br>
        use: for example, an ALTO Server transforms/aggregates the feed
        to
        <br>
        generate ALTO views (maps and graphs), and an PCE uses the feed
        to
        <br>
        maintain its TED. One could even imagine that ALTO builds a
        detailed TED
        <br>
        and feeds to PCE, but this beyond the scope of this discussion.
        The
        <br>
        master BGP-LS is BGP-LS in and BGP-LS out. It is also possible
        that the
        <br>
        master BGP-LS does not push to any other entities and simply
        maintains
        <br>
        an internal DB for others to query.
        <br>
        <br>
        Do I understand it correctly?
        <br>
        <br>
        Now, we can take a look at more specifics of BGP-LS.
        <br>
        <br>
        A first perspective is the semantics of the content. If the
        objective is
        <br>
        to solve the aforementioned deployment issue, then an
        alternative
        <br>
        solution is to introduce a simple LS update tunneling protocol,
        where a
        <br>
        link-state proxy forwards LS messages to a collector. The
        current design
        <br>
        of BGP-LS starts with such a feeling (i.e., an NLRI starts with
        the
        <br>
        Protocol ID, which indicates it is from IS-IS level 1 IS-IS
        level 2,
        <br>
        OSPF, etc).&nbsp; However, the protocol appears to (try to) go beyond
        simple
        <br>
        tunneling and introduces a common LS schema, by
        converting/filtering
        <br>
        individual IGP LS messages to some common format. I feel that it
        can be
        <br>
        helpful to first specify the schema (LS data model) instead of
        the
        <br>
        specific encoding. For example, OSPF specifies LS Age, and this
        is
        <br>
        filtered. (Please correct me if I missed it). On the other hand,
        one can
        <br>
        think that some Age info can be helpful for one to understand
        the
        <br>
        "freshness" of the LS. A problem studied in database is
        heterogeneous
        <br>
        databases, to merge multiple data sources (IS-IS, OSPF, etc) to
        a single
        <br>
        schema, and there can be many problems. If there is such a
        study, please
        <br>
        do share a pointer.
        <br>
        <br>
        A second perspective is using BGP as the transport. What key
        features
        <br>
        from BGP do we really need (yes, weak-typed TLV encoding offers
        a lot of
        <br>
        flexibility)? What features of BGP do we not need (e.g., BGP is
        a
        <br>
        routing protocol and hence builds in features to handle
        convergence such
        <br>
        as dampening)? What may be missing (e.g., a capability of pull
        or
        <br>
        filtering of push). I feel that these issues should be
        discussed. If
        <br>
        they have already been discussed, please do share a pointer, as
        I am
        <br>
        definitely a new comer.
        <br>
        <br>
        Thanks!
        <br>
        <br>
        Richard
        <br>
        <br>
        On 8/2/12 11:54 PM, stefano previdi wrote:
        <br>
        <blockquote type="cite">All,
          <br>
          <br>
          as co-author of both BGP-LS and ALTO drafts, I'd try to
          clarify a few
          <br>
          things:
          <br>
          <br>
          ALTO has been designed in order to deliver to applications
          (through
          <br>
          http/json):
          <br>
          <br>
          1. two maps representing the network topology in an abstracted
          view
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; (or set of views through multiple maps). The map does not
          include
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; the details of a link-state database and therefore have
          little use
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; for any element that would need to retrieve from the
          network the
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; detailed/complete network layer topology, for example:
          link
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; addresses or link BW resources, etc. IOW: ALTO maps do
          not have
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; the granularity of a link-state database and ALTO
          protocol is not
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; designed to deliver such details.
          <br>
          <br>
          and/or
          <br>
          <br>
          2. Ranking services where a client sends a bunch of IP
          addresses in
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; a message and the ALTO server replies by ranking these
          addresses
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; based on their topological/network distance (or whatever
          criteria
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; the ALTO server has been configured for). This is called:
          Endpoint
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; Cost Service.
          <br>
          <br>
          When using ALTO maps, and the ALTO protocol being http/pull
          based,
          <br>
          there's no such concept of unsolicited routing updates. An
          ALTO
          <br>
          client is typically a browser that will pull the maps from an
          ALTO
          <br>
          server using http. The ALTO server will make no effort to
          ensure the
          <br>
          client has the latest view of the topology (i.e.: It's the
          role of the
          <br>
          client to poll for new maps time to time).
          <br>
          <br>
          Now, in order for an ALTO server to deliver Maps or Ranking
          services,
          <br>
          it needs to build some form of topology and in order to
          achieve this,
          <br>
          it needs somehow to be fed by either the operator
          (configuration) or
          <br>
          to receive dynamically topology information from the network
          (e.g.:
          <br>
          ISIS/OSPF/BGP).
          <br>
          <br>
          Here we had two options:
          <br>
          1. ALTO server to implement ISIS/OSPF/BGP and establish IGP
          adjacencies
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; to ABRs or L1L2 routers in each area so to retrieve the
          LSDB from
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; each area. In practice I know no SP willing/accepting to
          open their
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; IGP to an ALTO server. Also, IGP requires direct
          connectivity
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; (adjacency) so from an operation point of view is complex
          and not
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; desired.
          <br>
          2. Use a database distribution protocol running on top of a
          reliable
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; transport layer that would allow an ALTO server to
          connect to a
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; _single_ and _remote_ Route Reflector (i.e.: no need to
          be directly
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; connected) and grab the whole network topology that will
          be updated
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; using standard routing protocol mechanisms (i.e.: routing
          updates)
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; and that would allow the operator to control (through
          policies and
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; filters) what to distribute and to which server.
          <br>
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; Benefits: single (or dual at most for redundancy)
          connection for
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; the ALTO server (rather than having a direct adjacency
          with each
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; ABR) and, from the operator perspective, single point of
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; distribution of network topology (easier to apply
          policies and
          <br>
          &nbsp;&nbsp;&nbsp;&nbsp; control what you deliver). This is what BGP-LS is about.
          <br>
          <br>
          BGP-LS defines new AFI/SAFI for a new NLRI and attributes that
          convey
          <br>
          link-state and, more generically, any type of topology
          information.
          <br>
          The draft specifies NLRI and attributes that map whatever you
          can
          <br>
          find in a link-state database.
          <br>
          <br>
          We currently have draft-gredler-idr-ls-distribution in the
          process
          <br>
          (hopefully) to be adopted as WG item in IDR WG (so far we 're
          getting
          <br>
          good support). We also have 3 implementations of BGP-LS.
          <br>
          <br>
          Deployment-wise: BGP-LS is not yet deployed, however, we
          already have
          <br>
          deployments (I know at least two) where an ALTO server
          retrieves IP
          <br>
          prefix information from remote BGP RRs (for ipv4). The same
          scheme
          <br>
          will be used once BGP-LS will be deployed (so to say that it
          is not
          <br>
          something that we invented after a couple of beers but
          corresponds to
          <br>
          requirements for delivering network services to upper layers
          while
          <br>
          still controlling efficiently what you distribute, to whom and
          in
          <br>
          which form (note that, often, beers are anyway part of the
          process).
          <br>
          <br>
          More information here:
          <br>
<a class="moz-txt-link-freetext" href="http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02">http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02</a>
          <br>
          <a class="moz-txt-link-freetext" href="http://tools.ietf.org/html/draft-ietf-alto-protocol-12">http://tools.ietf.org/html/draft-ietf-alto-protocol-12</a>
          <br>
          <br>
          Hope this helps.
          <br>
          <br>
          s.
          <br>
          <br>
          <br>
          <br>
          On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:
          <br>
          <blockquote type="cite">Tom,
            <br>
            <br>
            <blockquote type="cite">I agree that one of the top work
              items for this effort should be a
              <br>
              standardized topology function, and one that is accessible
              via a
              <br>
              non-routing protocol.
              <br>
            </blockquote>
            So if the requirement is to have topology export via
            non-routing protocol then I think we should seriously
            revisit or repackage the
            draft-gredler-idr-ls-distribution-01 which works for for
            both OSPF and ISIS.
            <br>
            <br>
            However before that let's really understand the requirement
            why it must be exported via non-routing protocol .... Keep
            in mind that just to parse BGP UPDATE messages and retrieve
            interesting pieces out it it requires very little code
            rather then full BGP implementation.
            <br>
            <br>
            The particular feature I like about
            draft-gredler-idr-ls-distribution-01 is that it is read-only
            ;)
            <br>
            <br>
            R.
            <br>
            <br>
            <blockquote type="cite">&nbsp;&nbsp;&nbsp;&nbsp;I agree that one of the top work
              items for this effort should be a
              <br>
              standardized topology function, and one that is accessible
              via a
              <br>
              non-routing protocol.&nbsp; While not exactly "low hanging
              fruit", it is
              <br>
              something that (to me) is a clear work item with clear
              goals that should
              <br>
              be tackled straight away.
              <br>
              <br>
              &nbsp;&nbsp;&nbsp;&nbsp;--Tom
              <br>
              <br>
              <br>
              <br>
              On 8/2/12 3:24 PM, "James Kempf"
              <a class="moz-txt-link-rfc2396E" href="mailto:james.kempf@ericsson.com">&lt;james.kempf@ericsson.com&gt;</a> wrote:
              <br>
              <br>
              <blockquote type="cite">So after seeing part of Alia's
                talk this morning (I had to leave in the
                <br>
                middle unfortunately), I'd like to make a couple
                suggestions. There were
                <br>
                a lot of ideas presented in the talk, enough for an
                entire IETF Area. I
                <br>
                think to make tangible progress, the work needs to be
                focussed on a small
                <br>
                subset that would be of immediate interest and
                usability.
                <br>
                <br>
                There are a couple areas that suggest themselves, but
                one that would be
                <br>
                useful in work that I've been involved in is a
                standardized format for
                <br>
                network topology representation and a protocol for
                exchanging it. The
                <br>
                Onix OpenFlow controller has a network information base
                with a
                <br>
                specialized format for network topology, and every
                OpenFlow controller
                <br>
                requires this. Having a standardized way to represent it
                might foster a
                <br>
                common topology database package. Another application is
                network
                <br>
                management. Every network management system needs some
                kind of topology
                <br>
                representation. Finally, though I am not an expert in
                PCE construction,
                <br>
                it would seem to me that a PCE would need some kind of
                topology
                <br>
                representation in order to perform path calculations.
                Having a way,for
                <br>
                example, for the OpenFlow controller and the PCE to
                exchange topology
                <br>
                information would be really useful.&nbsp; I would say to
                start with physical
                <br>
                topology because that is fundamental, but make the
                format flexible enough
                <br>
                to support
                <br>
                virtual topology representation.
                <br>
                <br>
                &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; jak
                <br>
                _______________________________________________
                <br>
                irs-discuss mailing list
                <br>
                <a class="moz-txt-link-abbreviated" href="mailto:irs-discuss@ietf.org">irs-discuss@ietf.org</a>
                <br>
                <a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/irs-discuss">https://www.ietf.org/mailman/listinfo/irs-discuss</a>
                <br>
              </blockquote>
              _______________________________________________
              <br>
              irs-discuss mailing list
              <br>
              <a class="moz-txt-link-abbreviated" href="mailto:irs-discuss@ietf.org">irs-discuss@ietf.org</a>
              <br>
              <a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/irs-discuss">https://www.ietf.org/mailman/listinfo/irs-discuss</a>
              <br>
              <br>
              <br>
            </blockquote>
            _______________________________________________
            <br>
            irs-discuss mailing list
            <br>
            <a class="moz-txt-link-abbreviated" href="mailto:irs-discuss@ietf.org">irs-discuss@ietf.org</a>
            <br>
            <a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/irs-discuss">https://www.ietf.org/mailman/listinfo/irs-discuss</a>
            <br>
            <br>
          </blockquote>
          _______________________________________________
          <br>
          irs-discuss mailing list
          <br>
          <a class="moz-txt-link-abbreviated" href="mailto:irs-discuss@ietf.org">irs-discuss@ietf.org</a>
          <br>
          <a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/irs-discuss">https://www.ietf.org/mailman/listinfo/irs-discuss</a>
          <br>
        </blockquote>
        _______________________________________________
        <br>
        irs-discuss mailing list
        <br>
        <a class="moz-txt-link-abbreviated" href="mailto:irs-discuss@ietf.org">irs-discuss@ietf.org</a>
        <br>
        <a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/irs-discuss">https://www.ietf.org/mailman/listinfo/irs-discuss</a>
        <br>
      </blockquote>
      <br>
      _______________________________________________
      <br>
      alto mailing list
      <br>
      <a class="moz-txt-link-abbreviated" href="mailto:alto@ietf.org">alto@ietf.org</a>
      <br>
      <a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/alto">https://www.ietf.org/mailman/listinfo/alto</a>
      <br>
      <br>
      <br>
    </blockquote>
    <br>
    <br>
    <pre class="moz-signature" cols="72">-- 
===================================================
Dr Greg Bernstein, Grotto Networking (510) 573-2237

</pre>
  </body>
</html>

--------------090209040004060208090600--

From gregb@grotto-networking.com  Mon Aug  6 11:10:02 2012
Return-Path: <gregb@grotto-networking.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 66A4D21E808E; Mon,  6 Aug 2012 11:10:01 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.073
X-Spam-Level: 
X-Spam-Status: No, score=-2.073 tagged_above=-999 required=5 tests=[AWL=-0.075, BAYES_00=-2.599, HTML_MESSAGE=0.001, J_CHICKENPOX_33=0.6]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id AIG58CXYz1Qd; Mon,  6 Aug 2012 11:09:58 -0700 (PDT)
Received: from smtp.webfaction.com (mail6.webfaction.com [74.55.86.74]) by ietfa.amsl.com (Postfix) with ESMTP id E041221E808B; Mon,  6 Aug 2012 11:09:57 -0700 (PDT)
Received: from [192.168.0.124] (c-67-170-243-110.hsd1.ca.comcast.net [67.170.243.110]) by smtp.webfaction.com (Postfix) with ESMTP id D14BD2131D37; Mon,  6 Aug 2012 13:09:56 -0500 (CDT)
Message-ID: <50200870.3000000@grotto-networking.com>
Date: Mon, 06 Aug 2012 11:09:52 -0700
From: Greg Bernstein <gregb@grotto-networking.com>
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:14.0) Gecko/20120713 Thunderbird/14.0
MIME-Version: 1.0
To: Alia Atlas <akatlas@gmail.com>
References: <CC404D94.2D5D%tnadeau@juniper.net> <501B0D75.4090009@raszuk.net> <7E89A05A-CE4E-4FCF-81AB-8F39B42FBF8E@cisco.com> <501C128D.10809@cs.yale.edu> <501C2047.1000100@bell-labs.com> <501C3999.3080804@cs.yale.edu> <CAG4d1rdo3BuEukb9F0aR=wOkKFEM9_esaJVJvn+e_UndQ3jELg@mail.gmail.com>
In-Reply-To: <CAG4d1rdo3BuEukb9F0aR=wOkKFEM9_esaJVJvn+e_UndQ3jELg@mail.gmail.com>
Content-Type: multipart/alternative; boundary="------------080305080401020808030000"
Cc: idr@ietf.org, Volker Hilt <volker.hilt@bell-labs.com>, irs-discuss@ietf.org, James Kempf <james.kempf@ericsson.com>, IETF ALTO <alto@ietf.org>
Subject: Re: [alto] [irs-discuss] About ALTO Vs. BGP-LS
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 06 Aug 2012 18:10:03 -0000

This is a multi-part message in MIME format.
--------------080305080401020808030000
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Hi Alia, one key use case that needs a different level of abstraction is 
the "large bandwidth" use-case. In the P4P paper Richard Yang and 
co-authors were able to affect traffic engineering by changing the path 
costs over time.  However in the "large bandwidth" case we need explicit 
information on bandwidth constraints 
(http://tools.ietf.org/html/draft-bernstein-alto-large-bandwidth-cases-02).

Another dimension to the "abstraction" levels was that of the scope/span 
of users of the information.  Currently ALTO is aimed at open, 
"Internet" -like environments.  However some of the extensions being 
proposed may apply in more "controlled" or "restricted" environments.

Although we are trying to hide network details from the ALTO users there 
may be cases with particular technology information is important. For 
example in our "large bandwidth" use-case document we showed how to 
shared multiple spanning tree instance (MSTI) Ethernet graphs in reduced 
form for VLAN to MSTI assignment in a data center.  We also gave  
optical (WDM and TDM) examples.

Cheers

Greg B.

On 8/3/2012 4:57 PM, Alia Atlas wrote:
>
> I strongly agree that we need to start collecting the use-cases for 
> multiple abstraction levels as well as physical/virtual/private.
>
> Alia
>
> On Aug 3, 2012 4:51 PM, "Y. Richard Yang" <yry@cs.yale.edu 
> <mailto:yry@cs.yale.edu>> wrote:
>
>     Hi Volker,
>
>     On 8/3/12 12:02 PM, Volker Hilt wrote:
>
>
>         I believe that the capability to provide a simplified view on
>         the network topology to an application is a key feature rather
>         than a bug. Applications that want to have a view on network
>         topology don't need need a fine grained view on the topology
>         in most casts and benefit from having an abstracted view. This
>         will simplify processing for the application and enables
>         providers to control the exposure of details.
>
>     Agreed.
>
>         We've seen some numbers for topology data sizes in the
>         incremental updates presentation in the ALTO meeting, which
>         provide some insights into the amounts of data needed for
>         different topology sizes.
>
>         I like the idea of enabling an ALTO server to offer different
>         levels of details. This will enable a server to tailor
>         responses to the specific client. It will add complexity as
>         the ALTO server itself will have to maintain the most complex
>         topology it is offering and will have to keep this topology up
>         to date. But this is an interesting question for discussion in
>         the WG.
>
>     Glad that you also like the idea of different levels of details of
>     the network topology. If the ALTO Server is given a detailed
>     topology (constructed from multiple sources, such as routing feed,
>     LSP feed, configuration files), we can offer multiple topology
>     operators/aggregators to explore the detailed topology, according
>     to need and policy. A simple example operator is level (i.e.,
>     hierarchy such as at the area level), but one might specify other
>     operators (e.g., fisheye focus). There are studies on
>     representation of multi-level graphs that we can try to take
>     advantage of. This can be a subject for the group to explore.
>
>     We may need to study/collect use cases for multiple level
>     topology. One interesting example immediately coming to mind is
>     the Abstract Node concept (specify IP prefix/ASN) used in RSVP-TE.
>
>     Cheers,
>
>     Richard
>
>         Cheers,
>
>         Volker
>
>
>
>
>
>
>
>         On 03.08.2012 11:03, Y. Richard Yang wrote:
>
>             Hi Stefano,
>
>             Good post! I added the ALTO mailing list, given the
>             relevance. I hope
>             that this is OK cross posting.
>
>             First, a few comments on ALTO:
>
>             View granularity:
>
>             - ALTO currently defines two abstract network topology
>             data structures:
>             Network Map and Cost Map(s). More link-state oriented maps
>             (graphs),
>             with aggregations and transformations, can be added
>             efficiently to ALTO.
>             Some initial efforts are already on the way (e.g., see the
>             graph
>             representation proposal at page 9:
>             http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf).
>             Hence,
>             I see a natural next-step is for ALTO to provide a
>             link-state view to
>             external entities.
>
>             - It is a good comment on the level of details that ALTO
>             should
>             delivery. This is a good question for the ALTO working
>             group and the
>             community to discuss. I feel that ALTO should provide
>             multi-levels of
>             granularity of views, and we should discuss in the working
>             group.
>
>             Pull vs push delivery mechanism:
>
>             - There are more discussions on adding the incremental
>             update in ALTO.
>             Multiple mechanisms have been discussed. I feel that it is
>             the right
>             direction for ALTO.
>
>             Now let me understand the deployment model of BGP-LS. Your
>             explanation
>             on the issues of acquiring routing state is excellent. Let
>             me start by
>             understanding more details on the deployment model of
>             BGP-LS inside a
>             network:
>
>             - A set N_igp of BGP-LS instances are deployed to collect
>             IGP info. You
>             need multiple instances because IGP needs direct connectivity
>             (adjacency). Each instance here receives (potentially
>             multiple) IGP
>             updates and streams (relays) to an another (remote)
>             entity, which I
>             assume is a master BGP-LS instance. So each of these N_igp
>             instances is
>             IGP-in and BGP-LS out. This appears to be shown in Figure 1 of
>             draft-gredler-idr-ls-distribution.
>
>             - A set N_egp of BGP-LS instances are deployed to collect
>             BGP feeds. You
>             also may need multiple instances because the network does
>             not want to
>             see aggregated states but raw states. These instances will
>             feed to the
>             master BGP-LS as well.
>
>             - The master BGP-LS aggregates the multiple BGP-LS ins
>             (and maybe some
>             direct IGP/EGP ins) and pushes (after policy) to other
>             BGP-LS peers to
>             use: for example, an ALTO Server transforms/aggregates the
>             feed to
>             generate ALTO views (maps and graphs), and an PCE uses the
>             feed to
>             maintain its TED. One could even imagine that ALTO builds
>             a detailed TED
>             and feeds to PCE, but this beyond the scope of this
>             discussion. The
>             master BGP-LS is BGP-LS in and BGP-LS out. It is also
>             possible that the
>             master BGP-LS does not push to any other entities and
>             simply maintains
>             an internal DB for others to query.
>
>             Do I understand it correctly?
>
>             Now, we can take a look at more specifics of BGP-LS.
>
>             A first perspective is the semantics of the content. If
>             the objective is
>             to solve the aforementioned deployment issue, then an
>             alternative
>             solution is to introduce a simple LS update tunneling
>             protocol, where a
>             link-state proxy forwards LS messages to a collector. The
>             current design
>             of BGP-LS starts with such a feeling (i.e., an NLRI starts
>             with the
>             Protocol ID, which indicates it is from IS-IS level 1
>             IS-IS level 2,
>             OSPF, etc). However, the protocol appears to (try to) go
>             beyond simple
>             tunneling and introduces a common LS schema, by
>             converting/filtering
>             individual IGP LS messages to some common format. I feel
>             that it can be
>             helpful to first specify the schema (LS data model)
>             instead of the
>             specific encoding. For example, OSPF specifies LS Age, and
>             this is
>             filtered. (Please correct me if I missed it). On the other
>             hand, one can
>             think that some Age info can be helpful for one to
>             understand the
>             "freshness" of the LS. A problem studied in database is
>             heterogeneous
>             databases, to merge multiple data sources (IS-IS, OSPF,
>             etc) to a single
>             schema, and there can be many problems. If there is such a
>             study, please
>             do share a pointer.
>
>             A second perspective is using BGP as the transport. What
>             key features
>             from BGP do we really need (yes, weak-typed TLV encoding
>             offers a lot of
>             flexibility)? What features of BGP do we not need (e.g.,
>             BGP is a
>             routing protocol and hence builds in features to handle
>             convergence such
>             as dampening)? What may be missing (e.g., a capability of
>             pull or
>             filtering of push). I feel that these issues should be
>             discussed. If
>             they have already been discussed, please do share a
>             pointer, as I am
>             definitely a new comer.
>
>             Thanks!
>
>             Richard
>
>             On 8/2/12 11:54 PM, stefano previdi wrote:
>
>                 All,
>
>                 as co-author of both BGP-LS and ALTO drafts, I'd try
>                 to clarify a few
>                 things:
>
>                 ALTO has been designed in order to deliver to
>                 applications (through
>                 http/json):
>
>                 1. two maps representing the network topology in an
>                 abstracted view
>                 (or set of views through multiple maps). The map does
>                 not include
>                 the details of a link-state database and therefore
>                 have little use
>                 for any element that would need to retrieve from the
>                 network the
>                 detailed/complete network layer topology, for example:
>                 link
>                 addresses or link BW resources, etc. IOW: ALTO maps do
>                 not have
>                 the granularity of a link-state database and ALTO
>                 protocol is not
>                 designed to deliver such details.
>
>                 and/or
>
>                 2. Ranking services where a client sends a bunch of IP
>                 addresses in
>                 a message and the ALTO server replies by ranking these
>                 addresses
>                 based on their topological/network distance (or
>                 whatever criteria
>                 the ALTO server has been configured for). This is
>                 called: Endpoint
>                 Cost Service.
>
>                 When using ALTO maps, and the ALTO protocol being
>                 http/pull based,
>                 there's no such concept of unsolicited routing
>                 updates. An ALTO
>                 client is typically a browser that will pull the maps
>                 from an ALTO
>                 server using http. The ALTO server will make no effort
>                 to ensure the
>                 client has the latest view of the topology (i.e.: It's
>                 the role of the
>                 client to poll for new maps time to time).
>
>                 Now, in order for an ALTO server to deliver Maps or
>                 Ranking services,
>                 it needs to build some form of topology and in order
>                 to achieve this,
>                 it needs somehow to be fed by either the operator
>                 (configuration) or
>                 to receive dynamically topology information from the
>                 network (e.g.:
>                 ISIS/OSPF/BGP).
>
>                 Here we had two options:
>                 1. ALTO server to implement ISIS/OSPF/BGP and
>                 establish IGP adjacencies
>                 to ABRs or L1L2 routers in each area so to retrieve
>                 the LSDB from
>                 each area. In practice I know no SP willing/accepting
>                 to open their
>                 IGP to an ALTO server. Also, IGP requires direct
>                 connectivity
>                 (adjacency) so from an operation point of view is
>                 complex and not
>                 desired.
>                 2. Use a database distribution protocol running on top
>                 of a reliable
>                 transport layer that would allow an ALTO server to
>                 connect to a
>                 _single_ and _remote_ Route Reflector (i.e.: no need
>                 to be directly
>                 connected) and grab the whole network topology that
>                 will be updated
>                 using standard routing protocol mechanisms (i.e.:
>                 routing updates)
>                 and that would allow the operator to control (through
>                 policies and
>                 filters) what to distribute and to which server.
>
>                 Benefits: single (or dual at most for redundancy)
>                 connection for
>                 the ALTO server (rather than having a direct adjacency
>                 with each
>                 ABR) and, from the operator perspective, single point of
>                 distribution of network topology (easier to apply
>                 policies and
>                 control what you deliver). This is what BGP-LS is about.
>
>                 BGP-LS defines new AFI/SAFI for a new NLRI and
>                 attributes that convey
>                 link-state and, more generically, any type of topology
>                 information.
>                 The draft specifies NLRI and attributes that map
>                 whatever you can
>                 find in a link-state database.
>
>                 We currently have draft-gredler-idr-ls-distribution in
>                 the process
>                 (hopefully) to be adopted as WG item in IDR WG (so far
>                 we 're getting
>                 good support). We also have 3 implementations of BGP-LS.
>
>                 Deployment-wise: BGP-LS is not yet deployed, however,
>                 we already have
>                 deployments (I know at least two) where an ALTO server
>                 retrieves IP
>                 prefix information from remote BGP RRs (for ipv4). The
>                 same scheme
>                 will be used once BGP-LS will be deployed (so to say
>                 that it is not
>                 something that we invented after a couple of beers but
>                 corresponds to
>                 requirements for delivering network services to upper
>                 layers while
>                 still controlling efficiently what you distribute, to
>                 whom and in
>                 which form (note that, often, beers are anyway part of
>                 the process).
>
>                 More information here:
>                 http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02
>                 http://tools.ietf.org/html/draft-ietf-alto-protocol-12
>
>                 Hope this helps.
>
>                 s.
>
>
>
>                 On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:
>
>                     Tom,
>
>                         I agree that one of the top work items for
>                         this effort should be a
>                         standardized topology function, and one that
>                         is accessible via a
>                         non-routing protocol.
>
>                     So if the requirement is to have topology export
>                     via non-routing
>                     protocol then I think we should seriously revisit
>                     or repackage the
>                     draft-gredler-idr-ls-distribution-01 which works
>                     for for both OSPF
>                     and ISIS.
>
>                     However before that let's really understand the
>                     requirement why it
>                     must be exported via non-routing protocol ....
>                     Keep in mind that just
>                     to parse BGP UPDATE messages and retrieve
>                     interesting pieces out it
>                     it requires very little code rather then full BGP
>                     implementation.
>
>                     The particular feature I like about
>                     draft-gredler-idr-ls-distribution-01 is that it is
>                     read-only ;)
>
>                     R.
>
>                         I agree that one of the top work items for
>                         this effort should be a
>                         standardized topology function, and one that
>                         is accessible via a
>                         non-routing protocol. While not exactly "low
>                         hanging fruit", it is
>                         something that (to me) is a clear work item
>                         with clear goals that
>                         should
>                         be tackled straight away.
>
>                         --Tom
>
>
>
>                         On 8/2/12 3:24 PM, "James Kempf"
>                         <james.kempf@ericsson.com
>                         <mailto:james.kempf@ericsson.com>> wrote:
>
>                             So after seeing part of Alia's talk this
>                             morning (I had to leave in
>                             the
>                             middle unfortunately), I'd like to make a
>                             couple suggestions. There
>                             were
>                             a lot of ideas presented in the talk,
>                             enough for an entire IETF
>                             Area. I
>                             think to make tangible progress, the work
>                             needs to be focussed on a
>                             small
>                             subset that would be of immediate interest
>                             and usability.
>
>                             There are a couple areas that suggest
>                             themselves, but one that
>                             would be
>                             useful in work that I've been involved in
>                             is a standardized format for
>                             network topology representation and a
>                             protocol for exchanging it. The
>                             Onix OpenFlow controller has a network
>                             information base with a
>                             specialized format for network topology,
>                             and every OpenFlow controller
>                             requires this. Having a standardized way
>                             to represent it might
>                             foster a
>                             common topology database package. Another
>                             application is network
>                             management. Every network management
>                             system needs some kind of
>                             topology
>                             representation. Finally, though I am not
>                             an expert in PCE
>                             construction,
>                             it would seem to me that a PCE would need
>                             some kind of topology
>                             representation in order to perform path
>                             calculations. Having a way,for
>                             example, for the OpenFlow controller and
>                             the PCE to exchange topology
>                             information would be really useful. I
>                             would say to start with physical
>                             topology because that is fundamental, but
>                             make the format flexible
>                             enough
>                             to support
>                             virtual topology representation.
>
>                             jak
>                             _______________________________________________
>                             irs-discuss mailing list
>                             irs-discuss@ietf.org
>                             <mailto:irs-discuss@ietf.org>
>                             https://www.ietf.org/mailman/listinfo/irs-discuss
>
>                         _______________________________________________
>                         irs-discuss mailing list
>                         irs-discuss@ietf.org <mailto:irs-discuss@ietf.org>
>                         https://www.ietf.org/mailman/listinfo/irs-discuss
>
>
>                     _______________________________________________
>                     irs-discuss mailing list
>                     irs-discuss@ietf.org <mailto:irs-discuss@ietf.org>
>                     https://www.ietf.org/mailman/listinfo/irs-discuss
>
>                 _______________________________________________
>                 irs-discuss mailing list
>                 irs-discuss@ietf.org <mailto:irs-discuss@ietf.org>
>                 https://www.ietf.org/mailman/listinfo/irs-discuss
>
>
>             _______________________________________________
>             irs-discuss mailing list
>             irs-discuss@ietf.org <mailto:irs-discuss@ietf.org>
>             https://www.ietf.org/mailman/listinfo/irs-discuss
>
>     _______________________________________________
>     irs-discuss mailing list
>     irs-discuss@ietf.org <mailto:irs-discuss@ietf.org>
>     https://www.ietf.org/mailman/listinfo/irs-discuss
>
>
>
> _______________________________________________
> alto mailing list
> alto@ietf.org
> https://www.ietf.org/mailman/listinfo/alto


-- 
===================================================
Dr Greg Bernstein, Grotto Networking (510) 573-2237


--------------080305080401020808030000
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <div class="moz-cite-prefix">Hi Alia, one key use case that needs a
      different level of abstraction is the "large bandwidth" use-case.&nbsp;
      In the P4P paper Richard Yang and co-authors were able to affect
      traffic engineering by changing the path costs over time.&nbsp; However
      in the "large bandwidth" case we need explicit information on
      bandwidth constraints (<a
href="http://tools.ietf.org/html/draft-bernstein-alto-large-bandwidth-cases-02">http://tools.ietf.org/html/draft-bernstein-alto-large-bandwidth-cases-02</a>).<br>
      <br>
      Another dimension to the "abstraction" levels was that of the
      scope/span of users of the information.&nbsp; Currently ALTO is aimed
      at open, "Internet" -like environments.&nbsp; However some of the
      extensions being proposed may apply in more "controlled" or
      "restricted" environments.<br>
      <br>
      Although we are trying to hide network details from the ALTO users
      there may be cases with particular technology information is
      important. For example in our "large bandwidth" use-case document
      we showed how to shared multiple spanning tree instance (MSTI)
      Ethernet graphs in reduced form for VLAN to MSTI assignment in a
      data center.&nbsp; We also gave&nbsp; optical (WDM and TDM) examples.<br>
      <br>
      Cheers<br>
      <br>
      Greg B.<br>
      <br>
      On 8/3/2012 4:57 PM, Alia Atlas wrote:<br>
    </div>
    <blockquote
cite="mid:CAG4d1rdo3BuEukb9F0aR=wOkKFEM9_esaJVJvn+e_UndQ3jELg@mail.gmail.com"
      type="cite">
      <p>I strongly agree that we need to start collecting the use-cases
        for multiple abstraction levels as well as
        physical/virtual/private.</p>
      <p>Alia</p>
      <div class="gmail_quote">On Aug 3, 2012 4:51 PM, "Y. Richard Yang"
        &lt;<a moz-do-not-send="true" href="mailto:yry@cs.yale.edu">yry@cs.yale.edu</a>&gt;
        wrote:<br type="attribution">
        <blockquote class="gmail_quote" style="margin:0 0 0
          .8ex;border-left:1px #ccc solid;padding-left:1ex">
          Hi Volker,<br>
          <br>
          On 8/3/12 12:02 PM, Volker Hilt wrote:<br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            <br>
            I believe that the capability to provide a simplified view
            on the network topology to an application is a key feature
            rather than a bug. Applications that want to have a view on
            network topology don't need need a fine grained view on the
            topology in most casts and benefit from having an abstracted
            view. This will simplify processing for the application and
            enables providers to control the exposure of details. <br>
          </blockquote>
          Agreed.<br>
          <br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            We've seen some numbers for topology data sizes in the
            incremental updates presentation in the ALTO meeting, which
            provide some insights into the amounts of data needed for
            different topology sizes.<br>
            <br>
            I like the idea of enabling an ALTO server to offer
            different levels of details. This will enable a server to
            tailor responses to the specific client. It will add
            complexity as the ALTO server itself will have to maintain
            the most complex topology it is offering and will have to
            keep this topology up to date. But this is an interesting
            question for discussion in the WG.<br>
            <br>
          </blockquote>
          Glad that you also like the idea of different levels of
          details of the network topology. If the ALTO Server is given a
          detailed topology (constructed from multiple sources, such as
          routing feed, LSP feed, configuration files), we can offer
          multiple topology operators/aggregators to explore the
          detailed topology, according to need and policy. A simple
          example operator is level (i.e., hierarchy such as at the area
          level), but one might specify other operators (e.g., fisheye
          focus). There are studies on representation of multi-level
          graphs that we can try to take advantage of. This can be a
          subject for the group to explore.<br>
          <br>
          We may need to study/collect use cases for multiple level
          topology. One interesting example immediately coming to mind
          is the Abstract Node concept (specify IP prefix/ASN) used in
          RSVP-TE.<br>
          <br>
          Cheers,<br>
          <br>
          Richard<br>
          <br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            Cheers,<br>
            <br>
            Volker<br>
            <br>
            <br>
            <br>
            <br>
            <br>
            <br>
            <br>
            On 03.08.2012 11:03, Y. Richard Yang wrote:<br>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              Hi Stefano,<br>
              <br>
              Good post! I added the ALTO mailing list, given the
              relevance. I hope<br>
              that this is OK cross posting.<br>
              <br>
              First, a few comments on ALTO:<br>
              <br>
              View granularity:<br>
              <br>
              - ALTO currently defines two abstract network topology
              data structures:<br>
              Network Map and Cost Map(s). More link-state oriented maps
              (graphs),<br>
              with aggregations and transformations, can be added
              efficiently to ALTO.<br>
              Some initial efforts are already on the way (e.g., see the
              graph<br>
              representation proposal at page 9:<br>
              <a moz-do-not-send="true"
                href="http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf"
                target="_blank">http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf</a>).
              Hence,<br>
              I see a natural next-step is for ALTO to provide a
              link-state view to<br>
              external entities.<br>
              <br>
              - It is a good comment on the level of details that ALTO
              should<br>
              delivery. This is a good question for the ALTO working
              group and the<br>
              community to discuss. I feel that ALTO should provide
              multi-levels of<br>
              granularity of views, and we should discuss in the working
              group.<br>
              <br>
              Pull vs push delivery mechanism:<br>
              <br>
              - There are more discussions on adding the incremental
              update in ALTO.<br>
              Multiple mechanisms have been discussed. I feel that it is
              the right<br>
              direction for ALTO.<br>
              <br>
              Now let me understand the deployment model of BGP-LS. Your
              explanation<br>
              on the issues of acquiring routing state is excellent. Let
              me start by<br>
              understanding more details on the deployment model of
              BGP-LS inside a<br>
              network:<br>
              <br>
              - A set N_igp of BGP-LS instances are deployed to collect
              IGP info. You<br>
              need multiple instances because IGP needs direct
              connectivity<br>
              (adjacency). Each instance here receives (potentially
              multiple) IGP<br>
              updates and streams (relays) to an another (remote)
              entity, which I<br>
              assume is a master BGP-LS instance. So each of these N_igp
              instances is<br>
              IGP-in and BGP-LS out. This appears to be shown in Figure
              1 of<br>
              draft-gredler-idr-ls-distribution.<br>
              <br>
              - A set N_egp of BGP-LS instances are deployed to collect
              BGP feeds. You<br>
              also may need multiple instances because the network does
              not want to<br>
              see aggregated states but raw states. These instances will
              feed to the<br>
              master BGP-LS as well.<br>
              <br>
              - The master BGP-LS aggregates the multiple BGP-LS ins
              (and maybe some<br>
              direct IGP/EGP ins) and pushes (after policy) to other
              BGP-LS peers to<br>
              use: for example, an ALTO Server transforms/aggregates the
              feed to<br>
              generate ALTO views (maps and graphs), and an PCE uses the
              feed to<br>
              maintain its TED. One could even imagine that ALTO builds
              a detailed TED<br>
              and feeds to PCE, but this beyond the scope of this
              discussion. The<br>
              master BGP-LS is BGP-LS in and BGP-LS out. It is also
              possible that the<br>
              master BGP-LS does not push to any other entities and
              simply maintains<br>
              an internal DB for others to query.<br>
              <br>
              Do I understand it correctly?<br>
              <br>
              Now, we can take a look at more specifics of BGP-LS.<br>
              <br>
              A first perspective is the semantics of the content. If
              the objective is<br>
              to solve the aforementioned deployment issue, then an
              alternative<br>
              solution is to introduce a simple LS update tunneling
              protocol, where a<br>
              link-state proxy forwards LS messages to a collector. The
              current design<br>
              of BGP-LS starts with such a feeling (i.e., an NLRI starts
              with the<br>
              Protocol ID, which indicates it is from IS-IS level 1
              IS-IS level 2,<br>
              OSPF, etc). However, the protocol appears to (try to) go
              beyond simple<br>
              tunneling and introduces a common LS schema, by
              converting/filtering<br>
              individual IGP LS messages to some common format. I feel
              that it can be<br>
              helpful to first specify the schema (LS data model)
              instead of the<br>
              specific encoding. For example, OSPF specifies LS Age, and
              this is<br>
              filtered. (Please correct me if I missed it). On the other
              hand, one can<br>
              think that some Age info can be helpful for one to
              understand the<br>
              "freshness" of the LS. A problem studied in database is
              heterogeneous<br>
              databases, to merge multiple data sources (IS-IS, OSPF,
              etc) to a single<br>
              schema, and there can be many problems. If there is such a
              study, please<br>
              do share a pointer.<br>
              <br>
              A second perspective is using BGP as the transport. What
              key features<br>
              from BGP do we really need (yes, weak-typed TLV encoding
              offers a lot of<br>
              flexibility)? What features of BGP do we not need (e.g.,
              BGP is a<br>
              routing protocol and hence builds in features to handle
              convergence such<br>
              as dampening)? What may be missing (e.g., a capability of
              pull or<br>
              filtering of push). I feel that these issues should be
              discussed. If<br>
              they have already been discussed, please do share a
              pointer, as I am<br>
              definitely a new comer.<br>
              <br>
              Thanks!<br>
              <br>
              Richard<br>
              <br>
              On 8/2/12 11:54 PM, stefano previdi wrote:<br>
              <blockquote class="gmail_quote" style="margin:0 0 0
                .8ex;border-left:1px #ccc solid;padding-left:1ex">
                All,<br>
                <br>
                as co-author of both BGP-LS and ALTO drafts, I'd try to
                clarify a few<br>
                things:<br>
                <br>
                ALTO has been designed in order to deliver to
                applications (through<br>
                http/json):<br>
                <br>
                1. two maps representing the network topology in an
                abstracted view<br>
                (or set of views through multiple maps). The map does
                not include<br>
                the details of a link-state database and therefore have
                little use<br>
                for any element that would need to retrieve from the
                network the<br>
                detailed/complete network layer topology, for example:
                link<br>
                addresses or link BW resources, etc. IOW: ALTO maps do
                not have<br>
                the granularity of a link-state database and ALTO
                protocol is not<br>
                designed to deliver such details.<br>
                <br>
                and/or<br>
                <br>
                2. Ranking services where a client sends a bunch of IP
                addresses in<br>
                a message and the ALTO server replies by ranking these
                addresses<br>
                based on their topological/network distance (or whatever
                criteria<br>
                the ALTO server has been configured for). This is
                called: Endpoint<br>
                Cost Service.<br>
                <br>
                When using ALTO maps, and the ALTO protocol being
                http/pull based,<br>
                there's no such concept of unsolicited routing updates.
                An ALTO<br>
                client is typically a browser that will pull the maps
                from an ALTO<br>
                server using http. The ALTO server will make no effort
                to ensure the<br>
                client has the latest view of the topology (i.e.: It's
                the role of the<br>
                client to poll for new maps time to time).<br>
                <br>
                Now, in order for an ALTO server to deliver Maps or
                Ranking services,<br>
                it needs to build some form of topology and in order to
                achieve this,<br>
                it needs somehow to be fed by either the operator
                (configuration) or<br>
                to receive dynamically topology information from the
                network (e.g.:<br>
                ISIS/OSPF/BGP).<br>
                <br>
                Here we had two options:<br>
                1. ALTO server to implement ISIS/OSPF/BGP and establish
                IGP adjacencies<br>
                to ABRs or L1L2 routers in each area so to retrieve the
                LSDB from<br>
                each area. In practice I know no SP willing/accepting to
                open their<br>
                IGP to an ALTO server. Also, IGP requires direct
                connectivity<br>
                (adjacency) so from an operation point of view is
                complex and not<br>
                desired.<br>
                2. Use a database distribution protocol running on top
                of a reliable<br>
                transport layer that would allow an ALTO server to
                connect to a<br>
                _single_ and _remote_ Route Reflector (i.e.: no need to
                be directly<br>
                connected) and grab the whole network topology that will
                be updated<br>
                using standard routing protocol mechanisms (i.e.:
                routing updates)<br>
                and that would allow the operator to control (through
                policies and<br>
                filters) what to distribute and to which server.<br>
                <br>
                Benefits: single (or dual at most for redundancy)
                connection for<br>
                the ALTO server (rather than having a direct adjacency
                with each<br>
                ABR) and, from the operator perspective, single point of<br>
                distribution of network topology (easier to apply
                policies and<br>
                control what you deliver). This is what BGP-LS is about.<br>
                <br>
                BGP-LS defines new AFI/SAFI for a new NLRI and
                attributes that convey<br>
                link-state and, more generically, any type of topology
                information.<br>
                The draft specifies NLRI and attributes that map
                whatever you can<br>
                find in a link-state database.<br>
                <br>
                We currently have draft-gredler-idr-ls-distribution in
                the process<br>
                (hopefully) to be adopted as WG item in IDR WG (so far
                we 're getting<br>
                good support). We also have 3 implementations of BGP-LS.<br>
                <br>
                Deployment-wise: BGP-LS is not yet deployed, however, we
                already have<br>
                deployments (I know at least two) where an ALTO server
                retrieves IP<br>
                prefix information from remote BGP RRs (for ipv4). The
                same scheme<br>
                will be used once BGP-LS will be deployed (so to say
                that it is not<br>
                something that we invented after a couple of beers but
                corresponds to<br>
                requirements for delivering network services to upper
                layers while<br>
                still controlling efficiently what you distribute, to
                whom and in<br>
                which form (note that, often, beers are anyway part of
                the process).<br>
                <br>
                More information here:<br>
                <a moz-do-not-send="true"
                  href="http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02"
                  target="_blank">http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02</a><br>
                <a moz-do-not-send="true"
                  href="http://tools.ietf.org/html/draft-ietf-alto-protocol-12"
                  target="_blank">http://tools.ietf.org/html/draft-ietf-alto-protocol-12</a><br>
                <br>
                Hope this helps.<br>
                <br>
                s.<br>
                <br>
                <br>
                <br>
                On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:<br>
                <blockquote class="gmail_quote" style="margin:0 0 0
                  .8ex;border-left:1px #ccc solid;padding-left:1ex">
                  Tom,<br>
                  <br>
                  <blockquote class="gmail_quote" style="margin:0 0 0
                    .8ex;border-left:1px #ccc solid;padding-left:1ex">
                    I agree that one of the top work items for this
                    effort should be a<br>
                    standardized topology function, and one that is
                    accessible via a<br>
                    non-routing protocol.<br>
                  </blockquote>
                  So if the requirement is to have topology export via
                  non-routing<br>
                  protocol then I think we should seriously revisit or
                  repackage the<br>
                  draft-gredler-idr-ls-distribution-01 which works for
                  for both OSPF<br>
                  and ISIS.<br>
                  <br>
                  However before that let's really understand the
                  requirement why it<br>
                  must be exported via non-routing protocol .... Keep in
                  mind that just<br>
                  to parse BGP UPDATE messages and retrieve interesting
                  pieces out it<br>
                  it requires very little code rather then full BGP
                  implementation.<br>
                  <br>
                  The particular feature I like about<br>
                  draft-gredler-idr-ls-distribution-01 is that it is
                  read-only ;)<br>
                  <br>
                  R.<br>
                  <br>
                  <blockquote class="gmail_quote" style="margin:0 0 0
                    .8ex;border-left:1px #ccc solid;padding-left:1ex">
                    I agree that one of the top work items for this
                    effort should be a<br>
                    standardized topology function, and one that is
                    accessible via a<br>
                    non-routing protocol. While not exactly "low hanging
                    fruit", it is<br>
                    something that (to me) is a clear work item with
                    clear goals that<br>
                    should<br>
                    be tackled straight away.<br>
                    <br>
                    --Tom<br>
                    <br>
                    <br>
                    <br>
                    On 8/2/12 3:24 PM, "James Kempf" &lt;<a
                      moz-do-not-send="true"
                      href="mailto:james.kempf@ericsson.com"
                      target="_blank">james.kempf@ericsson.com</a>&gt;
                    wrote:<br>
                    <br>
                    <blockquote class="gmail_quote" style="margin:0 0 0
                      .8ex;border-left:1px #ccc solid;padding-left:1ex">
                      So after seeing part of Alia's talk this morning
                      (I had to leave in<br>
                      the<br>
                      middle unfortunately), I'd like to make a couple
                      suggestions. There<br>
                      were<br>
                      a lot of ideas presented in the talk, enough for
                      an entire IETF<br>
                      Area. I<br>
                      think to make tangible progress, the work needs to
                      be focussed on a<br>
                      small<br>
                      subset that would be of immediate interest and
                      usability.<br>
                      <br>
                      There are a couple areas that suggest themselves,
                      but one that<br>
                      would be<br>
                      useful in work that I've been involved in is a
                      standardized format for<br>
                      network topology representation and a protocol for
                      exchanging it. The<br>
                      Onix OpenFlow controller has a network information
                      base with a<br>
                      specialized format for network topology, and every
                      OpenFlow controller<br>
                      requires this. Having a standardized way to
                      represent it might<br>
                      foster a<br>
                      common topology database package. Another
                      application is network<br>
                      management. Every network management system needs
                      some kind of<br>
                      topology<br>
                      representation. Finally, though I am not an expert
                      in PCE<br>
                      construction,<br>
                      it would seem to me that a PCE would need some
                      kind of topology<br>
                      representation in order to perform path
                      calculations. Having a way,for<br>
                      example, for the OpenFlow controller and the PCE
                      to exchange topology<br>
                      information would be really useful. I would say to
                      start with physical<br>
                      topology because that is fundamental, but make the
                      format flexible<br>
                      enough<br>
                      to support<br>
                      virtual topology representation.<br>
                      <br>
                      jak<br>
                      _______________________________________________<br>
                      irs-discuss mailing list<br>
                      <a moz-do-not-send="true"
                        href="mailto:irs-discuss@ietf.org"
                        target="_blank">irs-discuss@ietf.org</a><br>
                      <a moz-do-not-send="true"
                        href="https://www.ietf.org/mailman/listinfo/irs-discuss"
                        target="_blank">https://www.ietf.org/mailman/listinfo/irs-discuss</a><br>
                    </blockquote>
                    _______________________________________________<br>
                    irs-discuss mailing list<br>
                    <a moz-do-not-send="true"
                      href="mailto:irs-discuss@ietf.org" target="_blank">irs-discuss@ietf.org</a><br>
                    <a moz-do-not-send="true"
                      href="https://www.ietf.org/mailman/listinfo/irs-discuss"
                      target="_blank">https://www.ietf.org/mailman/listinfo/irs-discuss</a><br>
                    <br>
                    <br>
                  </blockquote>
                  _______________________________________________<br>
                  irs-discuss mailing list<br>
                  <a moz-do-not-send="true"
                    href="mailto:irs-discuss@ietf.org" target="_blank">irs-discuss@ietf.org</a><br>
                  <a moz-do-not-send="true"
                    href="https://www.ietf.org/mailman/listinfo/irs-discuss"
                    target="_blank">https://www.ietf.org/mailman/listinfo/irs-discuss</a><br>
                  <br>
                </blockquote>
                _______________________________________________<br>
                irs-discuss mailing list<br>
                <a moz-do-not-send="true"
                  href="mailto:irs-discuss@ietf.org" target="_blank">irs-discuss@ietf.org</a><br>
                <a moz-do-not-send="true"
                  href="https://www.ietf.org/mailman/listinfo/irs-discuss"
                  target="_blank">https://www.ietf.org/mailman/listinfo/irs-discuss</a><br>
              </blockquote>
              <br>
              _______________________________________________<br>
              irs-discuss mailing list<br>
              <a moz-do-not-send="true"
                href="mailto:irs-discuss@ietf.org" target="_blank">irs-discuss@ietf.org</a><br>
              <a moz-do-not-send="true"
                href="https://www.ietf.org/mailman/listinfo/irs-discuss"
                target="_blank">https://www.ietf.org/mailman/listinfo/irs-discuss</a><br>
            </blockquote>
          </blockquote>
          _______________________________________________<br>
          irs-discuss mailing list<br>
          <a moz-do-not-send="true" href="mailto:irs-discuss@ietf.org"
            target="_blank">irs-discuss@ietf.org</a><br>
          <a moz-do-not-send="true"
            href="https://www.ietf.org/mailman/listinfo/irs-discuss"
            target="_blank">https://www.ietf.org/mailman/listinfo/irs-discuss</a><br>
        </blockquote>
      </div>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
      <pre wrap="">_______________________________________________
alto mailing list
<a class="moz-txt-link-abbreviated" href="mailto:alto@ietf.org">alto@ietf.org</a>
<a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/alto">https://www.ietf.org/mailman/listinfo/alto</a>
</pre>
    </blockquote>
    <br>
    <br>
    <pre class="moz-signature" cols="72">-- 
===================================================
Dr Greg Bernstein, Grotto Networking (510) 573-2237

</pre>
  </body>
</html>

--------------080305080401020808030000--

From volker.hilt@bell-labs.com  Tue Aug  7 04:17:05 2012
Return-Path: <volker.hilt@bell-labs.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 2E8C821F86A4; Tue,  7 Aug 2012 04:17:05 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -9.649
X-Spam-Level: 
X-Spam-Status: No, score=-9.649 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HELO_EQ_FR=0.35, J_CHICKENPOX_33=0.6, RCVD_IN_DNSWL_HI=-8]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Cz0kY+SW9f7W; Tue,  7 Aug 2012 04:17:03 -0700 (PDT)
Received: from smail5.alcatel.fr (smail5.alcatel.fr [64.208.49.27]) by ietfa.amsl.com (Postfix) with ESMTP id 4804C21F86A2; Tue,  7 Aug 2012 04:17:03 -0700 (PDT)
Received: from FRMRSSXCHHUB02.dc-m.alcatel-lucent.com (FRMRSSXCHHUB02.dc-m.alcatel-lucent.com [135.120.45.62]) by smail5.alcatel.fr (8.14.3/8.14.3/ICT) with ESMTP id q77BGs5V020020 (version=TLSv1/SSLv3 cipher=RC4-MD5 bits=128 verify=NOT); Tue, 7 Aug 2012 13:16:56 +0200
Received: from [149.204.61.15] (135.120.57.7) by FRMRSSXCHHUB02.dc-m.alcatel-lucent.com (135.120.45.62) with Microsoft SMTP Server (TLS) id 8.3.213.0; Tue, 7 Aug 2012 13:16:55 +0200
Message-ID: <5020F926.707@bell-labs.com>
Date: Tue, 7 Aug 2012 13:16:54 +0200
From: Volker Hilt <volker.hilt@bell-labs.com>
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:13.0) Gecko/20120614 Thunderbird/13.0.1
MIME-Version: 1.0
To: "Y. Richard Yang" <yry@cs.yale.edu>
References: <CC404D94.2D5D%tnadeau@juniper.net> <501B0D75.4090009@raszuk.net> <7E89A05A-CE4E-4FCF-81AB-8F39B42FBF8E@cisco.com> <501C128D.10809@cs.yale.edu> <501C2047.1000100@bell-labs.com> <501C3999.3080804@cs.yale.edu>
In-Reply-To: <501C3999.3080804@cs.yale.edu>
Content-Type: text/plain; charset="ISO-8859-1"; format=flowed
Content-Transfer-Encoding: 7bit
X-Scanned-By: MIMEDefang 2.69 on 155.132.188.13
X-Mailman-Approved-At: Tue, 07 Aug 2012 08:01:09 -0700
Cc: idr@ietf.org, irs-discuss@ietf.org, James Kempf <james.kempf@ericsson.com>, IETF ALTO <alto@ietf.org>
Subject: Re: [alto] [irs-discuss] About ALTO Vs. BGP-LS
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 07 Aug 2012 11:17:05 -0000

Richard,
>>
>> I like the idea of enabling an ALTO server to offer different levels
>> of details. This will enable a server to tailor responses to the
>> specific client. It will add complexity as the ALTO server itself will
>> have to maintain the most complex topology it is offering and will
>> have to keep this topology up to date. But this is an interesting
>> question for discussion in the WG.
>>
> Glad that you also like the idea of different levels of details of the
> network topology. If the ALTO Server is given a detailed topology
> (constructed from multiple sources, such as routing feed, LSP feed,
> configuration files), we can offer multiple topology
> operators/aggregators to explore the detailed topology, according to
> need and policy. A simple example operator is level (i.e., hierarchy
> such as at the area level), but one might specify other operators (e.g.,
> fisheye focus). There are studies on representation of multi-level
> graphs that we can try to take advantage of. This can be a subject for
> the group to explore.
>
Yes, I agree - this is an interesting area to explore. Another operator 
could be a view for a specific use, e.g., a view for CDN use. The 
operator can then populate this view with the desired abstraction level 
and policy.

> We may need to study/collect use cases for multiple level topology. One
> interesting example immediately coming to mind is the Abstract Node
> concept (specify IP prefix/ASN) used in RSVP-TE.
>
Indeed.

Volker



>> On 03.08.2012 11:03, Y. Richard Yang wrote:
>>> Hi Stefano,
>>>
>>> Good post! I added the ALTO mailing list, given the relevance. I hope
>>> that this is OK cross posting.
>>>
>>> First, a few comments on ALTO:
>>>
>>> View granularity:
>>>
>>> - ALTO currently defines two abstract network topology data structures:
>>> Network Map and Cost Map(s). More link-state oriented maps (graphs),
>>> with aggregations and transformations, can be added efficiently to ALTO.
>>> Some initial efforts are already on the way (e.g., see the graph
>>> representation proposal at page 9:
>>> http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf). Hence,
>>> I see a natural next-step is for ALTO to provide a link-state view to
>>> external entities.
>>>
>>> - It is a good comment on the level of details that ALTO should
>>> delivery. This is a good question for the ALTO working group and the
>>> community to discuss. I feel that ALTO should provide multi-levels of
>>> granularity of views, and we should discuss in the working group.
>>>
>>> Pull vs push delivery mechanism:
>>>
>>> - There are more discussions on adding the incremental update in ALTO.
>>> Multiple mechanisms have been discussed. I feel that it is the right
>>> direction for ALTO.
>>>
>>> Now let me understand the deployment model of BGP-LS. Your explanation
>>> on the issues of acquiring routing state is excellent. Let me start by
>>> understanding more details on the deployment model of BGP-LS inside a
>>> network:
>>>
>>> - A set N_igp of BGP-LS instances are deployed to collect IGP info. You
>>> need multiple instances because IGP needs direct connectivity
>>> (adjacency). Each instance here receives (potentially multiple) IGP
>>> updates and streams (relays) to an another (remote) entity, which I
>>> assume is a master BGP-LS instance. So each of these N_igp instances is
>>> IGP-in and BGP-LS out. This appears to be shown in Figure 1 of
>>> draft-gredler-idr-ls-distribution.
>>>
>>> - A set N_egp of BGP-LS instances are deployed to collect BGP feeds. You
>>> also may need multiple instances because the network does not want to
>>> see aggregated states but raw states. These instances will feed to the
>>> master BGP-LS as well.
>>>
>>> - The master BGP-LS aggregates the multiple BGP-LS ins (and maybe some
>>> direct IGP/EGP ins) and pushes (after policy) to other BGP-LS peers to
>>> use: for example, an ALTO Server transforms/aggregates the feed to
>>> generate ALTO views (maps and graphs), and an PCE uses the feed to
>>> maintain its TED. One could even imagine that ALTO builds a detailed TED
>>> and feeds to PCE, but this beyond the scope of this discussion. The
>>> master BGP-LS is BGP-LS in and BGP-LS out. It is also possible that the
>>> master BGP-LS does not push to any other entities and simply maintains
>>> an internal DB for others to query.
>>>
>>> Do I understand it correctly?
>>>
>>> Now, we can take a look at more specifics of BGP-LS.
>>>
>>> A first perspective is the semantics of the content. If the objective is
>>> to solve the aforementioned deployment issue, then an alternative
>>> solution is to introduce a simple LS update tunneling protocol, where a
>>> link-state proxy forwards LS messages to a collector. The current design
>>> of BGP-LS starts with such a feeling (i.e., an NLRI starts with the
>>> Protocol ID, which indicates it is from IS-IS level 1 IS-IS level 2,
>>> OSPF, etc). However, the protocol appears to (try to) go beyond simple
>>> tunneling and introduces a common LS schema, by converting/filtering
>>> individual IGP LS messages to some common format. I feel that it can be
>>> helpful to first specify the schema (LS data model) instead of the
>>> specific encoding. For example, OSPF specifies LS Age, and this is
>>> filtered. (Please correct me if I missed it). On the other hand, one can
>>> think that some Age info can be helpful for one to understand the
>>> "freshness" of the LS. A problem studied in database is heterogeneous
>>> databases, to merge multiple data sources (IS-IS, OSPF, etc) to a single
>>> schema, and there can be many problems. If there is such a study, please
>>> do share a pointer.
>>>
>>> A second perspective is using BGP as the transport. What key features
>>> from BGP do we really need (yes, weak-typed TLV encoding offers a lot of
>>> flexibility)? What features of BGP do we not need (e.g., BGP is a
>>> routing protocol and hence builds in features to handle convergence such
>>> as dampening)? What may be missing (e.g., a capability of pull or
>>> filtering of push). I feel that these issues should be discussed. If
>>> they have already been discussed, please do share a pointer, as I am
>>> definitely a new comer.
>>>
>>> Thanks!
>>>
>>> Richard
>>>
>>> On 8/2/12 11:54 PM, stefano previdi wrote:
>>>> All,
>>>>
>>>> as co-author of both BGP-LS and ALTO drafts, I'd try to clarify a few
>>>> things:
>>>>
>>>> ALTO has been designed in order to deliver to applications (through
>>>> http/json):
>>>>
>>>> 1. two maps representing the network topology in an abstracted view
>>>> (or set of views through multiple maps). The map does not include
>>>> the details of a link-state database and therefore have little use
>>>> for any element that would need to retrieve from the network the
>>>> detailed/complete network layer topology, for example: link
>>>> addresses or link BW resources, etc. IOW: ALTO maps do not have
>>>> the granularity of a link-state database and ALTO protocol is not
>>>> designed to deliver such details.
>>>>
>>>> and/or
>>>>
>>>> 2. Ranking services where a client sends a bunch of IP addresses in
>>>> a message and the ALTO server replies by ranking these addresses
>>>> based on their topological/network distance (or whatever criteria
>>>> the ALTO server has been configured for). This is called: Endpoint
>>>> Cost Service.
>>>>
>>>> When using ALTO maps, and the ALTO protocol being http/pull based,
>>>> there's no such concept of unsolicited routing updates. An ALTO
>>>> client is typically a browser that will pull the maps from an ALTO
>>>> server using http. The ALTO server will make no effort to ensure the
>>>> client has the latest view of the topology (i.e.: It's the role of the
>>>> client to poll for new maps time to time).
>>>>
>>>> Now, in order for an ALTO server to deliver Maps or Ranking services,
>>>> it needs to build some form of topology and in order to achieve this,
>>>> it needs somehow to be fed by either the operator (configuration) or
>>>> to receive dynamically topology information from the network (e.g.:
>>>> ISIS/OSPF/BGP).
>>>>
>>>> Here we had two options:
>>>> 1. ALTO server to implement ISIS/OSPF/BGP and establish IGP adjacencies
>>>> to ABRs or L1L2 routers in each area so to retrieve the LSDB from
>>>> each area. In practice I know no SP willing/accepting to open their
>>>> IGP to an ALTO server. Also, IGP requires direct connectivity
>>>> (adjacency) so from an operation point of view is complex and not
>>>> desired.
>>>> 2. Use a database distribution protocol running on top of a reliable
>>>> transport layer that would allow an ALTO server to connect to a
>>>> _single_ and _remote_ Route Reflector (i.e.: no need to be directly
>>>> connected) and grab the whole network topology that will be updated
>>>> using standard routing protocol mechanisms (i.e.: routing updates)
>>>> and that would allow the operator to control (through policies and
>>>> filters) what to distribute and to which server.
>>>>
>>>> Benefits: single (or dual at most for redundancy) connection for
>>>> the ALTO server (rather than having a direct adjacency with each
>>>> ABR) and, from the operator perspective, single point of
>>>> distribution of network topology (easier to apply policies and
>>>> control what you deliver). This is what BGP-LS is about.
>>>>
>>>> BGP-LS defines new AFI/SAFI for a new NLRI and attributes that convey
>>>> link-state and, more generically, any type of topology information.
>>>> The draft specifies NLRI and attributes that map whatever you can
>>>> find in a link-state database.
>>>>
>>>> We currently have draft-gredler-idr-ls-distribution in the process
>>>> (hopefully) to be adopted as WG item in IDR WG (so far we 're getting
>>>> good support). We also have 3 implementations of BGP-LS.
>>>>
>>>> Deployment-wise: BGP-LS is not yet deployed, however, we already have
>>>> deployments (I know at least two) where an ALTO server retrieves IP
>>>> prefix information from remote BGP RRs (for ipv4). The same scheme
>>>> will be used once BGP-LS will be deployed (so to say that it is not
>>>> something that we invented after a couple of beers but corresponds to
>>>> requirements for delivering network services to upper layers while
>>>> still controlling efficiently what you distribute, to whom and in
>>>> which form (note that, often, beers are anyway part of the process).
>>>>
>>>> More information here:
>>>> http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02
>>>> http://tools.ietf.org/html/draft-ietf-alto-protocol-12
>>>>
>>>> Hope this helps.
>>>>
>>>> s.
>>>>
>>>>
>>>>
>>>> On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:
>>>>> Tom,
>>>>>
>>>>>> I agree that one of the top work items for this effort should be a
>>>>>> standardized topology function, and one that is accessible via a
>>>>>> non-routing protocol.
>>>>> So if the requirement is to have topology export via non-routing
>>>>> protocol then I think we should seriously revisit or repackage the
>>>>> draft-gredler-idr-ls-distribution-01 which works for for both OSPF
>>>>> and ISIS.
>>>>>
>>>>> However before that let's really understand the requirement why it
>>>>> must be exported via non-routing protocol .... Keep in mind that just
>>>>> to parse BGP UPDATE messages and retrieve interesting pieces out it
>>>>> it requires very little code rather then full BGP implementation.
>>>>>
>>>>> The particular feature I like about
>>>>> draft-gredler-idr-ls-distribution-01 is that it is read-only ;)
>>>>>
>>>>> R.
>>>>>
>>>>>> I agree that one of the top work items for this effort should be a
>>>>>> standardized topology function, and one that is accessible via a
>>>>>> non-routing protocol. While not exactly "low hanging fruit", it is
>>>>>> something that (to me) is a clear work item with clear goals that
>>>>>> should
>>>>>> be tackled straight away.
>>>>>>
>>>>>> --Tom
>>>>>>
>>>>>>
>>>>>>
>>>>>> On 8/2/12 3:24 PM, "James Kempf" <james.kempf@ericsson.com> wrote:
>>>>>>
>>>>>>> So after seeing part of Alia's talk this morning (I had to leave in
>>>>>>> the
>>>>>>> middle unfortunately), I'd like to make a couple suggestions. There
>>>>>>> were
>>>>>>> a lot of ideas presented in the talk, enough for an entire IETF
>>>>>>> Area. I
>>>>>>> think to make tangible progress, the work needs to be focussed on a
>>>>>>> small
>>>>>>> subset that would be of immediate interest and usability.
>>>>>>>
>>>>>>> There are a couple areas that suggest themselves, but one that
>>>>>>> would be
>>>>>>> useful in work that I've been involved in is a standardized
>>>>>>> format for
>>>>>>> network topology representation and a protocol for exchanging it.
>>>>>>> The
>>>>>>> Onix OpenFlow controller has a network information base with a
>>>>>>> specialized format for network topology, and every OpenFlow
>>>>>>> controller
>>>>>>> requires this. Having a standardized way to represent it might
>>>>>>> foster a
>>>>>>> common topology database package. Another application is network
>>>>>>> management. Every network management system needs some kind of
>>>>>>> topology
>>>>>>> representation. Finally, though I am not an expert in PCE
>>>>>>> construction,
>>>>>>> it would seem to me that a PCE would need some kind of topology
>>>>>>> representation in order to perform path calculations. Having a
>>>>>>> way,for
>>>>>>> example, for the OpenFlow controller and the PCE to exchange
>>>>>>> topology
>>>>>>> information would be really useful. I would say to start with
>>>>>>> physical
>>>>>>> topology because that is fundamental, but make the format flexible
>>>>>>> enough
>>>>>>> to support
>>>>>>> virtual topology representation.
>>>>>>>
>>>>>>> jak
>>>>>>> _______________________________________________
>>>>>>> irs-discuss mailing list
>>>>>>> irs-discuss@ietf.org
>>>>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>>>> _______________________________________________
>>>>>> irs-discuss mailing list
>>>>>> irs-discuss@ietf.org
>>>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>>>>
>>>>>>
>>>>> _______________________________________________
>>>>> irs-discuss mailing list
>>>>> irs-discuss@ietf.org
>>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>>>
>>>> _______________________________________________
>>>> irs-discuss mailing list
>>>> irs-discuss@ietf.org
>>>> https://www.ietf.org/mailman/listinfo/irs-discuss
>>>
>>> _______________________________________________
>>> irs-discuss mailing list
>>> irs-discuss@ietf.org
>>> https://www.ietf.org/mailman/listinfo/irs-discuss


From jnc@mercury.lcs.mit.edu  Tue Aug  7 08:15:30 2012
Return-Path: <jnc@mercury.lcs.mit.edu>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 5DD1521F8550; Tue,  7 Aug 2012 08:15:30 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.27
X-Spam-Level: 
X-Spam-Status: No, score=-6.27 tagged_above=-999 required=5 tests=[AWL=0.329,  BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id JL2GS3NR3xE5; Tue,  7 Aug 2012 08:15:29 -0700 (PDT)
Received: from mercury.lcs.mit.edu (mercury.lcs.mit.edu [18.26.0.122]) by ietfa.amsl.com (Postfix) with ESMTP id 71F7F21F854D; Tue,  7 Aug 2012 08:15:29 -0700 (PDT)
Received: by mercury.lcs.mit.edu (Postfix, from userid 11178) id 1898F18C0A9; Tue,  7 Aug 2012 11:15:29 -0400 (EDT)
To: irs-discuss@ietf.org
Message-Id: <20120807151529.1898F18C0A9@mercury.lcs.mit.edu>
Date: Tue,  7 Aug 2012 11:15:29 -0400 (EDT)
From: jnc@mercury.lcs.mit.edu (Noel Chiappa)
X-Mailman-Approved-At: Tue, 07 Aug 2012 10:18:56 -0700
Cc: idr@ietf.org, alto@ietf.org, jnc@mercury.lcs.mit.edu
Subject: Re: [alto] [irs-discuss] About ALTO Vs. BGP-LS
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 07 Aug 2012 15:15:30 -0000

    > From: "Y. Richard Yang" <yry@cs.yale.edu>

    > Glad that you also like the idea of different levels of details of the
    > network topology. If the ALTO Server is given a detailed topology
    > ... we can offer multiple topology operators/aggregators to explore the
    > detailed topology, according to need and policy.

Too bad we have a routing architecture that deals (as its fundamental
currency) in destination tables, rather than topology - and, in particular,
in a multi-level representation of the toplogy. Then all this stuff would
just fall out naturally.

(And if I sound somewhat cranky, perhaps people will forgive me - although I
expect not that many on this list will know what I'm implicitly referring to.)


    > There are studies on representation of multi-level graphs that we can
    > try to take advantage of. This can be a subject for the group to explore.

There's a PhD thesis which is relevant to this area (since part of it is
about clustering, i.e. in representing an area of the graph without full
detail, one has to decide where to set the boundaries of said part):

  Jacob Hagouel, "Issues in Routing for Large and Dynamic Netoworks",
  Columbia University, 1983

A lot of it is irrelevant (to me, at least), since it's talking about
distributed path computation (which I have long since concluded is 'buggy
whip' technology), but the clustering, etc, content is good.

	Noel

From yang.r.yang@gmail.com  Tue Aug  7 11:34:38 2012
Return-Path: <yang.r.yang@gmail.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 6F3FA21F859A; Tue,  7 Aug 2012 11:34:38 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.671
X-Spam-Level: 
X-Spam-Status: No, score=-2.671 tagged_above=-999 required=5 tests=[AWL=0.305,  BAYES_00=-2.599, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-1]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id EYitW0Gt8--0; Tue,  7 Aug 2012 11:34:36 -0700 (PDT)
Received: from mail-ob0-f172.google.com (mail-ob0-f172.google.com [209.85.214.172]) by ietfa.amsl.com (Postfix) with ESMTP id CD50E21F8608; Tue,  7 Aug 2012 11:34:35 -0700 (PDT)
Received: by obbwc20 with SMTP id wc20so8938530obb.31 for <multiple recipients>; Tue, 07 Aug 2012 11:34:28 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:sender:in-reply-to:references:date :x-google-sender-auth:message-id:subject:from:to:cc:content-type; bh=wR5qESaNP2wvFY5Sv5q0yQJ3gjwL+WoMWfFzRWZ5dhU=; b=LqCBEZWIoLAX/q4K5hQFZTuq5pczqGpMRJBaKVHDBimJ2SUEIFUh+azAQPZTUw+PNf h4V4HJTD7qUIozxZrosGcnYhWC/KcPaS4l+H1zq6r/E+k2bQyK5DwLCDKIIIo04iNsAc ULwP/2BgVgxez176vqpHHWTxUSmCnBeM5NjrCrAso6w8q05GqOS+vwG4WTd1AqVlCnts 9B1VszEZoUmt9Rr0r76lmQMOz6bux5+LXsythAB4a4Hx1PUyoED2gpB7JOTKRwXGQdXD A0KRZsSouKS8Jsp3Ncf5ENX4SyBVjG9Z7rf/w0KIIWgE6w0RbG9FuCi2QU/ab5Wy4mla RcAA==
MIME-Version: 1.0
Received: by 10.182.231.6 with SMTP id tc6mr26192898obc.63.1344364468041; Tue, 07 Aug 2012 11:34:28 -0700 (PDT)
Sender: yang.r.yang@gmail.com
Received: by 10.76.86.136 with HTTP; Tue, 7 Aug 2012 11:34:27 -0700 (PDT)
In-Reply-To: <20120807151529.1898F18C0A9@mercury.lcs.mit.edu>
References: <20120807151529.1898F18C0A9@mercury.lcs.mit.edu>
Date: Tue, 7 Aug 2012 14:34:27 -0400
X-Google-Sender-Auth: uouNH7IXlJxwGkqzpjxeg8MBvqk
Message-ID: <CANUuoLp_EcqaBKjvgKXpFQXG=pvye7KvBi78So+xnHV5PpPVyQ@mail.gmail.com>
From: "Y. Richard Yang" <yry@cs.yale.edu>
To: Noel Chiappa <jnc@mercury.lcs.mit.edu>
Content-Type: multipart/alternative; boundary=f46d0446312c90f8b604c6b13e5d
Cc: idr@ietf.org, alto@ietf.org, irs-discuss@ietf.org
Subject: Re: [alto] [irs-discuss] About ALTO Vs. BGP-LS
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 07 Aug 2012 18:34:39 -0000

--f46d0446312c90f8b604c6b13e5d
Content-Type: text/plain; charset=ISO-8859-1

On Tue, Aug 7, 2012 at 11:15 AM, Noel Chiappa <jnc@mercury.lcs.mit.edu>wrote:

>     > From: "Y. Richard Yang" <yry@cs.yale.edu>
>
>     > Glad that you also like the idea of different levels of details of
> the
>     > network topology. If the ALTO Server is given a detailed topology
>     > ... we can offer multiple topology operators/aggregators to explore
> the
>     > detailed topology, according to need and policy.
>
> Too bad we have a routing architecture that deals (as its fundamental
> currency) in destination tables, rather than topology - and, in particular,
> in a multi-level representation of the toplogy. Then all this stuff would
> just fall out naturally.
>
>
It appears to have increasing efforts focusing on building
abstractions/network
computations based on topology. Hope that it moves somewhere. We are finding
more benefits of using topology-based network routing/control, e.g.,
McNettle
(Multi-core Nettle), an SDN controlled being developed here, uses STM to
manage
topology in a highly concurrent controller and achieves pretty good
scalability.



> (And if I sound somewhat cranky, perhaps people will forgive me - although
> I
> expect not that many on this list will know what I'm implicitly referring
> to.)
>
>
>     > There are studies on representation of multi-level graphs that we can
>     > try to take advantage of. This can be a subject for the group to
> explore.
>
> There's a PhD thesis which is relevant to this area (since part of it is
> about clustering, i.e. in representing an area of the graph without full
> detail, one has to decide where to set the boundaries of said part):
>
>   Jacob Hagouel, "Issues in Routing for Large and Dynamic Netoworks",
>   Columbia University, 1983
>
> A lot of it is irrelevant (to me, at least), since it's talking about
> distributed path computation (which I have long since concluded is 'buggy
> whip' technology), but the clustering, etc, content is good.
>

Thanks a lot for the pointer. Definitely will read.

Richard


>
>         Noel
>

--f46d0446312c90f8b604c6b13e5d
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<br><br><div class=3D"gmail_quote">On Tue, Aug 7, 2012 at 11:15 AM, Noel Ch=
iappa <span dir=3D"ltr">&lt;<a href=3D"mailto:jnc@mercury.lcs.mit.edu" targ=
et=3D"_blank">jnc@mercury.lcs.mit.edu</a>&gt;</span> wrote:<br><blockquote =
class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid=
;padding-left:1ex">
=A0 =A0 &gt; From: &quot;Y. Richard Yang&quot; &lt;<a href=3D"mailto:yry@cs=
.yale.edu">yry@cs.yale.edu</a>&gt;<br>
<div class=3D"im"><br>
=A0 =A0 &gt; Glad that you also like the idea of different levels of detail=
s of the<br>
=A0 =A0 &gt; network topology. If the ALTO Server is given a detailed topol=
ogy<br>
</div>=A0 =A0 &gt; ... we can offer multiple topology operators/aggregators=
 to explore the<br>
<div class=3D"im">=A0 =A0 &gt; detailed topology, according to need and pol=
icy.<br>
<br>
</div>Too bad we have a routing architecture that deals (as its fundamental=
<br>
currency) in destination tables, rather than topology - and, in particular,=
<br>
in a multi-level representation of the toplogy. Then all this stuff would<b=
r>
just fall out naturally.<br>
<br></blockquote><div>=A0</div><div>It appears to have increasing efforts f=
ocusing on building abstractions/network</div><div>computations based on to=
pology. Hope that it moves somewhere. We are finding</div><div>more benefit=
s of using topology-based network routing/control, e.g., McNettle</div>
<div>(Multi-core Nettle), an SDN controlled being developed here, uses STM =
to manage</div><div>topology in a highly concurrent controller and achieves=
 pretty good scalability.</div><div><br></div><div>=A0</div><blockquote cla=
ss=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;pa=
dding-left:1ex">

(And if I sound somewhat cranky, perhaps people will forgive me - although =
I<br>
expect not that many on this list will know what I&#39;m implicitly referri=
ng to.)<br>
<div class=3D"im"><br>
<br>
=A0 =A0 &gt; There are studies on representation of multi-level graphs that=
 we can<br>
=A0 =A0 &gt; try to take advantage of. This can be a subject for the group =
to explore.<br>
<br>
</div>There&#39;s a PhD thesis which is relevant to this area (since part o=
f it is<br>
about clustering, i.e. in representing an area of the graph without full<br=
>
detail, one has to decide where to set the boundaries of said part):<br>
<br>
=A0 Jacob Hagouel, &quot;Issues in Routing for Large and Dynamic Netoworks&=
quot;,<br>
=A0 Columbia University, 1983<br>
<br>
A lot of it is irrelevant (to me, at least), since it&#39;s talking about<b=
r>
distributed path computation (which I have long since concluded is &#39;bug=
gy<br>
whip&#39; technology), but the clustering, etc, content is good.<br></block=
quote><div><br></div><div>Thanks a lot for the pointer. Definitely will rea=
d.</div><div><br></div><div>Richard</div><div>=A0</div><blockquote class=3D=
"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding=
-left:1ex">

<span class=3D"HOEnZb"><font color=3D"#888888"><br>
=A0 =A0 =A0 =A0 Noel<br>
</font></span></blockquote></div><br>

--f46d0446312c90f8b604c6b13e5d--

From gregb@grotto-networking.com  Thu Aug  9 08:47:20 2012
Return-Path: <gregb@grotto-networking.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 5BC3B21F87AB; Thu,  9 Aug 2012 08:47:20 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.615
X-Spam-Level: 
X-Spam-Status: No, score=-1.615 tagged_above=-999 required=5 tests=[AWL=0.383,  BAYES_00=-2.599, HTML_MESSAGE=0.001, J_CHICKENPOX_33=0.6]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id BQEICb0rqvOD; Thu,  9 Aug 2012 08:47:17 -0700 (PDT)
Received: from smtp.webfaction.com (mail6.webfaction.com [74.55.86.74]) by ietfa.amsl.com (Postfix) with ESMTP id 7DC7721F87AA; Thu,  9 Aug 2012 08:47:17 -0700 (PDT)
Received: from [192.168.0.124] (c-67-170-243-110.hsd1.ca.comcast.net [67.170.243.110]) by smtp.webfaction.com (Postfix) with ESMTP id 5E096213ED44; Thu,  9 Aug 2012 10:47:15 -0500 (CDT)
Message-ID: <5023DB7D.9010803@grotto-networking.com>
Date: Thu, 09 Aug 2012 08:47:09 -0700
From: Greg Bernstein <gregb@grotto-networking.com>
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:14.0) Gecko/20120713 Thunderbird/14.0
MIME-Version: 1.0
To: Alia Atlas <akatlas@gmail.com>
References: <CC404D94.2D5D%tnadeau@juniper.net> <501B0D75.4090009@raszuk.net> <7E89A05A-CE4E-4FCF-81AB-8F39B42FBF8E@cisco.com> <501C128D.10809@cs.yale.edu> <501C2047.1000100@bell-labs.com> <501C3999.3080804@cs.yale.edu> <CAG4d1rdo3BuEukb9F0aR=wOkKFEM9_esaJVJvn+e_UndQ3jELg@mail.gmail.com>
In-Reply-To: <CAG4d1rdo3BuEukb9F0aR=wOkKFEM9_esaJVJvn+e_UndQ3jELg@mail.gmail.com>
Content-Type: multipart/alternative; boundary="------------080100050901050805090804"
Cc: idr@ietf.org, Volker Hilt <volker.hilt@bell-labs.com>, irs-discuss@ietf.org, James Kempf <james.kempf@ericsson.com>, IETF ALTO <alto@ietf.org>
Subject: Re: [alto] [irs-discuss] About ALTO, BGP-LS, IRS
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 09 Aug 2012 15:47:20 -0000

This is a multi-part message in MIME format.
--------------080100050901050805090804
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Some questions and comments on IRS, IDR, and ALTO...

Would IRS apply to GMPLS controlled boxes such as TDM (G.709, SDH) or 
WDM (WSON) network elements?
Setting state in a GMPLS controlled box can be rather trivial, i.e., 
just setting up a cross connect. The penalties for screwing up the state 
in a WDM box are rather terrible... Would IRS want state for all the 
LSP's traversing a box?

I'm a infrastructure plumber and trying to get folks to use the nifty 
optical control plane more. In GMPLS we have technology specific data 
models (SDH, G.709, WSON) that are fairly independent of specific 
routing protocols (OSPF, IS-IS) so these can help with the modeling 
aspect of IRS.  However, in the early days of GMPLS folks thought 
everything should be done in the same IGP instance, later the OSPF folks 
decided multiple instances were better since the service impacts of 
routing protocols is much different for circuit switched technologies 
than IP.

The model we are interested for in large-bandwidth use-cases in ALTO 
tries to abstract away as many layers as possible.  For example, suppose 
that the use-case is to set up a large IP pipe between mulitple data 
centers.  We could show an abstracted graph that helps guide paths 
choices via costs and bandwidth constraints.  The actual implementation 
technology could be IP-MPLS-G.709-MPLS-IP  or 
IP-OpenFlow-WSON-OpenFlow-IP  (here I just mean a simple OpenFlow 
controlled Ethernet switch that that can feed into our optical "express 
lane").
In other cases the data centers might want some special ethernet 
connectivity over optical pipes (a different type of service).

When I think of an IDR and sharing information between ASs then one may 
want to (or need to) keep (sub-IP) layer specific for compatibility, 
i.e., connecting the pipes between domains and adapting the higher layer 
protocols back out.  It's useful to look at publicly available service 
maps for examples, I like Level3's maps 
(http://maps.level3.com/default/#.UCFpjJYu9rh) --compare wavelength 
services to other services -- and CenturyLink 
(http://www.centurylink-business.com/demos/network-maps.html) -- note 
how wavelength service availability is limited compared to other 
services (our big pipe networks are generally simpler than other service 
networks) --.

It seems what BGP-LS would deliver to an ALTO server to digest and serve 
up in abstracted form could be very different from what might be shared 
between ASes.

Cheers
Greg


On 8/3/2012 4:57 PM, Alia Atlas wrote:
>
> I strongly agree that we need to start collecting the use-cases for 
> multiple abstraction levels as well as physical/virtual/private.
>
> Alia
>
> On Aug 3, 2012 4:51 PM, "Y. Richard Yang" <yry@cs.yale.edu 
> <mailto:yry@cs.yale.edu>> wrote:
>
>     Hi Volker,
>
>     On 8/3/12 12:02 PM, Volker Hilt wrote:
>
>
>         I believe that the capability to provide a simplified view on
>         the network topology to an application is a key feature rather
>         than a bug. Applications that want to have a view on network
>         topology don't need need a fine grained view on the topology
>         in most casts and benefit from having an abstracted view. This
>         will simplify processing for the application and enables
>         providers to control the exposure of details.
>
>     Agreed.
>
>         We've seen some numbers for topology data sizes in the
>         incremental updates presentation in the ALTO meeting, which
>         provide some insights into the amounts of data needed for
>         different topology sizes.
>
>         I like the idea of enabling an ALTO server to offer different
>         levels of details. This will enable a server to tailor
>         responses to the specific client. It will add complexity as
>         the ALTO server itself will have to maintain the most complex
>         topology it is offering and will have to keep this topology up
>         to date. But this is an interesting question for discussion in
>         the WG.
>
>     Glad that you also like the idea of different levels of details of
>     the network topology. If the ALTO Server is given a detailed
>     topology (constructed from multiple sources, such as routing feed,
>     LSP feed, configuration files), we can offer multiple topology
>     operators/aggregators to explore the detailed topology, according
>     to need and policy. A simple example operator is level (i.e.,
>     hierarchy such as at the area level), but one might specify other
>     operators (e.g., fisheye focus). There are studies on
>     representation of multi-level graphs that we can try to take
>     advantage of. This can be a subject for the group to explore.
>
>     We may need to study/collect use cases for multiple level
>     topology. One interesting example immediately coming to mind is
>     the Abstract Node concept (specify IP prefix/ASN) used in RSVP-TE.
>
>     Cheers,
>
>     Richard
>
>         Cheers,
>
>         Volker
>
>
>
>
>
>
>
>         On 03.08.2012 11:03, Y. Richard Yang wrote:
>
>             Hi Stefano,
>
>             Good post! I added the ALTO mailing list, given the
>             relevance. I hope
>             that this is OK cross posting.
>
>             First, a few comments on ALTO:
>
>             View granularity:
>
>             - ALTO currently defines two abstract network topology
>             data structures:
>             Network Map and Cost Map(s). More link-state oriented maps
>             (graphs),
>             with aggregations and transformations, can be added
>             efficiently to ALTO.
>             Some initial efforts are already on the way (e.g., see the
>             graph
>             representation proposal at page 9:
>             http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf).
>             Hence,
>             I see a natural next-step is for ALTO to provide a
>             link-state view to
>             external entities.
>
>             - It is a good comment on the level of details that ALTO
>             should
>             delivery. This is a good question for the ALTO working
>             group and the
>             community to discuss. I feel that ALTO should provide
>             multi-levels of
>             granularity of views, and we should discuss in the working
>             group.
>
>             Pull vs push delivery mechanism:
>
>             - There are more discussions on adding the incremental
>             update in ALTO.
>             Multiple mechanisms have been discussed. I feel that it is
>             the right
>             direction for ALTO.
>
>             Now let me understand the deployment model of BGP-LS. Your
>             explanation
>             on the issues of acquiring routing state is excellent. Let
>             me start by
>             understanding more details on the deployment model of
>             BGP-LS inside a
>             network:
>
>             - A set N_igp of BGP-LS instances are deployed to collect
>             IGP info. You
>             need multiple instances because IGP needs direct connectivity
>             (adjacency). Each instance here receives (potentially
>             multiple) IGP
>             updates and streams (relays) to an another (remote)
>             entity, which I
>             assume is a master BGP-LS instance. So each of these N_igp
>             instances is
>             IGP-in and BGP-LS out. This appears to be shown in Figure 1 of
>             draft-gredler-idr-ls-distribution.
>
>             - A set N_egp of BGP-LS instances are deployed to collect
>             BGP feeds. You
>             also may need multiple instances because the network does
>             not want to
>             see aggregated states but raw states. These instances will
>             feed to the
>             master BGP-LS as well.
>
>             - The master BGP-LS aggregates the multiple BGP-LS ins
>             (and maybe some
>             direct IGP/EGP ins) and pushes (after policy) to other
>             BGP-LS peers to
>             use: for example, an ALTO Server transforms/aggregates the
>             feed to
>             generate ALTO views (maps and graphs), and an PCE uses the
>             feed to
>             maintain its TED. One could even imagine that ALTO builds
>             a detailed TED
>             and feeds to PCE, but this beyond the scope of this
>             discussion. The
>             master BGP-LS is BGP-LS in and BGP-LS out. It is also
>             possible that the
>             master BGP-LS does not push to any other entities and
>             simply maintains
>             an internal DB for others to query.
>
>             Do I understand it correctly?
>
>             Now, we can take a look at more specifics of BGP-LS.
>
>             A first perspective is the semantics of the content. If
>             the objective is
>             to solve the aforementioned deployment issue, then an
>             alternative
>             solution is to introduce a simple LS update tunneling
>             protocol, where a
>             link-state proxy forwards LS messages to a collector. The
>             current design
>             of BGP-LS starts with such a feeling (i.e., an NLRI starts
>             with the
>             Protocol ID, which indicates it is from IS-IS level 1
>             IS-IS level 2,
>             OSPF, etc). However, the protocol appears to (try to) go
>             beyond simple
>             tunneling and introduces a common LS schema, by
>             converting/filtering
>             individual IGP LS messages to some common format. I feel
>             that it can be
>             helpful to first specify the schema (LS data model)
>             instead of the
>             specific encoding. For example, OSPF specifies LS Age, and
>             this is
>             filtered. (Please correct me if I missed it). On the other
>             hand, one can
>             think that some Age info can be helpful for one to
>             understand the
>             "freshness" of the LS. A problem studied in database is
>             heterogeneous
>             databases, to merge multiple data sources (IS-IS, OSPF,
>             etc) to a single
>             schema, and there can be many problems. If there is such a
>             study, please
>             do share a pointer.
>
>             A second perspective is using BGP as the transport. What
>             key features
>             from BGP do we really need (yes, weak-typed TLV encoding
>             offers a lot of
>             flexibility)? What features of BGP do we not need (e.g.,
>             BGP is a
>             routing protocol and hence builds in features to handle
>             convergence such
>             as dampening)? What may be missing (e.g., a capability of
>             pull or
>             filtering of push). I feel that these issues should be
>             discussed. If
>             they have already been discussed, please do share a
>             pointer, as I am
>             definitely a new comer.
>
>             Thanks!
>
>             Richard
>
>             On 8/2/12 11:54 PM, stefano previdi wrote:
>
>                 All,
>
>                 as co-author of both BGP-LS and ALTO drafts, I'd try
>                 to clarify a few
>                 things:
>
>                 ALTO has been designed in order to deliver to
>                 applications (through
>                 http/json):
>
>                 1. two maps representing the network topology in an
>                 abstracted view
>                 (or set of views through multiple maps). The map does
>                 not include
>                 the details of a link-state database and therefore
>                 have little use
>                 for any element that would need to retrieve from the
>                 network the
>                 detailed/complete network layer topology, for example:
>                 link
>                 addresses or link BW resources, etc. IOW: ALTO maps do
>                 not have
>                 the granularity of a link-state database and ALTO
>                 protocol is not
>                 designed to deliver such details.
>
>                 and/or
>
>                 2. Ranking services where a client sends a bunch of IP
>                 addresses in
>                 a message and the ALTO server replies by ranking these
>                 addresses
>                 based on their topological/network distance (or
>                 whatever criteria
>                 the ALTO server has been configured for). This is
>                 called: Endpoint
>                 Cost Service.
>
>                 When using ALTO maps, and the ALTO protocol being
>                 http/pull based,
>                 there's no such concept of unsolicited routing
>                 updates. An ALTO
>                 client is typically a browser that will pull the maps
>                 from an ALTO
>                 server using http. The ALTO server will make no effort
>                 to ensure the
>                 client has the latest view of the topology (i.e.: It's
>                 the role of the
>                 client to poll for new maps time to time).
>
>                 Now, in order for an ALTO server to deliver Maps or
>                 Ranking services,
>                 it needs to build some form of topology and in order
>                 to achieve this,
>                 it needs somehow to be fed by either the operator
>                 (configuration) or
>                 to receive dynamically topology information from the
>                 network (e.g.:
>                 ISIS/OSPF/BGP).
>
>                 Here we had two options:
>                 1. ALTO server to implement ISIS/OSPF/BGP and
>                 establish IGP adjacencies
>                 to ABRs or L1L2 routers in each area so to retrieve
>                 the LSDB from
>                 each area. In practice I know no SP willing/accepting
>                 to open their
>                 IGP to an ALTO server. Also, IGP requires direct
>                 connectivity
>                 (adjacency) so from an operation point of view is
>                 complex and not
>                 desired.
>                 2. Use a database distribution protocol running on top
>                 of a reliable
>                 transport layer that would allow an ALTO server to
>                 connect to a
>                 _single_ and _remote_ Route Reflector (i.e.: no need
>                 to be directly
>                 connected) and grab the whole network topology that
>                 will be updated
>                 using standard routing protocol mechanisms (i.e.:
>                 routing updates)
>                 and that would allow the operator to control (through
>                 policies and
>                 filters) what to distribute and to which server.
>
>                 Benefits: single (or dual at most for redundancy)
>                 connection for
>                 the ALTO server (rather than having a direct adjacency
>                 with each
>                 ABR) and, from the operator perspective, single point of
>                 distribution of network topology (easier to apply
>                 policies and
>                 control what you deliver). This is what BGP-LS is about.
>
>                 BGP-LS defines new AFI/SAFI for a new NLRI and
>                 attributes that convey
>                 link-state and, more generically, any type of topology
>                 information.
>                 The draft specifies NLRI and attributes that map
>                 whatever you can
>                 find in a link-state database.
>
>                 We currently have draft-gredler-idr-ls-distribution in
>                 the process
>                 (hopefully) to be adopted as WG item in IDR WG (so far
>                 we 're getting
>                 good support). We also have 3 implementations of BGP-LS.
>
>                 Deployment-wise: BGP-LS is not yet deployed, however,
>                 we already have
>                 deployments (I know at least two) where an ALTO server
>                 retrieves IP
>                 prefix information from remote BGP RRs (for ipv4). The
>                 same scheme
>                 will be used once BGP-LS will be deployed (so to say
>                 that it is not
>                 something that we invented after a couple of beers but
>                 corresponds to
>                 requirements for delivering network services to upper
>                 layers while
>                 still controlling efficiently what you distribute, to
>                 whom and in
>                 which form (note that, often, beers are anyway part of
>                 the process).
>
>                 More information here:
>                 http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02
>                 http://tools.ietf.org/html/draft-ietf-alto-protocol-12
>
>                 Hope this helps.
>
>                 s.
>
>
>
>                 On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:
>
>                     Tom,
>
>                         I agree that one of the top work items for
>                         this effort should be a
>                         standardized topology function, and one that
>                         is accessible via a
>                         non-routing protocol.
>
>                     So if the requirement is to have topology export
>                     via non-routing
>                     protocol then I think we should seriously revisit
>                     or repackage the
>                     draft-gredler-idr-ls-distribution-01 which works
>                     for for both OSPF
>                     and ISIS.
>
>                     However before that let's really understand the
>                     requirement why it
>                     must be exported via non-routing protocol ....
>                     Keep in mind that just
>                     to parse BGP UPDATE messages and retrieve
>                     interesting pieces out it
>                     it requires very little code rather then full BGP
>                     implementation.
>
>                     The particular feature I like about
>                     draft-gredler-idr-ls-distribution-01 is that it is
>                     read-only ;)
>
>                     R.
>
>                         I agree that one of the top work items for
>                         this effort should be a
>                         standardized topology function, and one that
>                         is accessible via a
>                         non-routing protocol. While not exactly "low
>                         hanging fruit", it is
>                         something that (to me) is a clear work item
>                         with clear goals that
>                         should
>                         be tackled straight away.
>
>                         --Tom
>
>
>
>                         On 8/2/12 3:24 PM, "James Kempf"
>                         <james.kempf@ericsson.com
>                         <mailto:james.kempf@ericsson.com>> wrote:
>
>                             So after seeing part of Alia's talk this
>                             morning (I had to leave in
>                             the
>                             middle unfortunately), I'd like to make a
>                             couple suggestions. There
>                             were
>                             a lot of ideas presented in the talk,
>                             enough for an entire IETF
>                             Area. I
>                             think to make tangible progress, the work
>                             needs to be focussed on a
>                             small
>                             subset that would be of immediate interest
>                             and usability.
>
>                             There are a couple areas that suggest
>                             themselves, but one that
>                             would be
>                             useful in work that I've been involved in
>                             is a standardized format for
>                             network topology representation and a
>                             protocol for exchanging it. The
>                             Onix OpenFlow controller has a network
>                             information base with a
>                             specialized format for network topology,
>                             and every OpenFlow controller
>                             requires this. Having a standardized way
>                             to represent it might
>                             foster a
>                             common topology database package. Another
>                             application is network
>                             management. Every network management
>                             system needs some kind of
>                             topology
>                             representation. Finally, though I am not
>                             an expert in PCE
>                             construction,
>                             it would seem to me that a PCE would need
>                             some kind of topology
>                             representation in order to perform path
>                             calculations. Having a way,for
>                             example, for the OpenFlow controller and
>                             the PCE to exchange topology
>                             information would be really useful. I
>                             would say to start with physical
>                             topology because that is fundamental, but
>                             make the format flexible
>                             enough
>                             to support
>                             virtual topology representation.
>
>                             jak
>                             _______________________________________________
>                             irs-discuss mailing list
>                             irs-discuss@ietf.org
>                             <mailto:irs-discuss@ietf.org>
>                             https://www.ietf.org/mailman/listinfo/irs-discuss
>
>                         _______________________________________________
>                         irs-discuss mailing list
>                         irs-discuss@ietf.org <mailto:irs-discuss@ietf.org>
>                         https://www.ietf.org/mailman/listinfo/irs-discuss
>
>
>                     _______________________________________________
>                     irs-discuss mailing list
>                     irs-discuss@ietf.org <mailto:irs-discuss@ietf.org>
>                     https://www.ietf.org/mailman/listinfo/irs-discuss
>
>                 _______________________________________________
>                 irs-discuss mailing list
>                 irs-discuss@ietf.org <mailto:irs-discuss@ietf.org>
>                 https://www.ietf.org/mailman/listinfo/irs-discuss
>
>
>             _______________________________________________
>             irs-discuss mailing list
>             irs-discuss@ietf.org <mailto:irs-discuss@ietf.org>
>             https://www.ietf.org/mailman/listinfo/irs-discuss
>
>     _______________________________________________
>     irs-discuss mailing list
>     irs-discuss@ietf.org <mailto:irs-discuss@ietf.org>
>     https://www.ietf.org/mailman/listinfo/irs-discuss
>
>
>
> _______________________________________________
> alto mailing list
> alto@ietf.org
> https://www.ietf.org/mailman/listinfo/alto


-- 
===================================================
Dr Greg Bernstein, Grotto Networking (510) 573-2237


--------------080100050901050805090804
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <div class="moz-cite-prefix">Some questions and comments on IRS,
      IDR, and ALTO...<br>
      <br>
      Would IRS apply to GMPLS controlled boxes such as TDM (G.709, SDH)
      or WDM (WSON) network elements?
      <br>
      Setting state in a GMPLS controlled box can be rather trivial,
      i.e., just setting up a cross connect. The penalties for screwing
      up the state in a WDM box are rather terrible...
      Would IRS want state for all the LSP's traversing a box?<br>
      <br>
      I'm a infrastructure plumber and trying to get folks to use the
      nifty optical control plane more. In GMPLS we have technology
      specific data models (SDH, G.709, WSON) that are fairly
      independent of specific routing protocols (OSPF, IS-IS) so these
      can help with the modeling aspect of IRS.&nbsp; However, in the early
      days of GMPLS folks thought everything should be done in the same
      IGP instance, later the OSPF folks decided multiple instances were
      better since the service impacts of routing protocols is much
      different for circuit switched technologies than IP.<br>
      <br>
      The model we are interested for in large-bandwidth use-cases in
      ALTO tries to abstract away as many layers as possible.&nbsp; For
      example, suppose that the use-case is to set up a large IP pipe
      between mulitple data centers.&nbsp; We could show an abstracted graph
      that helps guide paths choices via costs and bandwidth
      constraints.&nbsp; The actual implementation technology could be
      IP-MPLS-G.709-MPLS-IP&nbsp; or IP-OpenFlow-WSON-OpenFlow-IP&nbsp; (here I
      just mean a simple OpenFlow controlled Ethernet switch that that
      can feed into our optical "express lane").
      <br>
      In other cases the data centers might want some special ethernet
      connectivity over optical pipes (a different type of service).
      <br>
      <br>
      When I think of an IDR and sharing information between ASs then
      one may want to (or need to) keep (sub-IP) layer specific for
      compatibility, i.e., connecting the pipes between domains and
      adapting the higher layer protocols back out.&nbsp; It's useful to look
      at publicly available service maps for examples, I like Level3's
      maps (<a class="moz-txt-link-freetext"
        href="http://maps.level3.com/default/#.UCFpjJYu9rh">http://maps.level3.com/default/#.UCFpjJYu9rh</a>)
      --compare wavelength services to other services -- and CenturyLink
      (<a class="moz-txt-link-freetext"
        href="http://www.centurylink-business.com/demos/network-maps.html">http://www.centurylink-business.com/demos/network-maps.html</a>)
      -- note how wavelength service availability is limited compared to
      other services (our big pipe networks are generally simpler than
      other service networks) --.
      <br>
      <br>
      It seems what BGP-LS would deliver to an ALTO server to digest and
      serve up in abstracted form could be very different from what
      might be shared between ASes.&nbsp;&nbsp; <br>
      <br>
      Cheers
      <br>
      Greg
      <br>
      <br>
      <br>
      On 8/3/2012 4:57 PM, Alia Atlas wrote:<br>
    </div>
    <blockquote
cite="mid:CAG4d1rdo3BuEukb9F0aR=wOkKFEM9_esaJVJvn+e_UndQ3jELg@mail.gmail.com"
      type="cite">
      <p>I strongly agree that we need to start collecting the use-cases
        for multiple abstraction levels as well as
        physical/virtual/private.</p>
      <p>Alia</p>
      <div class="gmail_quote">On Aug 3, 2012 4:51 PM, "Y. Richard Yang"
        &lt;<a moz-do-not-send="true" href="mailto:yry@cs.yale.edu">yry@cs.yale.edu</a>&gt;
        wrote:<br type="attribution">
        <blockquote class="gmail_quote" style="margin:0 0 0
          .8ex;border-left:1px #ccc solid;padding-left:1ex">
          Hi Volker,<br>
          <br>
          On 8/3/12 12:02 PM, Volker Hilt wrote:<br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            <br>
            I believe that the capability to provide a simplified view
            on the network topology to an application is a key feature
            rather than a bug. Applications that want to have a view on
            network topology don't need need a fine grained view on the
            topology in most casts and benefit from having an abstracted
            view. This will simplify processing for the application and
            enables providers to control the exposure of details. <br>
          </blockquote>
          Agreed.<br>
          <br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            We've seen some numbers for topology data sizes in the
            incremental updates presentation in the ALTO meeting, which
            provide some insights into the amounts of data needed for
            different topology sizes.<br>
            <br>
            I like the idea of enabling an ALTO server to offer
            different levels of details. This will enable a server to
            tailor responses to the specific client. It will add
            complexity as the ALTO server itself will have to maintain
            the most complex topology it is offering and will have to
            keep this topology up to date. But this is an interesting
            question for discussion in the WG.<br>
            <br>
          </blockquote>
          Glad that you also like the idea of different levels of
          details of the network topology. If the ALTO Server is given a
          detailed topology (constructed from multiple sources, such as
          routing feed, LSP feed, configuration files), we can offer
          multiple topology operators/aggregators to explore the
          detailed topology, according to need and policy. A simple
          example operator is level (i.e., hierarchy such as at the area
          level), but one might specify other operators (e.g., fisheye
          focus). There are studies on representation of multi-level
          graphs that we can try to take advantage of. This can be a
          subject for the group to explore.<br>
          <br>
          We may need to study/collect use cases for multiple level
          topology. One interesting example immediately coming to mind
          is the Abstract Node concept (specify IP prefix/ASN) used in
          RSVP-TE.<br>
          <br>
          Cheers,<br>
          <br>
          Richard<br>
          <br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            Cheers,<br>
            <br>
            Volker<br>
            <br>
            <br>
            <br>
            <br>
            <br>
            <br>
            <br>
            On 03.08.2012 11:03, Y. Richard Yang wrote:<br>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              Hi Stefano,<br>
              <br>
              Good post! I added the ALTO mailing list, given the
              relevance. I hope<br>
              that this is OK cross posting.<br>
              <br>
              First, a few comments on ALTO:<br>
              <br>
              View granularity:<br>
              <br>
              - ALTO currently defines two abstract network topology
              data structures:<br>
              Network Map and Cost Map(s). More link-state oriented maps
              (graphs),<br>
              with aggregations and transformations, can be added
              efficiently to ALTO.<br>
              Some initial efforts are already on the way (e.g., see the
              graph<br>
              representation proposal at page 9:<br>
              <a moz-do-not-send="true"
                href="http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf"
                target="_blank">http://www.ietf.org/proceedings/84/slides/slides-84-alto-1.pdf</a>).
              Hence,<br>
              I see a natural next-step is for ALTO to provide a
              link-state view to<br>
              external entities.<br>
              <br>
              - It is a good comment on the level of details that ALTO
              should<br>
              delivery. This is a good question for the ALTO working
              group and the<br>
              community to discuss. I feel that ALTO should provide
              multi-levels of<br>
              granularity of views, and we should discuss in the working
              group.<br>
              <br>
              Pull vs push delivery mechanism:<br>
              <br>
              - There are more discussions on adding the incremental
              update in ALTO.<br>
              Multiple mechanisms have been discussed. I feel that it is
              the right<br>
              direction for ALTO.<br>
              <br>
              Now let me understand the deployment model of BGP-LS. Your
              explanation<br>
              on the issues of acquiring routing state is excellent. Let
              me start by<br>
              understanding more details on the deployment model of
              BGP-LS inside a<br>
              network:<br>
              <br>
              - A set N_igp of BGP-LS instances are deployed to collect
              IGP info. You<br>
              need multiple instances because IGP needs direct
              connectivity<br>
              (adjacency). Each instance here receives (potentially
              multiple) IGP<br>
              updates and streams (relays) to an another (remote)
              entity, which I<br>
              assume is a master BGP-LS instance. So each of these N_igp
              instances is<br>
              IGP-in and BGP-LS out. This appears to be shown in Figure
              1 of<br>
              draft-gredler-idr-ls-distribution.<br>
              <br>
              - A set N_egp of BGP-LS instances are deployed to collect
              BGP feeds. You<br>
              also may need multiple instances because the network does
              not want to<br>
              see aggregated states but raw states. These instances will
              feed to the<br>
              master BGP-LS as well.<br>
              <br>
              - The master BGP-LS aggregates the multiple BGP-LS ins
              (and maybe some<br>
              direct IGP/EGP ins) and pushes (after policy) to other
              BGP-LS peers to<br>
              use: for example, an ALTO Server transforms/aggregates the
              feed to<br>
              generate ALTO views (maps and graphs), and an PCE uses the
              feed to<br>
              maintain its TED. One could even imagine that ALTO builds
              a detailed TED<br>
              and feeds to PCE, but this beyond the scope of this
              discussion. The<br>
              master BGP-LS is BGP-LS in and BGP-LS out. It is also
              possible that the<br>
              master BGP-LS does not push to any other entities and
              simply maintains<br>
              an internal DB for others to query.<br>
              <br>
              Do I understand it correctly?<br>
              <br>
              Now, we can take a look at more specifics of BGP-LS.<br>
              <br>
              A first perspective is the semantics of the content. If
              the objective is<br>
              to solve the aforementioned deployment issue, then an
              alternative<br>
              solution is to introduce a simple LS update tunneling
              protocol, where a<br>
              link-state proxy forwards LS messages to a collector. The
              current design<br>
              of BGP-LS starts with such a feeling (i.e., an NLRI starts
              with the<br>
              Protocol ID, which indicates it is from IS-IS level 1
              IS-IS level 2,<br>
              OSPF, etc). However, the protocol appears to (try to) go
              beyond simple<br>
              tunneling and introduces a common LS schema, by
              converting/filtering<br>
              individual IGP LS messages to some common format. I feel
              that it can be<br>
              helpful to first specify the schema (LS data model)
              instead of the<br>
              specific encoding. For example, OSPF specifies LS Age, and
              this is<br>
              filtered. (Please correct me if I missed it). On the other
              hand, one can<br>
              think that some Age info can be helpful for one to
              understand the<br>
              "freshness" of the LS. A problem studied in database is
              heterogeneous<br>
              databases, to merge multiple data sources (IS-IS, OSPF,
              etc) to a single<br>
              schema, and there can be many problems. If there is such a
              study, please<br>
              do share a pointer.<br>
              <br>
              A second perspective is using BGP as the transport. What
              key features<br>
              from BGP do we really need (yes, weak-typed TLV encoding
              offers a lot of<br>
              flexibility)? What features of BGP do we not need (e.g.,
              BGP is a<br>
              routing protocol and hence builds in features to handle
              convergence such<br>
              as dampening)? What may be missing (e.g., a capability of
              pull or<br>
              filtering of push). I feel that these issues should be
              discussed. If<br>
              they have already been discussed, please do share a
              pointer, as I am<br>
              definitely a new comer.<br>
              <br>
              Thanks!<br>
              <br>
              Richard<br>
              <br>
              On 8/2/12 11:54 PM, stefano previdi wrote:<br>
              <blockquote class="gmail_quote" style="margin:0 0 0
                .8ex;border-left:1px #ccc solid;padding-left:1ex">
                All,<br>
                <br>
                as co-author of both BGP-LS and ALTO drafts, I'd try to
                clarify a few<br>
                things:<br>
                <br>
                ALTO has been designed in order to deliver to
                applications (through<br>
                http/json):<br>
                <br>
                1. two maps representing the network topology in an
                abstracted view<br>
                (or set of views through multiple maps). The map does
                not include<br>
                the details of a link-state database and therefore have
                little use<br>
                for any element that would need to retrieve from the
                network the<br>
                detailed/complete network layer topology, for example:
                link<br>
                addresses or link BW resources, etc. IOW: ALTO maps do
                not have<br>
                the granularity of a link-state database and ALTO
                protocol is not<br>
                designed to deliver such details.<br>
                <br>
                and/or<br>
                <br>
                2. Ranking services where a client sends a bunch of IP
                addresses in<br>
                a message and the ALTO server replies by ranking these
                addresses<br>
                based on their topological/network distance (or whatever
                criteria<br>
                the ALTO server has been configured for). This is
                called: Endpoint<br>
                Cost Service.<br>
                <br>
                When using ALTO maps, and the ALTO protocol being
                http/pull based,<br>
                there's no such concept of unsolicited routing updates.
                An ALTO<br>
                client is typically a browser that will pull the maps
                from an ALTO<br>
                server using http. The ALTO server will make no effort
                to ensure the<br>
                client has the latest view of the topology (i.e.: It's
                the role of the<br>
                client to poll for new maps time to time).<br>
                <br>
                Now, in order for an ALTO server to deliver Maps or
                Ranking services,<br>
                it needs to build some form of topology and in order to
                achieve this,<br>
                it needs somehow to be fed by either the operator
                (configuration) or<br>
                to receive dynamically topology information from the
                network (e.g.:<br>
                ISIS/OSPF/BGP).<br>
                <br>
                Here we had two options:<br>
                1. ALTO server to implement ISIS/OSPF/BGP and establish
                IGP adjacencies<br>
                to ABRs or L1L2 routers in each area so to retrieve the
                LSDB from<br>
                each area. In practice I know no SP willing/accepting to
                open their<br>
                IGP to an ALTO server. Also, IGP requires direct
                connectivity<br>
                (adjacency) so from an operation point of view is
                complex and not<br>
                desired.<br>
                2. Use a database distribution protocol running on top
                of a reliable<br>
                transport layer that would allow an ALTO server to
                connect to a<br>
                _single_ and _remote_ Route Reflector (i.e.: no need to
                be directly<br>
                connected) and grab the whole network topology that will
                be updated<br>
                using standard routing protocol mechanisms (i.e.:
                routing updates)<br>
                and that would allow the operator to control (through
                policies and<br>
                filters) what to distribute and to which server.<br>
                <br>
                Benefits: single (or dual at most for redundancy)
                connection for<br>
                the ALTO server (rather than having a direct adjacency
                with each<br>
                ABR) and, from the operator perspective, single point of<br>
                distribution of network topology (easier to apply
                policies and<br>
                control what you deliver). This is what BGP-LS is about.<br>
                <br>
                BGP-LS defines new AFI/SAFI for a new NLRI and
                attributes that convey<br>
                link-state and, more generically, any type of topology
                information.<br>
                The draft specifies NLRI and attributes that map
                whatever you can<br>
                find in a link-state database.<br>
                <br>
                We currently have draft-gredler-idr-ls-distribution in
                the process<br>
                (hopefully) to be adopted as WG item in IDR WG (so far
                we 're getting<br>
                good support). We also have 3 implementations of BGP-LS.<br>
                <br>
                Deployment-wise: BGP-LS is not yet deployed, however, we
                already have<br>
                deployments (I know at least two) where an ALTO server
                retrieves IP<br>
                prefix information from remote BGP RRs (for ipv4). The
                same scheme<br>
                will be used once BGP-LS will be deployed (so to say
                that it is not<br>
                something that we invented after a couple of beers but
                corresponds to<br>
                requirements for delivering network services to upper
                layers while<br>
                still controlling efficiently what you distribute, to
                whom and in<br>
                which form (note that, often, beers are anyway part of
                the process).<br>
                <br>
                More information here:<br>
                <a moz-do-not-send="true"
                  href="http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02"
                  target="_blank">http://tools.ietf.org/html/draft-gredler-idr-ls-distribution-02</a><br>
                <a moz-do-not-send="true"
                  href="http://tools.ietf.org/html/draft-ietf-alto-protocol-12"
                  target="_blank">http://tools.ietf.org/html/draft-ietf-alto-protocol-12</a><br>
                <br>
                Hope this helps.<br>
                <br>
                s.<br>
                <br>
                <br>
                <br>
                On Aug 3, 2012, at 1:29 AM, Robert Raszuk wrote:<br>
                <blockquote class="gmail_quote" style="margin:0 0 0
                  .8ex;border-left:1px #ccc solid;padding-left:1ex">
                  Tom,<br>
                  <br>
                  <blockquote class="gmail_quote" style="margin:0 0 0
                    .8ex;border-left:1px #ccc solid;padding-left:1ex">
                    I agree that one of the top work items for this
                    effort should be a<br>
                    standardized topology function, and one that is
                    accessible via a<br>
                    non-routing protocol.<br>
                  </blockquote>
                  So if the requirement is to have topology export via
                  non-routing<br>
                  protocol then I think we should seriously revisit or
                  repackage the<br>
                  draft-gredler-idr-ls-distribution-01 which works for
                  for both OSPF<br>
                  and ISIS.<br>
                  <br>
                  However before that let's really understand the
                  requirement why it<br>
                  must be exported via non-routing protocol .... Keep in
                  mind that just<br>
                  to parse BGP UPDATE messages and retrieve interesting
                  pieces out it<br>
                  it requires very little code rather then full BGP
                  implementation.<br>
                  <br>
                  The particular feature I like about<br>
                  draft-gredler-idr-ls-distribution-01 is that it is
                  read-only ;)<br>
                  <br>
                  R.<br>
                  <br>
                  <blockquote class="gmail_quote" style="margin:0 0 0
                    .8ex;border-left:1px #ccc solid;padding-left:1ex">
                    I agree that one of the top work items for this
                    effort should be a<br>
                    standardized topology function, and one that is
                    accessible via a<br>
                    non-routing protocol. While not exactly "low hanging
                    fruit", it is<br>
                    something that (to me) is a clear work item with
                    clear goals that<br>
                    should<br>
                    be tackled straight away.<br>
                    <br>
                    --Tom<br>
                    <br>
                    <br>
                    <br>
                    On 8/2/12 3:24 PM, "James Kempf" &lt;<a
                      moz-do-not-send="true"
                      href="mailto:james.kempf@ericsson.com"
                      target="_blank">james.kempf@ericsson.com</a>&gt;
                    wrote:<br>
                    <br>
                    <blockquote class="gmail_quote" style="margin:0 0 0
                      .8ex;border-left:1px #ccc solid;padding-left:1ex">
                      So after seeing part of Alia's talk this morning
                      (I had to leave in<br>
                      the<br>
                      middle unfortunately), I'd like to make a couple
                      suggestions. There<br>
                      were<br>
                      a lot of ideas presented in the talk, enough for
                      an entire IETF<br>
                      Area. I<br>
                      think to make tangible progress, the work needs to
                      be focussed on a<br>
                      small<br>
                      subset that would be of immediate interest and
                      usability.<br>
                      <br>
                      There are a couple areas that suggest themselves,
                      but one that<br>
                      would be<br>
                      useful in work that I've been involved in is a
                      standardized format for<br>
                      network topology representation and a protocol for
                      exchanging it. The<br>
                      Onix OpenFlow controller has a network information
                      base with a<br>
                      specialized format for network topology, and every
                      OpenFlow controller<br>
                      requires this. Having a standardized way to
                      represent it might<br>
                      foster a<br>
                      common topology database package. Another
                      application is network<br>
                      management. Every network management system needs
                      some kind of<br>
                      topology<br>
                      representation. Finally, though I am not an expert
                      in PCE<br>
                      construction,<br>
                      it would seem to me that a PCE would need some
                      kind of topology<br>
                      representation in order to perform path
                      calculations. Having a way,for<br>
                      example, for the OpenFlow controller and the PCE
                      to exchange topology<br>
                      information would be really useful. I would say to
                      start with physical<br>
                      topology because that is fundamental, but make the
                      format flexible<br>
                      enough<br>
                      to support<br>
                      virtual topology representation.<br>
                      <br>
                      jak<br>
                      _______________________________________________<br>
                      irs-discuss mailing list<br>
                      <a moz-do-not-send="true"
                        href="mailto:irs-discuss@ietf.org"
                        target="_blank">irs-discuss@ietf.org</a><br>
                      <a moz-do-not-send="true"
                        href="https://www.ietf.org/mailman/listinfo/irs-discuss"
                        target="_blank">https://www.ietf.org/mailman/listinfo/irs-discuss</a><br>
                    </blockquote>
                    _______________________________________________<br>
                    irs-discuss mailing list<br>
                    <a moz-do-not-send="true"
                      href="mailto:irs-discuss@ietf.org" target="_blank">irs-discuss@ietf.org</a><br>
                    <a moz-do-not-send="true"
                      href="https://www.ietf.org/mailman/listinfo/irs-discuss"
                      target="_blank">https://www.ietf.org/mailman/listinfo/irs-discuss</a><br>
                    <br>
                    <br>
                  </blockquote>
                  _______________________________________________<br>
                  irs-discuss mailing list<br>
                  <a moz-do-not-send="true"
                    href="mailto:irs-discuss@ietf.org" target="_blank">irs-discuss@ietf.org</a><br>
                  <a moz-do-not-send="true"
                    href="https://www.ietf.org/mailman/listinfo/irs-discuss"
                    target="_blank">https://www.ietf.org/mailman/listinfo/irs-discuss</a><br>
                  <br>
                </blockquote>
                _______________________________________________<br>
                irs-discuss mailing list<br>
                <a moz-do-not-send="true"
                  href="mailto:irs-discuss@ietf.org" target="_blank">irs-discuss@ietf.org</a><br>
                <a moz-do-not-send="true"
                  href="https://www.ietf.org/mailman/listinfo/irs-discuss"
                  target="_blank">https://www.ietf.org/mailman/listinfo/irs-discuss</a><br>
              </blockquote>
              <br>
              _______________________________________________<br>
              irs-discuss mailing list<br>
              <a moz-do-not-send="true"
                href="mailto:irs-discuss@ietf.org" target="_blank">irs-discuss@ietf.org</a><br>
              <a moz-do-not-send="true"
                href="https://www.ietf.org/mailman/listinfo/irs-discuss"
                target="_blank">https://www.ietf.org/mailman/listinfo/irs-discuss</a><br>
            </blockquote>
          </blockquote>
          _______________________________________________<br>
          irs-discuss mailing list<br>
          <a moz-do-not-send="true" href="mailto:irs-discuss@ietf.org"
            target="_blank">irs-discuss@ietf.org</a><br>
          <a moz-do-not-send="true"
            href="https://www.ietf.org/mailman/listinfo/irs-discuss"
            target="_blank">https://www.ietf.org/mailman/listinfo/irs-discuss</a><br>
        </blockquote>
      </div>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
      <pre wrap="">_______________________________________________
alto mailing list
<a class="moz-txt-link-abbreviated" href="mailto:alto@ietf.org">alto@ietf.org</a>
<a class="moz-txt-link-freetext" href="https://www.ietf.org/mailman/listinfo/alto">https://www.ietf.org/mailman/listinfo/alto</a>
</pre>
    </blockquote>
    <br>
    <br>
    <pre class="moz-signature" cols="72">-- 
===================================================
Dr Greg Bernstein, Grotto Networking (510) 573-2237

</pre>
  </body>
</html>

--------------080100050901050805090804--

From vkg@bell-labs.com  Fri Aug 10 07:07:05 2012
Return-Path: <vkg@bell-labs.com>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 4066021F8522 for <alto@ietfa.amsl.com>; Fri, 10 Aug 2012 07:07:05 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -107.188
X-Spam-Level: 
X-Spam-Status: No, score=-107.188 tagged_above=-999 required=5 tests=[AWL=-2.078, BAYES_05=-1.11, RCVD_IN_DNSWL_MED=-4, USER_IN_WHITELIST=-100]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 2EUK1muTM34J for <alto@ietfa.amsl.com>; Fri, 10 Aug 2012 07:07:04 -0700 (PDT)
Received: from ihemail4.lucent.com (ihemail4.lucent.com [135.245.0.39]) by ietfa.amsl.com (Postfix) with ESMTP id 7AB3821F84DA for <alto@ietf.org>; Fri, 10 Aug 2012 07:07:04 -0700 (PDT)
Received: from usnavsmail4.ndc.alcatel-lucent.com (usnavsmail4.ndc.alcatel-lucent.com [135.3.39.12]) by ihemail4.lucent.com (8.13.8/IER-o) with ESMTP id q7ADuxXh024471 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=FAIL) for <alto@ietf.org>; Fri, 10 Aug 2012 08:57:04 -0500 (CDT)
Received: from umail.lucent.com (umail-ce2.ndc.lucent.com [135.3.40.63]) by usnavsmail4.ndc.alcatel-lucent.com (8.14.3/8.14.3/GMO) with ESMTP id q7ADuwhN023368 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NOT) for <alto@ietf.org>; Fri, 10 Aug 2012 08:56:59 -0500
Received: from shoonya.ih.lucent.com (vkg.lra.lucent.com [135.244.40.216]) by umail.lucent.com (8.13.8/TPES) with ESMTP id q7ADuw6x021488 for <alto@ietf.org>; Fri, 10 Aug 2012 08:56:58 -0500 (CDT)
Message-ID: <5025132E.5040505@bell-labs.com>
Date: Fri, 10 Aug 2012 08:57:02 -0500
From: "Vijay K. Gurbani" <vkg@bell-labs.com>
Organization: Bell Laboratories, Alcatel-Lucent
User-Agent: Mozilla/5.0 (X11; Linux i686; rv:13.0) Gecko/20120605 Thunderbird/13.0
MIME-Version: 1.0
To: alto <alto@ietf.org>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-Scanned-By: MIMEDefang 2.57 on 135.245.2.39
X-Scanned-By: MIMEDefang 2.64 on 135.3.39.12
Subject: [alto] Draft minutes for Vancouver IETF
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 10 Aug 2012 14:07:05 -0000

Folks: The draft minutes for the Vancouver IETF are now available in the
meetings material (please see
http://www.ietf.org/proceedings/84/minutes/minutes-84-alto)

A round of thanks to Brian Trammell and Fabio Picconi for taking the
notes.

Please let me and Enrico know if any of the links to the MP3 files or
Meetecho recordings do not work.  And, please post any corrections to
the list for inclusion in the final minutes.

Thanks,

- vijay
-- 
Vijay K. Gurbani, Bell Laboratories, Alcatel-Lucent
1960 Lucent Lane, Rm. 9C-533, Naperville, Illinois 60563 (USA)
Email: vkg@{bell-labs.com,acm.org} / vijay.gurbani@alcatel-lucent.com
Web:   http://ect.bell-labs.com/who/vkg/


From hannes@juniper.net  Thu Aug  9 08:59:00 2012
Return-Path: <hannes@juniper.net>
X-Original-To: alto@ietfa.amsl.com
Delivered-To: alto@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 216D621F87AA; Thu,  9 Aug 2012 08:59:00 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.19
X-Spam-Level: 
X-Spam-Status: No, score=-6.19 tagged_above=-999 required=5 tests=[AWL=-0.191,  BAYES_00=-2.599, J_CHICKENPOX_33=0.6, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id ayl4Gl3KJN+l; Thu,  9 Aug 2012 08:58:56 -0700 (PDT)
Received: from exprod7og117.obsmtp.com (exprod7og117.obsmtp.com [64.18.2.6]) by ietfa.amsl.com (Postfix) with ESMTP id D399E21F8795; Thu,  9 Aug 2012 08:58:34 -0700 (PDT)
Received: from P-EMHUB01-HQ.jnpr.net ([66.129.224.36]) (using TLSv1) by exprod7ob117.postini.com ([64.18.6.12]) with SMTP ID DSNKUCPeKB6yJoiQFLOdNyb8sRcVlkfKwFOz@postini.com; Thu, 09 Aug 2012 08:58:53 PDT
Received: from ubuntu (172.23.6.234) by P-EMHUB01-HQ.jnpr.net (172.24.192.33) with Microsoft SMTP Server id 8.3.213.0; Thu, 9 Aug 2012 08:57:42 -0700
Received: by ubuntu (Postfix, from userid 1000)	id 979512B2D6; Thu,  9 Aug 2012 17:57:45 +0200 (CEST)
Date: Thu, 9 Aug 2012 17:57:45 +0200
From: Hannes Gredler <hannes@juniper.net>
To: Greg Bernstein <gregb@grotto-networking.com>
Message-ID: <20120809155745.GA27488@juniper.net>
References: <CC404D94.2D5D%tnadeau@juniper.net> <501B0D75.4090009@raszuk.net> <7E89A05A-CE4E-4FCF-81AB-8F39B42FBF8E@cisco.com> <501C128D.10809@cs.yale.edu> <501C2047.1000100@bell-labs.com> <501C3999.3080804@cs.yale.edu> <CAG4d1rdo3BuEukb9F0aR=wOkKFEM9_esaJVJvn+e_UndQ3jELg@mail.gmail.com> <5023DB7D.9010803@grotto-networking.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <5023DB7D.9010803@grotto-networking.com>
User-Agent: Mutt/1.5.21 (2010-09-15)
X-Mailman-Approved-At: Fri, 10 Aug 2012 08:04:54 -0700
Cc: idr@ietf.org, Volker Hilt <volker.hilt@bell-labs.com>, irs-discuss@ietf.org, James Kempf <james.kempf@ericsson.com>, IETF ALTO <alto@ietf.org>, Alia Atlas <akatlas@gmail.com>
Subject: Re: [alto] [Idr]  [irs-discuss] About ALTO, BGP-LS, IRS
X-BeenThere: alto@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: "Application-Layer Traffic Optimization \(alto\) WG mailing list" <alto.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/alto>, <mailto:alto-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/alto>
List-Post: <mailto:alto@ietf.org>
List-Help: <mailto:alto-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/alto>, <mailto:alto-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 09 Aug 2012 15:59:00 -0000

hi greg,

see comments inline prefixed by HG>

On Thu, Aug 09, 2012 at 08:47:09AM -0700, Greg Bernstein wrote:
| Some questions and comments on IRS, IDR, and ALTO...
| 
| Would IRS apply to GMPLS controlled boxes such as TDM (G.709, SDH) or WDM
| (WSON) network elements?
| Setting state in a GMPLS controlled box can be rather trivial, i.e., just
| setting up a cross connect. The penalties for screwing up the state in a WDM
| box are rather terrible... Would IRS want state for all the LSP's traversing a
| box?
| 
| I'm a infrastructure plumber and trying to get folks to use the nifty optical
| control plane more. In GMPLS we have technology specific data models (SDH,
| G.709, WSON) that are fairly independent of specific routing protocols (OSPF,
| IS-IS) so these can help with the modeling aspect of IRS.  However, in the
| early days of GMPLS folks thought everything should be done in the same IGP
| instance, later the OSPF folks decided multiple instances were better since the
| service impacts of routing protocols is much different for circuit switched
| technologies than IP.
| 
| The model we are interested for in large-bandwidth use-cases in ALTO tries to
| abstract away as many layers as possible.  For example, suppose that the use-
| case is to set up a large IP pipe between mulitple data centers.  We could show
| an abstracted graph that helps guide paths choices via costs and bandwidth
| constraints.  The actual implementation technology could be IP-MPLS-G.709-MPLS-
| IP  or IP-OpenFlow-WSON-OpenFlow-IP  (here I just mean a simple OpenFlow
| controlled Ethernet switch that that can feed into our optical "express lane").
| 
| In other cases the data centers might want some special ethernet connectivity
| over optical pipes (a different type of service).
| 
| When I think of an IDR and sharing information between ASs then one may want to
| (or need to) keep (sub-IP) layer specific for compatibility, i.e., connecting
| the pipes between domains and adapting the higher layer protocols back out. 
| It's useful to look at publicly available service maps for examples, I like
| Level3's maps (http://maps.level3.com/default/#.UCFpjJYu9rh) --compare
| wavelength services to other services -- and CenturyLink (http://
| www.centurylink-business.com/demos/network-maps.html) -- note how wavelength
| service availability is limited compared to other services (our big pipe
| networks are generally simpler than other service networks) --.
| 
| It seems what BGP-LS would deliver to an ALTO server to digest and serve up in
| abstracted form could be very different from what might be shared between
| ASes.  

HG> BGP-LS feeding an L3 routing topology into an ALTO server is one particular
application of the protocol. as you may have noticed the encoding of the link entries
in BGP-LS is a superset of all IGP extensions most notably including the MI
(multiple instance) drafts. as such one could carry higher or lower-layer
graphs (e.g. graphs of application clusters / graphs of DWDM wavelengths).
sicne you are concerened about optical gear feel free to send text on the
GMPLS link attributes that you would like to have supported.


| On 8/3/2012 4:57 PM, Alia Atlas wrote:
|      I strongly agree that we need to start collecting the use-cases for
|      multiple abstraction levels as well as physical/virtual/private.
|      Alia
|      On Aug 3, 2012 4:51 PM, "Y. Richard Yang" <yry@cs.yale.edu> wrote:
|           Hi Volker,
| 
|           On 8/3/12 12:02 PM, Volker Hilt wrote:
| 
|                I believe that the capability to provide a
|                simplified view on the network topology to an
|                application is a key feature rather than a bug.
|                Applications that want to have a view on network
|                topology don't need need a fine grained view on
|                the topology in most casts and benefit from
|                having an abstracted view. This will simplify
|                processing for the application and enables
|                providers to control the exposure of details.
|           Agreed.
| 
|                We've seen some numbers for topology data sizes
|                in the incremental updates presentation in the
|                ALTO meeting, which provide some insights into
|                the amounts of data needed for different topology
|                sizes.
| 
|                I like the idea of enabling an ALTO server to
|                offer different levels of details. This will
|                enable a server to tailor responses to the
|                specific client. It will add complexity as the
|                ALTO server itself will have to maintain the most
|                complex topology it is offering and will have to
|                keep this topology up to date. But this is an
|                interesting question for discussion in the WG.
| 
|           Glad that you also like the idea of different levels of
|           details of the network topology. If the ALTO Server is
|           given a detailed topology (constructed from multiple
|           sources, such as routing feed, LSP feed, configuration
|           files), we can offer multiple topology operators/
|           aggregators to explore the detailed topology, according to
|           need and policy. A simple example operator is level (i.e.,
|           hierarchy such as at the area level), but one might specify
|           other operators (e.g., fisheye focus). There are studies on
|           representation of multi-level graphs that we can try to
|           take advantage of. This can be a subject for the group to
|           explore.
| 
|           We may need to study/collect use cases for multiple level
|           topology. One interesting example immediately coming to
|           mind is the Abstract Node concept (specify IP prefix/ASN)
|           used in RSVP-TE.
| 
|           Cheers,
| 
|           Richard
| 
|                Cheers,
| 
|                Volker
| 
| 
| 
| 
| 
| 
| 
|                On 03.08.2012 11:03, Y. Richard Yang wrote:
|                     Hi Stefano,
| 
|                     Good post! I added the ALTO mailing
|                     list, given the relevance. I hope
|                     that this is OK cross posting.
| 
|                     First, a few comments on ALTO:
| 
|                     View granularity:
| 
|                     - ALTO currently defines two abstract
|                     network topology data structures:
|                     Network Map and Cost Map(s). More link-
|                     state oriented maps (graphs),
|                     with aggregations and transformations,
|                     can be added efficiently to ALTO.
|                     Some initial efforts are already on the
|                     way (e.g., see the graph
|                     representation proposal at page 9:
|                     http://www.ietf.org/proceedings/84/
|                     slides/slides-84-alto-1.pdf). Hence,
|                     I see a natural next-step is for ALTO
|                     to provide a link-state view to
|                     external entities.
| 
|                     - It is a good comment on the level of
|                     details that ALTO should
|                     delivery. This is a good question for
|                     the ALTO working group and the
|                     community to discuss. I feel that ALTO
|                     should provide multi-levels of
|                     granularity of views, and we should
|                     discuss in the working group.
| 
|                     Pull vs push delivery mechanism:
| 
|                     - There are more discussions on adding
|                     the incremental update in ALTO.
|                     Multiple mechanisms have been
|                     discussed. I feel that it is the right
|                     direction for ALTO.
| 
|                     Now let me understand the deployment
|                     model of BGP-LS. Your explanation
|                     on the issues of acquiring routing
|                     state is excellent. Let me start by
|                     understanding more details on the
|                     deployment model of BGP-LS inside a
|                     network:
| 
|                     - A set N_igp of BGP-LS instances are
|                     deployed to collect IGP info. You
|                     need multiple instances because IGP
|                     needs direct connectivity
|                     (adjacency). Each instance here
|                     receives (potentially multiple) IGP
|                     updates and streams (relays) to an
|                     another (remote) entity, which I
|                     assume is a master BGP-LS instance. So
|                     each of these N_igp instances is
|                     IGP-in and BGP-LS out. This appears to
|                     be shown in Figure 1 of
|                     draft-gredler-idr-ls-distribution.
| 
|                     - A set N_egp of BGP-LS instances are
|                     deployed to collect BGP feeds. You
|                     also may need multiple instances
|                     because the network does not want to
|                     see aggregated states but raw states.
|                     These instances will feed to the
|                     master BGP-LS as well.
| 
|                     - The master BGP-LS aggregates the
|                     multiple BGP-LS ins (and maybe some
|                     direct IGP/EGP ins) and pushes (after
|                     policy) to other BGP-LS peers to
|                     use: for example, an ALTO Server
|                     transforms/aggregates the feed to
|                     generate ALTO views (maps and graphs),
|                     and an PCE uses the feed to
|                     maintain its TED. One could even
|                     imagine that ALTO builds a detailed TED
|                     and feeds to PCE, but this beyond the
|                     scope of this discussion. The
|                     master BGP-LS is BGP-LS in and BGP-LS
|                     out. It is also possible that the
|                     master BGP-LS does not push to any
|                     other entities and simply maintains
|                     an internal DB for others to query.
| 
|                     Do I understand it correctly?
| 
|                     Now, we can take a look at more
|                     specifics of BGP-LS.
| 
|                     A first perspective is the semantics of
|                     the content. If the objective is
|                     to solve the aforementioned deployment
|                     issue, then an alternative
|                     solution is to introduce a simple LS
|                     update tunneling protocol, where a
|                     link-state proxy forwards LS messages
|                     to a collector. The current design
|                     of BGP-LS starts with such a feeling
|                     (i.e., an NLRI starts with the
|                     Protocol ID, which indicates it is from
|                     IS-IS level 1 IS-IS level 2,
|                     OSPF, etc). However, the protocol
|                     appears to (try to) go beyond simple
|                     tunneling and introduces a common LS
|                     schema, by converting/filtering
|                     individual IGP LS messages to some
|                     common format. I feel that it can be
|                     helpful to first specify the schema (LS
|                     data model) instead of the
|                     specific encoding. For example, OSPF
|                     specifies LS Age, and this is
|                     filtered. (Please correct me if I
|                     missed it). On the other hand, one can
|                     think that some Age info can be helpful
|                     for one to understand the
|                     "freshness" of the LS. A problem
|                     studied in database is heterogeneous
|                     databases, to merge multiple data
|                     sources (IS-IS, OSPF, etc) to a single
|                     schema, and there can be many problems.
|                     If there is such a study, please
|                     do share a pointer.
| 
|                     A second perspective is using BGP as
|                     the transport. What key features
|                     from BGP do we really need (yes, weak-
|                     typed TLV encoding offers a lot of
|                     flexibility)? What features of BGP do
|                     we not need (e.g., BGP is a
|                     routing protocol and hence builds in
|                     features to handle convergence such
|                     as dampening)? What may be missing
|                     (e.g., a capability of pull or
|                     filtering of push). I feel that these
|                     issues should be discussed. If
|                     they have already been discussed,
|                     please do share a pointer, as I am
|                     definitely a new comer.
| 
|                     Thanks!
| 
|                     Richard
| 
|                     On 8/2/12 11:54 PM, stefano previdi
|                     wrote:
|                          All,
| 
|                          as co-author of both BGP-LS
|                          and ALTO drafts, I'd try to
|                          clarify a few
|                          things:
| 
|                          ALTO has been designed in
|                          order to deliver to
|                          applications (through
|                          http/json):
| 
|                          1. two maps representing the
|                          network topology in an
|                          abstracted view
|                          (or set of views through
|                          multiple maps). The map does
|                          not include
|                          the details of a link-state
|                          database and therefore have
|                          little use
|                          for any element that would
|                          need to retrieve from the
|                          network the
|                          detailed/complete network
|                          layer topology, for example:
|                          link
|                          addresses or link BW
|                          resources, etc. IOW: ALTO
|                          maps do not have
|                          the granularity of a link-
|                          state database and ALTO
|                          protocol is not
|                          designed to deliver such
|                          details.
| 
|                          and/or
| 
|                          2. Ranking services where a
|                          client sends a bunch of IP
|                          addresses in
|                          a message and the ALTO server
|                          replies by ranking these
|                          addresses
|                          based on their topological/
|                          network distance (or whatever
|                          criteria
|                          the ALTO server has been
|                          configured for). This is
|                          called: Endpoint
|                          Cost Service.
| 
|                          When using ALTO maps, and the
|                          ALTO protocol being http/pull
|                          based,
|                          there's no such concept of
|                          unsolicited routing updates.
|                          An ALTO
|                          client is typically a browser
|                          that will pull the maps from
|                          an ALTO
|                          server using http. The ALTO
|                          server will make no effort to
|                          ensure the
|                          client has the latest view of
|                          the topology (i.e.: It's the
|                          role of the
|                          client to poll for new maps
|                          time to time).
| 
|                          Now, in order for an ALTO
|                          server to deliver Maps or
|                          Ranking services,
|                          it needs to build some form
|                          of topology and in order to
|                          achieve this,
|                          it needs somehow to be fed by
|                          either the operator
|                          (configuration) or
|                          to receive dynamically
|                          topology information from the
|                          network (e.g.:
|                          ISIS/OSPF/BGP).
| 
|                          Here we had two options:
|                          1. ALTO server to implement
|                          ISIS/OSPF/BGP and establish
|                          IGP adjacencies
|                          to ABRs or L1L2 routers in
|                          each area so to retrieve the
|                          LSDB from
|                          each area. In practice I know
|                          no SP willing/accepting to
|                          open their
|                          IGP to an ALTO server. Also,
|                          IGP requires direct
|                          connectivity
|                          (adjacency) so from an
|                          operation point of view is
|                          complex and not
|                          desired.
|                          2. Use a database
|                          distribution protocol running
|                          on top of a reliable
|                          transport layer that would
|                          allow an ALTO server to
|                          connect to a
|                          _single_ and _remote_ Route
|                          Reflector (i.e.: no need to
|                          be directly
|                          connected) and grab the whole
|                          network topology that will be
|                          updated
|                          using standard routing
|                          protocol mechanisms (i.e.:
|                          routing updates)
|                          and that would allow the
|                          operator to control (through
|                          policies and
|                          filters) what to distribute
|                          and to which server.
| 
|                          Benefits: single (or dual at
|                          most for redundancy)
|                          connection for
|                          the ALTO server (rather than
|                          having a direct adjacency
|                          with each
|                          ABR) and, from the operator
|                          perspective, single point of
|                          distribution of network
|                          topology (easier to apply
|                          policies and
|                          control what you deliver).
|                          This is what BGP-LS is about.
| 
|                          BGP-LS defines new AFI/SAFI
|                          for a new NLRI and attributes
|                          that convey
|                          link-state and, more
|                          generically, any type of
|                          topology information.
|                          The draft specifies NLRI and
|                          attributes that map whatever
|                          you can
|                          find in a link-state
|                          database.
| 
|                          We currently have draft-
|                          gredler-idr-ls-distribution
|                          in the process
|                          (hopefully) to be adopted as
|                          WG item in IDR WG (so far we
|                          're getting
|                          good support). We also have 3
|                          implementations of BGP-LS.
| 
|                          Deployment-wise: BGP-LS is
|                          not yet deployed, however, we
|                          already have
|                          deployments (I know at least
|                          two) where an ALTO server
|                          retrieves IP
|                          prefix information from
|                          remote BGP RRs (for ipv4).
|                          The same scheme
|                          will be used once BGP-LS will
|                          be deployed (so to say that
|                          it is not
|                          something that we invented
|                          after a couple of beers but
|                          corresponds to
|                          requirements for delivering
|                          network services to upper
|                          layers while
|                          still controlling efficiently
|                          what you distribute, to whom
|                          and in
|                          which form (note that, often,
|                          beers are anyway part of the
|                          process).
| 
|                          More information here:
|                          http://tools.ietf.org/html/
|                          draft-gredler-idr-ls-
|                          distribution-02
|                          http://tools.ietf.org/html/
|                          draft-ietf-alto-protocol-12
| 
|                          Hope this helps.
| 
|                          s.
| 
| 
| 
|                          On Aug 3, 2012, at 1:29 AM,
|                          Robert Raszuk wrote:
|                               Tom,
| 
|                                    I agree
|                                    that one
|                                    of the top
|                                    work items
|                                    for this
|                                    effort
|                                    should be
|                                    a
|                                    standardized
|                                    topology
|                                    function,
|                                    and one
|                                    that is
|                                    accessible
|                                    via a
|                                    non-
|                                    routing
|                                    protocol.
|                               So if the
|                               requirement is to
|                               have topology
|                               export via non-
|                               routing
|                               protocol then I
|                               think we should
|                               seriously revisit
|                               or repackage the
|                               draft-gredler-idr-
|                               ls-distribution-01
|                               which works for for
|                               both OSPF
|                               and ISIS.
| 
|                               However before that
|                               let's really
|                               understand the
|                               requirement why it
|                               must be exported
|                               via non-routing
|                               protocol .... Keep
|                               in mind that just
|                               to parse BGP UPDATE
|                               messages and
|                               retrieve
|                               interesting pieces
|                               out it
|                               it requires very
|                               little code rather
|                               then full BGP
|                               implementation.
| 
|                               The particular
|                               feature I like
|                               about
|                               draft-gredler-idr-
|                               ls-distribution-01
|                               is that it is read-
|                               only ;)
| 
|                               R.
| 
|                                    I agree
|                                    that one
|                                    of the top
|                                    work items
|                                    for this
|                                    effort
|                                    should be
|                                    a
|                                    standardized
|                                    topology
|                                    function,
|                                    and one
|                                    that is
|                                    accessible
|                                    via a
|                                    non-
|                                    routing
|                                    protocol.
|                                    While not
|                                    exactly
|                                    "low
|                                    hanging
|                                    fruit", it
|                                    is
|                                    something
|                                    that (to
|                                    me) is a
|                                    clear work
|                                    item with
|                                    clear
|                                    goals that
|                                    should
|                                    be tackled
|                                    straight
|                                    away.
| 
|                                    --Tom
| 
| 
| 
|                                    On 8/2/12
|                                    3:24 PM,
|                                    "James
|                                    Kempf"
|                                    <james.kempf@ericsson.com>
|                                    wrote:
| 
|                                         So
|                                         after
|                                         seeing
|                                         part
|                                         of
|                                         Alia's
|                                         talk
|                                         this
|                                         morning
|                                         (I
|                                         had
|                                         to
|                                         leave
|                                         in
|                                         the
|                                         middle
|                                         unfortunately),
|                                         I'd
|                                         like
|                                         to
|                                         make
|                                         a
|                                         couple
|                                         suggestions.
|                                         There
|                                         were
|                                         a lot
|                                         of
|                                         ideas
|                                         presented
|                                         in
|                                         the
|                                         talk,
|                                         enough
|                                         for
|                                         an
|                                         entire
|                                         IETF
|                                         Area.
|                                         I
|                                         think
|                                         to
|                                         make
|                                         tangible
|                                         progress,
|                                         the
|                                         work
|                                         needs
|                                         to be
|                                         focussed
|                                         on a
|                                         small
|                                         subset
|                                         that
|                                         would
|                                         be of
|                                         immediate
|                                         interest
|                                         and
|                                         usability.
| 
|                                         There
|                                         are a
|                                         couple
|                                         areas
|                                         that
|                                         suggest
|                                         themselves,
|                                         but
|                                         one
|                                         that
|                                         would
|                                         be
|                                         useful
|                                         in
|                                         work
|                                         that
|                                         I've
|                                         been
|                                         involved
|                                         in is
|                                         a
|                                         standardized
|                                         format
|                                         for
|                                         network
|                                         topology
|                                         representation
|                                         and a
|                                         protocol
|                                         for
|                                         exchanging
|                                         it.
|                                         The
|                                         Onix
|                                         OpenFlow
|                                         controller
|                                         has a
|                                         network
|                                         information
|                                         base
|                                         with
|                                         a
|                                         specialized
|                                         format
|                                         for
|                                         network
|                                         topology,
|                                         and
|                                         every
|                                         OpenFlow
|                                         controller
|                                         requires
|                                         this.
|                                         Having
|                                         a
|                                         standardized
|                                         way
|                                         to
|                                         represent
|                                         it
|                                         might
|                                         foster
|                                         a
|                                         common
|                                         topology
|                                         database
|                                         package.
|                                         Another
|                                         application
|                                         is
|                                         network
|                                         management.
|                                         Every
|                                         network
|                                         management
|                                         system
|                                         needs
|                                         some
|                                         kind
|                                         of
|                                         topology
|                                         representation.
|                                         Finally,
|                                         though
|                                         I am
|                                         not
|                                         an
|                                         expert
|                                         in
|                                         PCE
|                                         construction,
|                                         it
|                                         would
|                                         seem
|                                         to me
|                                         that
|                                         a PCE
|                                         would
|                                         need
|                                         some
|                                         kind
|                                         of
|                                         topology
|                                         representation
|                                         in
|                                         order
|                                         to
|                                         perform
|                                         path
|                                         calculations.
|                                         Having
|                                         a
|                                         way,for
|                                         example,
|                                         for
|                                         the
|                                         OpenFlow
|                                         controller
|                                         and
|                                         the
|                                         PCE
|                                         to
|                                         exchange
|                                         topology
|                                         information
|                                         would
|                                         be
|                                         really
|                                         useful.
|                                         I
|                                         would
|                                         say
|                                         to
|                                         start
|                                         with
|                                         physical
|                                         topology
|                                         because
|                                         that
|                                         is
|                                         fundamental,
|                                         but
|                                         make
|                                         the
|                                         format
|                                         flexible
|                                         enough
|                                         to
|                                         support
|                                         virtual
|                                         topology
|                                         representation.
| 
|                                         jak
|                                         _______________________________________________
|                                         irs-
|                                         discuss
|                                         mailing
|                                         list
|                                         irs-
|                                         discuss@ietf.org
|                                         https:
|                                         //
|                                         www.ietf.org/
|                                         mailman/
|                                         listinfo/
|                                         irs-
|                                         discuss
|                                    _______________________________________________
|                                    irs-
|                                    discuss
|                                    mailing
|                                    list
|                                    irs-
|                                    discuss@ietf.org
|                                    https://
|                                    www.ietf.org/
|                                    mailman/
|                                    listinfo/
|                                    irs-
|                                    discuss
| 
| 
|                               _______________________________________________
|                               irs-discuss mailing
|                               list
|                               irs-
|                               discuss@ietf.org
|                               https://
|                               www.ietf.org/
|                               mailman/listinfo/
|                               irs-discuss
| 
|                          _______________________________________________
|                          irs-discuss mailing list
|                          irs-discuss@ietf.org
|                          https://www.ietf.org/mailman/
|                          listinfo/irs-discuss
| 
|                     _______________________________________________
|                     irs-discuss mailing list
|                     irs-discuss@ietf.org
|                     https://www.ietf.org/mailman/listinfo/
|                     irs-discuss
|           _______________________________________________
|           irs-discuss mailing list
|           irs-discuss@ietf.org
|           https://www.ietf.org/mailman/listinfo/irs-discuss
| 
| 
|      _______________________________________________
|      alto mailing list
|      alto@ietf.org
|      https://www.ietf.org/mailman/listinfo/alto
| 
| 
| --
| ===================================================
| Dr Greg Bernstein, Grotto Networking (510) 573-2237

| _______________________________________________
| Idr mailing list
| Idr@ietf.org
| https://www.ietf.org/mailman/listinfo/idr

