From nea-bounces@ietf.org Thu Sep 06 17:31:43 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1ITOwt-0002Xq-NM; Thu, 06 Sep 2007 17:31:39 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1ITOws-0002XV-VZ
	for nea-confirm+ok@megatron.ietf.org; Thu, 06 Sep 2007 17:31:38 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1ITOws-0002XE-GE
	for nea@ietf.org; Thu, 06 Sep 2007 17:31:38 -0400
Received: from extu-mxob-1.symantec.com ([216.10.194.28])
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1ITOwr-0008H6-Gx
	for nea@ietf.org; Thu, 06 Sep 2007 17:31:38 -0400
Received: from tus1opsmtapin01.ges.symantec.com
	(tus1opsmtapin01.ges.symantec.com [192.168.214.43])
	by extu-mxob-1.symantec.com (8.14.1/8.14.1) with ESMTP id
	l86LVVKQ010650
	(version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO)
	for <nea@ietf.org>; Thu, 6 Sep 2007 14:31:36 -0700
Received: from reserved-155-64-230-19.ges.symantec.com ([155.64.230.19]
	helo=TUS1XCHECNPIN02.enterprise.veritas.com)
	by tus1opsmtapin01.ges.symantec.com with esmtp (Exim 4.67)
	(envelope-from <Paul_Sangster@symantec.com>) id 1ITOwb-0001D9-OI
	for nea@ietf.org; Thu, 06 Sep 2007 14:31:21 -0700
Received: from TUS1XCHEVSPIN05.enterprise.veritas.com ([155.64.231.28]) by
	TUS1XCHECNPIN02.enterprise.veritas.com with Microsoft
	SMTPSVC(6.0.3790.1830); Thu, 6 Sep 2007 14:31:21 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Date: Thu, 6 Sep 2007 14:31:04 -0700
Message-ID: <AB96CED633A7C246BDC661DBEE1CF01FC4607D@TUS1XCHCLUPIN12.enterprise.veritas.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: Should we downgrade PA security to a SHOULD, MAY or remove it?
Thread-Index: AcfwzTq2XBTSVc2lRlGEdm6kXscoXw==
From: "Paul Sangster" <Paul_Sangster@symantec.com>
To: <nea@ietf.org>
X-OriginalArrivalTime: 06 Sep 2007 21:31:21.0201 (UTC)
	FILETIME=[44870610:01C7F0CD]
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 3d7f2f6612d734db849efa86ea692407
Subject: [Nea] 
	Should we downgrade PA security to a SHOULD, MAY or remove it?
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Content-Type: multipart/mixed; boundary="===============1524917633=="
Errors-To: nea-bounces@ietf.org

This is a multi-part message in MIME format.

--===============1524917633==
Content-class: urn:content-classes:message
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C7F0CD.44A3A998"

This is a multi-part message in MIME format.

------_=_NextPart_001_01C7F0CD.44A3A998
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

Hopefully everyone noticed that our -04 revision was published and is=20
awaiting comments from the WG.  As mentioned in my prior e-mail on
8/24/07,
we've made a significant number of changes based on comments received=20
during WGLC.  We tried to address those comments but one significant
question that was not addressed relates to our requirement PA-5 below:
=20
 "PA-5 The PA protocol MUST provide authentication, integrity, and=20
       confidentiality of attributes communicated between a Posture=20
       Collector and Posture Validator. This enables end-to-end security

       across a NEA deployment that might involve traversal of several=20
       systems."
=20
The question is should this requirement be a SHOULD, MAY, or removed=20
entirely?  This requirement guides our selection process for the PA
protocol and currently would prevent candidate PA protocols from being
selected if they lack proper security protections.  In the past
we've discussed the overlapping requirements for security upon PA
(MUST),
PB (MAY) and PT (MUST) and in Prague this factored into our decision to
reduce PB to a MAY (instead of a MUST).  This topic is also discussed in

section 8.2 and other parts of the spec.
=20
The current PA security protections were included to provide for end to
end=20
(e.g. Posture Collector to Posture Validator) authentication, integrity
and
confidentiality protection of the PA attributes.  These attributes might
travel to remote=20
systems hosting the posture validator that are separate from the posture

transport server thus PT protections are no longer present.  Several
folks=20
have mentioned that the protection of the link between the posture
transport=20
server and remote posture validator could be provided in other less
complex=20
ways and so requiring a standards based PA security protocol isn't
critical=20
and will just make our deliverable more difficult.  We also would need
to
specify a solution consistent with the principals of RFC 4107 which
could
be challenging.
=20
We would like to request that the working group discuss whether we
should=20
downgrade PA-5 to SHOULD, MAY or remove it all together.  In order
to bound the discussion we would like to conclude the discussion by 10PM
PDT
on thursday, Sept 20th (2 weeks).  This will allows the editors time to
make=20
any revisions to the specification and re-submit in a timely manner so
we can
product a new I-D ready for a re-start the WGLC process.
=20
Paul

------_=_NextPart_001_01C7F0CD.44A3A998
Content-Type: text/html;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=3DContent-Type content=3D"text/html; =
charset=3Dus-ascii">
<META content=3D"MSHTML 6.00.2900.3157" name=3DGENERATOR></HEAD>
<BODY>
<DIV><FONT face=3DArial size=3D2>Hopefully everyone noticed that our -04 =
revision=20
was published and is <BR>awaiting comments from the WG.&nbsp; As =
mentioned in my=20
prior e-mail on 8/24/07,<BR>we've made a significant number of changes =
based on=20
comments received <BR>during WGLC.&nbsp; We tried to address those =
comments but=20
one significant<BR>question that was not addressed relates to our =
requirement=20
PA-5 below:</FONT></DIV>
<DIV>&nbsp;</DIV>
<DIV><FONT face=3DArial size=3D2>&nbsp;"PA-5 The PA protocol MUST =
provide=20
authentication, integrity, and <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;=20
confidentiality of attributes communicated between a Posture=20
<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Collector and Posture =
Validator. This=20
enables end-to-end security <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
across a=20
NEA deployment that might involve traversal of several=20
<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; systems."</FONT></DIV>
<DIV>&nbsp;</DIV>
<DIV><FONT face=3DArial size=3D2>The question is should this requirement =
be a=20
SHOULD, MAY, or removed <BR>entirely?&nbsp; This requirement guides our=20
selection process for the PA<BR>protocol and currently would prevent =
candidate=20
PA protocols from being<BR>selected if they lack proper security=20
protections.&nbsp; In the past<BR>we've discussed the overlapping =
requirements=20
for security upon PA (MUST),<BR>PB (MAY) and PT (MUST) and in Prague =
this=20
factored into our decision to<BR>reduce PB to a MAY (instead of a =
MUST).&nbsp;=20
This topic is also discussed in <BR>section 8.2 and other parts of the=20
spec.</FONT></DIV>
<DIV>&nbsp;</DIV>
<DIV><FONT face=3DArial size=3D2>The current PA security protections =
were included=20
to provide for end to end <BR>(e.g. Posture Collector to Posture =
Validator)=20
authentication, integrity and</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>confidentiality protection of the PA=20
attributes<SPAN class=3D734362521-06092007>.&nbsp; These=20
attributes&nbsp;m</SPAN>ight travel to remote <BR>systems hosting the =
posture=20
validator that are separate from the posture <BR>transport server thus =
PT=20
protections are no longer present.&nbsp; Several folks <BR>have =
mentioned that=20
the protection of the link between the posture transport <BR>server and =
remote=20
posture validator could be provided in other less complex <BR>ways and =
so=20
requiring a standards based PA security protocol isn't critical <BR>and =
will=20
just make our deliverable more difficult.&nbsp; We also would need =
to<BR>specify=20
a solution consistent with the principals of RFC 4107 which could<BR>be=20
challenging.</FONT></DIV>
<DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
<DIV><FONT face=3DArial size=3D2>We would like to request that the =
working group=20
discuss whether we should <BR>downgrade PA-5 to SHOULD, MAY or remove it =
all=20
together.&nbsp; In order<BR>to bound the discussion we would like to =
conclude=20
the discussion by 10PM PDT<BR>on thursday, Sept 20th (2 weeks).&nbsp; =
This will=20
allows the editors time to make <BR>any revisions to the specification =
and=20
re-submit in a timely manner so we can<BR><SPAN =
class=3D734362521-06092007>product=20
a new I-D ready for a </SPAN>re-start the WGLC process.</FONT></DIV>
<DIV>&nbsp;</DIV>
<DIV><FONT face=3DArial size=3D2>Paul</FONT></DIV></BODY></HTML>

------_=_NextPart_001_01C7F0CD.44A3A998--



--===============1524917633==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea

--===============1524917633==--





From nea-bounces@ietf.org Thu Sep 06 18:22:41 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1ITPkH-0000Um-Ho; Thu, 06 Sep 2007 18:22:41 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1ITPkG-0000Ug-Fc
	for nea-confirm+ok@megatron.ietf.org; Thu, 06 Sep 2007 18:22:40 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1ITPkG-0000UY-5c
	for nea@ietf.org; Thu, 06 Sep 2007 18:22:40 -0400
Received: from elasmtp-masked.atl.sa.earthlink.net ([209.86.89.68])
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1ITPkE-0005Z8-L0
	for nea@ietf.org; Thu, 06 Sep 2007 18:22:40 -0400
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=earthlink.net;
	b=hUNx4r96lAl8xncwmDQC0gFoN5PyKh+hZxK3j6gKZVPth9ZL3XiPdcsCbmdK0BOk;
	h=Received:Message-ID:From:To:References:Subject:Date:MIME-Version:Content-Type:X-Priority:X-MSMail-Priority:X-Mailer:X-MimeOLE:X-ELNK-Trace:X-Originating-IP;
Received: from [64.125.79.23] (helo=gw)
	by elasmtp-masked.atl.sa.earthlink.net with asmtp (Exim 4.34)
	id 1ITPkD-0000LQ-Tp; Thu, 06 Sep 2007 18:22:38 -0400
Message-ID: <00f701c7f0d4$6fa89280$174f7d40@home.glassey.com>
From: "todd glassey" <tglassey@earthlink.net>
To: "Paul Sangster" <Paul_Sangster@symantec.com>,
	<nea@ietf.org>
References: <AB96CED633A7C246BDC661DBEE1CF01FC4607D@TUS1XCHCLUPIN12.enterprise.veritas.com>
Subject: Re: [Nea] Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Date: Thu, 6 Sep 2007 15:22:36 -0700
MIME-Version: 1.0
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2900.3138
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.3138
X-ELNK-Trace: 01b7a7e171bdf5911aa676d7e74259b7b3291a7d08dfec79e8b53b7f842ed9985e9f74924b0734a3350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c
X-Originating-IP: 64.125.79.23
X-Spam-Score: 0.0 (/)
X-Scan-Signature: b6e18fadcfab41fa5e7faede753de4c2
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Content-Type: multipart/mixed; boundary="===============0032556310=="
Errors-To: nea-bounces@ietf.org

This is a multi-part message in MIME format.

--===============0032556310==
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_00F4_01C7F099.C0EE8C70"

This is a multi-part message in MIME format.

------=_NextPart_000_00F4_01C7F099.C0EE8C70
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable


  ----- Original Message -----=20
  From: Paul Sangster=20
  To: nea@ietf.org=20
  Sent: Thursday, September 06, 2007 2:31 PM
  Subject: [Nea] Should we downgrade PA security to a SHOULD, MAY or =
remove it?


  Hopefully everyone noticed that our -04 revision was published and is=20
  awaiting comments from the WG.  As mentioned in my prior e-mail on =
8/24/07,
  we've made a significant number of changes based on comments received=20
  during WGLC.  We tried to address those comments but one significant
  question that was not addressed relates to our requirement PA-5 below:

   "PA-5 The PA protocol MUST provide authentication, integrity, and=20
         confidentiality of attributes communicated between a Posture=20
         Collector and Posture Validator. This enables end-to-end =
security=20
         across a NEA deployment that might involve traversal of several =

         systems."

  The question is should this requirement be a SHOULD, MAY, or removed=20
  entirely?  This requirement guides our selection process for the PA
  protocol and currently would prevent candidate PA protocols from being
  selected if they lack proper security protections.  In the past
  we've discussed the overlapping requirements for security upon PA =
(MUST),
  PB (MAY) and PT (MUST) and in Prague this factored into our decision =
to
  reduce PB to a MAY (instead of a MUST).  This topic is also discussed =
in=20
  section 8.2 and other parts of the spec.

Paul - the issue is the use model. If NEA is used across an Enterprise =
Architecture or multiple campus type topologies then the policy controls =
"MAY" want to require this end-to-end functionality.=20

Likewise, if the NEA tools were used for something like Microsoft =
Genuine Advantage compliance, then it would definently be a MUST one =
would think.

The real issue is that the OPERATIONS POLICIES (OP)  will determine =
whether certain functions or criteria are put in place, so depending on =
the OP's the state of this control will changed so you need to be able =
to say MUST in some cases and MAY or OPTIONALLY in others.


  The current PA security protections were included to provide for end =
to end=20
  (e.g. Posture Collector to Posture Validator) authentication, =
integrity and
  confidentiality protection of the PA attributes.  These attributes =
might travel to remote=20
  systems hosting the posture validator that are separate from the =
posture=20
  transport server thus PT protections are no longer present. =20

OK so this could be a situation perhaps where a Trusted Third Party was =
operating the NEA master PV so it needs some tunnel to communicate with =
the other components of the system. I think this is 'doable' with the =
simple addition of a requirement "that all NEA components communicate in =
a secure manner" which meets Industry Security Requirements for those =
application's bases. This type of disclaimer means that the NEA Spec =
itself doesnt need to address these inter-nodal security issues IMHO.

  Several folks=20
  have mentioned that the protection of the link between the posture =
transport=20
  server and remote posture validator could be provided in other less =
complex=20
  ways and so requiring a standards based PA security protocol isn't =
critical=20
  and will just make our deliverable more difficult. =20

The problem here isnt NEA - its the use environments which NEA is to be =
used in and many of them already have their own security and audit =
requirements.=20


  We also would need to
  specify a solution consistent with the principals of RFC 4107 which =
could
  be challenging.

Understood - EAP is complex...

  We would like to request that the working group discuss whether we =
should=20
  downgrade PA-5 to SHOULD, MAY or remove it all together. =20

I think PA-5's status is dependant on the end use model and based =
therein, the existing language of its description needs to be changed to =
reflect that.


  In order
  to bound the discussion we would like to conclude the discussion by =
10PM PDT
  on thursday, Sept 20th (2 weeks).  This will allows the editors time =
to make=20
  any revisions to the specification and re-submit in a timely manner so =
we can
  product a new I-D ready for a re-start the WGLC process.

Existing:

 "PA-5 The PA protocol MUST provide authentication, integrity, and=20
       confidentiality of attributes communicated between a Posture=20
       Collector and Posture Validator. This enables end-to-end security =

       across a NEA deployment that might involve traversal of several=20
       systems."

Proposed:

 "PA-5 The PA protocol can be used in conformance with other
      tools and BCPs which will set at a policy level the requirements=20
      for which components of the NEA specification are mandatory,=20
      and which are optional. In all instances NEA systems rely on=20
      the integrity of the underlying Network System to operate on
     and so in these instances the PA MUST provide authentication, and=20
     confidentiality of attributes communicated between a Posture=20
     NEA Components (Posture Collector's and Posture Validator).=20
     This Integrity and confidentiality of communications may be =
integrated
     into NEA implementations or those implementations may rely on other
     protocols and technologies to supply those operation's features for =
the
     NEA suite.

    This flexibility in NEA design and deployment enables=20
    end-to-end security across a NEA deployment which may=20
    involve traversal of several heterogenous networks and their =
systems."


/TSG

  Paul


-------------------------------------------------------------------------=
-----


  _______________________________________________
  Nea mailing list
  Nea@ietf.org
  https://www1.ietf.org/mailman/listinfo/nea

------=_NextPart_000_00F4_01C7F099.C0EE8C70
Content-Type: text/html;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=3DContent-Type content=3D"text/html; =
charset=3Diso-8859-1">
<META content=3D"MSHTML 6.00.6000.16525" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3D"Times New Roman" size=3D2></FONT>&nbsp;</DIV>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <DIV style=3D"FONT: 10pt arial">----- Original Message ----- </DIV>
  <DIV=20
  style=3D"BACKGROUND: #e4e4e4; FONT: 10pt arial; font-color: =
black"><B>From:</B>=20
  <A title=3DPaul_Sangster@symantec.com=20
  href=3D"mailto:Paul_Sangster@symantec.com">Paul Sangster</A> </DIV>
  <DIV style=3D"FONT: 10pt arial"><B>To:</B> <A title=3Dnea@ietf.org=20
  href=3D"mailto:nea@ietf.org">nea@ietf.org</A> </DIV>
  <DIV style=3D"FONT: 10pt arial"><B>Sent:</B> Thursday, September 06, =
2007 2:31=20
  PM</DIV>
  <DIV style=3D"FONT: 10pt arial"><B>Subject:</B> [Nea] Should we =
downgrade PA=20
  security to a SHOULD, MAY or remove it?</DIV>
  <DIV><BR></DIV>
  <DIV><FONT face=3DArial size=3D2>Hopefully everyone noticed that our =
-04 revision=20
  was published and is <BR>awaiting comments from the WG.&nbsp; As =
mentioned in=20
  my prior e-mail on 8/24/07,<BR>we've made a significant number of =
changes=20
  based on comments received <BR>during WGLC.&nbsp; We tried to address =
those=20
  comments but one significant<BR>question that was not addressed =
relates to our=20
  requirement PA-5 below:</FONT></DIV>
  <DIV>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>&nbsp;"PA-5 The PA protocol MUST =
provide=20
  authentication, integrity, and =
<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;=20
  confidentiality of attributes communicated between a Posture=20
  <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Collector and Posture =
Validator. This=20
  enables end-to-end security <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
across a=20
  NEA deployment that might involve traversal of several=20
  <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; systems."</FONT></DIV>
  <DIV>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>The question is should this =
requirement be a=20
  SHOULD, MAY, or removed <BR>entirely?&nbsp; This requirement guides =
our=20
  selection process for the PA<BR>protocol and currently would prevent =
candidate=20
  PA protocols from being<BR>selected if they lack proper security=20
  protections.&nbsp; In the past<BR>we've discussed the overlapping =
requirements=20
  for security upon PA (MUST),<BR>PB (MAY) and PT (MUST) and in Prague =
this=20
  factored into our decision to<BR>reduce PB to a MAY (instead of a =
MUST).&nbsp;=20
  This topic is also discussed in <BR>section 8.2 and other parts of the =

  spec.</FONT></DIV>
  <DIV><FONT size=3D2></FONT>&nbsp;</DIV></BLOCKQUOTE>
<DIV><FONT size=3D2>Paul - the issue is the use model. If NEA is used =
across an=20
Enterprise Architecture or multiple campus type topologies then the =
policy=20
controls "MAY" want to require this end-to-end functionality. =
</FONT></DIV>
<DIV><FONT size=3D2></FONT>&nbsp;</DIV>
<DIV><FONT size=3D2>Likewise, if the NEA tools were used for something =
like=20
Microsoft Genuine Advantage compliance, then it would definently be a =
MUST one=20
would think.</FONT></DIV>
<DIV><FONT size=3D2></FONT>&nbsp;</DIV>
<DIV><FONT size=3D2>The real issue is that the OPERATIONS POLICIES (OP) =
&nbsp;will=20
determine whether certain functions or criteria are put in place, so =
depending=20
on the OP's the state of this control will changed so you need to be =
able to say=20
MUST in some cases and MAY or OPTIONALLY in others.</FONT></DIV>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <DIV><FONT size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT face=3D"Times New Roman" size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>The current PA security protections =
were included=20
  to provide for end to end <BR>(e.g. Posture Collector to Posture =
Validator)=20
  authentication, integrity and</FONT></DIV>
  <DIV><FONT face=3DArial size=3D2>confidentiality protection of the PA=20
  attributes<SPAN class=3D734362521-06092007>.&nbsp; These=20
  attributes&nbsp;m</SPAN>ight travel to remote <BR>systems hosting the =
posture=20
  validator that are separate from the posture <BR>transport server thus =
PT=20
  protections are no longer present.&nbsp; </FONT></DIV>
  <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV></BLOCKQUOTE>
<DIV><FONT face=3DArial size=3D2>OK so this could be a situation perhaps =
where a=20
Trusted Third Party was operating the NEA master PV so it needs some =
tunnel to=20
communicate with the other components of the system. I think this is =
'doable'=20
with the simple addition of a requirement "that all NEA components =
communicate=20
in a secure manner" which meets Industry Security Requirements for those =

application's bases. This type of disclaimer means that the NEA Spec =
itself=20
doesnt need to address these inter-nodal security issues =
IMHO.</FONT></DIV>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>Several folks <BR>have mentioned that =
the=20
  protection of the link between the posture transport <BR>server and =
remote=20
  posture validator could be provided in other less complex <BR>ways and =
so=20
  requiring a standards based PA security protocol isn't critical =
<BR>and will=20
  just make our deliverable more difficult.&nbsp; </FONT></DIV>
  <DIV><FONT size=3D2></FONT>&nbsp;</DIV></BLOCKQUOTE>
<DIV><FONT size=3D2>The problem here isnt NEA - its the use environments =
which NEA=20
is to be used in and many of them already have their own security and =
audit=20
requirements.&nbsp;</FONT></DIV>
<DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>We also would need to<BR>specify a =
solution=20
  consistent with the principals of RFC 4107 which could<BR>be=20
  challenging.</FONT></DIV>
  <DIV><FONT size=3D2></FONT>&nbsp;</DIV></BLOCKQUOTE>
<DIV><FONT size=3D2>Understood - EAP is complex...</FONT></DIV>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <DIV><FONT size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>We would like to request that the =
working group=20
  discuss whether we should <BR>downgrade PA-5 to SHOULD, MAY or remove =
it all=20
  together.&nbsp; </FONT></DIV>
  <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV></BLOCKQUOTE>
<DIV><FONT face=3DArial size=3D2>I think PA-5's status is dependant on =
the end use=20
model and based therein, the existing language of its description needs =
to be=20
changed to reflect that.</FONT></DIV>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>In order<BR>to bound the discussion =
we would like=20
  to conclude the discussion by 10PM PDT<BR>on thursday, Sept 20th (2=20
  weeks).&nbsp; This will allows the editors time to make <BR>any =
revisions to=20
  the specification and re-submit in a timely manner so we can<BR><SPAN=20
  class=3D734362521-06092007>product a new I-D ready for a =
</SPAN>re-start the=20
  WGLC process.</FONT></DIV>
  <DIV><FONT size=3D2></FONT>&nbsp;</DIV></BLOCKQUOTE>
<DIV>
<DIV><FONT face=3D"Times New Roman" size=3D2>Existing:</FONT></DIV>
<DIV><FONT face=3D"Times New Roman" size=3D2></FONT>&nbsp;</DIV>
<DIV><FONT face=3DArial size=3D2>&nbsp;"PA-5 The PA protocol MUST =
provide=20
authentication, integrity, and <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;=20
confidentiality of attributes communicated between a Posture=20
<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Collector and Posture =
Validator. This=20
enables end-to-end security <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
across a=20
NEA deployment that might involve traversal of several=20
<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; systems."</FONT></DIV>
<DIV><FONT size=3D2></FONT>&nbsp;</DIV>
<DIV><FONT size=3D2>Proposed:</FONT></DIV>
<DIV>
<DIV>&nbsp;</DIV>
<DIV><FONT face=3DArial size=3D2>&nbsp;"PA-5 The PA protocol can be used =
in=20
conformance with other</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; tools =
and BCPs which=20
will set at a policy level the requirements </FONT></DIV>
<DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; for =
which components=20
of the NEA specification are mandatory, </FONT></DIV>
<DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; and =
which are=20
optional. In all instances NEA systems rely on&nbsp;</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;the =
integrity=20
of the underlying Network System to operate on</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp;&nbsp; and so in =
these instances=20
the PA MUST provide=20
authentication,&nbsp;and&nbsp;<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;confident=
iality=20
of attributes communicated between a=20
Posture&nbsp;<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;NEA Components=20
(Posture&nbsp;Collector's and Posture Validator). </FONT></DIV>
<DIV><FONT face=3D"Times New Roman" =
size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;This=20
Integrity and confidentiality of communications may be =
integrated</FONT></DIV>
<DIV><FONT face=3D"Times New Roman" =
size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;into NEA=20
implementations or those implementations may rely on other</FONT></DIV>
<DIV><FONT face=3D"Times New Roman" =
size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;protocols=20
and technologies to supply those operation's features for =
the</FONT></DIV>
<DIV><FONT face=3D"Times New Roman" size=3D2>&nbsp;&nbsp;&nbsp;&nbsp; =
NEA=20
suite.</FONT></DIV>
<DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
<DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp; This flexibility in =
NEA design=20
and deployment enables </FONT></DIV>
<DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp; end-to-end =
security&nbsp;across=20
a NEA</FONT><FONT face=3DArial size=3D2>&nbsp;deployment&nbsp;which may=20
</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp; involve traversal of =

several&nbsp;heterogenous</FONT><FONT face=3DArial size=3D2> networks =
and their=20
systems."</FONT></DIV>
<DIV><FONT size=3D2></FONT>&nbsp;</DIV>
<DIV><FONT size=3D2></FONT>&nbsp;</DIV>
<DIV><FONT size=3D2>/TSG</FONT></DIV></DIV></DIV>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <DIV><FONT face=3D"Times New Roman" size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>Paul</FONT></DIV>
  <P>
  <HR>

  <P></P>_______________________________________________<BR>Nea mailing=20
  =
list<BR>Nea@ietf.org<BR>https://www1.ietf.org/mailman/listinfo/nea<BR></B=
LOCKQUOTE></BODY></HTML>

------=_NextPart_000_00F4_01C7F099.C0EE8C70--




--===============0032556310==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea

--===============0032556310==--






From nea-bounces@ietf.org Thu Sep 06 18:33:29 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1ITPui-0001Y4-Eu; Thu, 06 Sep 2007 18:33:28 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1ITPug-0001XI-On
	for nea-confirm+ok@megatron.ietf.org; Thu, 06 Sep 2007 18:33:26 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1ITPug-0001X7-BB
	for nea@ietf.org; Thu, 06 Sep 2007 18:33:26 -0400
Received: from extu-mxob-2.symantec.com ([216.10.194.135])
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1ITPuc-0001r7-JN
	for nea@ietf.org; Thu, 06 Sep 2007 18:33:26 -0400
Received: from tus1opsmtapin02.ges.symantec.com
	(tus1opsmtapin02.ges.symantec.com [192.168.214.44])
	by extu-mxob-2.symantec.com (8.14.1/8.14.1) with ESMTP id
	l86MXKK0026915
	(version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO);
	Thu, 6 Sep 2007 15:33:20 -0700
Received: from reserved-155-64-230-19.ges.symantec.com ([155.64.230.19]
	helo=TUS1XCHECNPIN02.enterprise.veritas.com)
	by tus1opsmtapin02.ges.symantec.com with esmtp (Exim 4.67)
	(envelope-from <Paul_Sangster@symantec.com>)
	id 1ITPua-0000pV-G6; Thu, 06 Sep 2007 15:33:20 -0700
Received: from TUS1XCHEVSPIN05.enterprise.veritas.com ([155.64.231.28]) by
	TUS1XCHECNPIN02.enterprise.veritas.com with Microsoft
	SMTPSVC(6.0.3790.1830); Thu, 6 Sep 2007 15:33:19 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Subject: RE: [Nea] Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Date: Thu, 6 Sep 2007 15:33:01 -0700
Message-ID: <AB96CED633A7C246BDC661DBEE1CF01FC460F8@TUS1XCHCLUPIN12.enterprise.veritas.com>
In-Reply-To: <00f701c7f0d4$6fa89280$174f7d40@home.glassey.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Thread-Index: Acfw1HuXZ2YvxV0VRUuWmthaujzY2QAANd5g
References: <AB96CED633A7C246BDC661DBEE1CF01FC4607D@TUS1XCHCLUPIN12.enterprise.veritas.com>
	<00f701c7f0d4$6fa89280$174f7d40@home.glassey.com>
From: "Paul Sangster" <Paul_Sangster@symantec.com>
To: "todd glassey" <tglassey@earthlink.net>, <nea@ietf.org>
X-OriginalArrivalTime: 06 Sep 2007 22:33:19.0880 (UTC)
	FILETIME=[ED086880:01C7F0D5]
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 4b21264b25b2584da3ddf2bd579ca48f
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Content-Type: multipart/mixed; boundary="===============0179671529=="
Errors-To: nea-bounces@ietf.org

This is a multi-part message in MIME format.

--===============0179671529==
Content-class: urn:content-classes:message
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C7F0D5.ECF1AC00"

This is a multi-part message in MIME format.

------_=_NextPart_001_01C7F0D5.ECF1AC00
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

Thanks you've made some good points.  I just want to re-iterate that the
goal of these protocol requirements is to state what the WG criteria
will be when selecting the PA, PB and PT protocols.  Thus if we leave
PA-5 as it currently stands, the WG must only select a PA candidate
protocol capable of providing authentication, integrity and
confidentiality protection.  Whether a deployment chooses to use those
protections is up to the deployer (assuming these features are supported
by the selected protocol).  So, if we change MUST to SHOULD then this
means the WG can consider and potentially select a PA protocol which
lacks some or all of the described security protections.


________________________________

	From: todd glassey [mailto:tglassey@earthlink.net]=20
	Sent: Thursday, September 06, 2007 3:23 PM
	To: Paul Sangster; nea@ietf.org
	Subject: Re: [Nea] Should we downgrade PA security to a
SHOULD,MAY or remove it?
=09
=09
	=20

		----- Original Message -----=20
		From: Paul Sangster <mailto:Paul_Sangster@symantec.com>

		To: nea@ietf.org=20
		Sent: Thursday, September 06, 2007 2:31 PM
		Subject: [Nea] Should we downgrade PA security to a
SHOULD, MAY or remove it?

		Hopefully everyone noticed that our -04 revision was
published and is=20
		awaiting comments from the WG.  As mentioned in my prior
e-mail on 8/24/07,
		we've made a significant number of changes based on
comments received=20
		during WGLC.  We tried to address those comments but one
significant
		question that was not addressed relates to our
requirement PA-5 below:
		=20
		 "PA-5 The PA protocol MUST provide authentication,
integrity, and=20
		       confidentiality of attributes communicated
between a Posture=20
		       Collector and Posture Validator. This enables
end-to-end security=20
		       across a NEA deployment that might involve
traversal of several=20
		       systems."
		=20
		The question is should this requirement be a SHOULD,
MAY, or removed=20
		entirely?  This requirement guides our selection process
for the PA
		protocol and currently would prevent candidate PA
protocols from being
		selected if they lack proper security protections.  In
the past
		we've discussed the overlapping requirements for
security upon PA (MUST),
		PB (MAY) and PT (MUST) and in Prague this factored into
our decision to
		reduce PB to a MAY (instead of a MUST).  This topic is
also discussed in=20
		section 8.2 and other parts of the spec.
		=20

	Paul - the issue is the use model. If NEA is used across an
Enterprise Architecture or multiple campus type topologies then the
policy controls "MAY" want to require this end-to-end functionality.=20
	=20
	Likewise, if the NEA tools were used for something like
Microsoft Genuine Advantage compliance, then it would definently be a
MUST one would think.
	=20
	The real issue is that the OPERATIONS POLICIES (OP)  will
determine whether certain functions or criteria are put in place, so
depending on the OP's the state of this control will changed so you need
to be able to say MUST in some cases and MAY or OPTIONALLY in others.

		=20
		=20
		The current PA security protections were included to
provide for end to end=20
		(e.g. Posture Collector to Posture Validator)
authentication, integrity and
		confidentiality protection of the PA attributes.  These
attributes might travel to remote=20
		systems hosting the posture validator that are separate
from the posture=20
		transport server thus PT protections are no longer
present. =20
		=20

	OK so this could be a situation perhaps where a Trusted Third
Party was operating the NEA master PV so it needs some tunnel to
communicate with the other components of the system. I think this is
'doable' with the simple addition of a requirement "that all NEA
components communicate in a secure manner" which meets Industry Security
Requirements for those application's bases. This type of disclaimer
means that the NEA Spec itself doesnt need to address these inter-nodal
security issues IMHO.

		=20
		Several folks=20
		have mentioned that the protection of the link between
the posture transport=20
		server and remote posture validator could be provided in
other less complex=20
		ways and so requiring a standards based PA security
protocol isn't critical=20
		and will just make our deliverable more difficult. =20
		=20

	The problem here isnt NEA - its the use environments which NEA
is to be used in and many of them already have their own security and
audit requirements.=20
	=20

		=20
		We also would need to
		specify a solution consistent with the principals of RFC
4107 which could
		be challenging.
		=20

	Understood - EAP is complex...

		=20
		We would like to request that the working group discuss
whether we should=20
		downgrade PA-5 to SHOULD, MAY or remove it all together.

		=20

	I think PA-5's status is dependant on the end use model and
based therein, the existing language of its description needs to be
changed to reflect that.

		=20
		=20
		In order
		to bound the discussion we would like to conclude the
discussion by 10PM PDT
		on thursday, Sept 20th (2 weeks).  This will allows the
editors time to make=20
		any revisions to the specification and re-submit in a
timely manner so we can
		product a new I-D ready for a re-start the WGLC process.
		=20

	Existing:
	=20
	 "PA-5 The PA protocol MUST provide authentication, integrity,
and=20
	       confidentiality of attributes communicated between a
Posture=20
	       Collector and Posture Validator. This enables end-to-end
security=20
	       across a NEA deployment that might involve traversal of
several=20
	       systems."
	=20
	Proposed:
	=20
	 "PA-5 The PA protocol can be used in conformance with other
	      tools and BCPs which will set at a policy level the
requirements=20
	      for which components of the NEA specification are
mandatory,=20
	      and which are optional. In all instances NEA systems rely
on=20
	      the integrity of the underlying Network System to operate
on
	     and so in these instances the PA MUST provide
authentication, and=20
	     confidentiality of attributes communicated between a
Posture=20
	     NEA Components (Posture Collector's and Posture Validator).

	     This Integrity and confidentiality of communications may be
integrated
	     into NEA implementations or those implementations may rely
on other
	     protocols and technologies to supply those operation's
features for the
	     NEA suite.
	=20
	    This flexibility in NEA design and deployment enables=20
	    end-to-end security across a NEA deployment which may=20
	    involve traversal of several heterogenous networks and their
systems."
	=20
	=20
	/TSG

		=20
		Paul

	=09
________________________________


	=09

		_______________________________________________
		Nea mailing list
		Nea@ietf.org
		https://www1.ietf.org/mailman/listinfo/nea
	=09


------_=_NextPart_001_01C7F0D5.ECF1AC00
Content-Type: text/html;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=3DContent-Type content=3D"text/html; =
charset=3Dus-ascii">
<META content=3D"MSHTML 6.00.2900.3157" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV dir=3Dltr align=3Dleft><SPAN class=3D593012922-06092007><FONT =
face=3DArial=20
color=3D#0000ff size=3D2>Thanks you've made some good points.&nbsp; I =
just want to=20
re-iterate that the goal of these protocol requirements is to state what =
the WG=20
criteria will be when selecting the PA, PB and PT protocols.&nbsp; Thus =
if we=20
leave PA-5 as it currently stands, the WG must only select a PA =
candidate=20
protocol capable of providing authentication, integrity and =
confidentiality=20
protection.&nbsp; Whether a deployment chooses to use those protections =
is up to=20
the deployer (assuming these features are supported by the selected=20
protocol).&nbsp; So, if we change MUST to SHOULD then this means the WG =
can=20
consider and potentially select a PA protocol which lacks some or all of =
the=20
described security protections.</FONT></SPAN></DIV><BR>
<BLOCKQUOTE dir=3Dltr=20
style=3D"PADDING-LEFT: 5px; MARGIN-LEFT: 5px; BORDER-LEFT: #0000ff 2px =
solid; MARGIN-RIGHT: 0px">
  <DIV class=3DOutlookMessageHeader lang=3Den-us dir=3Dltr align=3Dleft>
  <HR tabIndex=3D-1>
  <FONT face=3DTahoma size=3D2><B>From:</B> todd glassey=20
  [mailto:tglassey@earthlink.net] <BR><B>Sent:</B> Thursday, September =
06, 2007=20
  3:23 PM<BR><B>To:</B> Paul Sangster; nea@ietf.org<BR><B>Subject:</B> =
Re: [Nea]=20
  Should we downgrade PA security to a SHOULD,MAY or remove=20
  it?<BR></FONT><BR></DIV>
  <DIV></DIV>
  <DIV><FONT face=3D"Times New Roman" size=3D2></FONT>&nbsp;</DIV>
  <BLOCKQUOTE=20
  style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
    <DIV style=3D"FONT: 10pt arial">----- Original Message ----- </DIV>
    <DIV=20
    style=3D"BACKGROUND: #e4e4e4; FONT: 10pt arial; font-color: =
black"><B>From:</B>=20
    <A title=3DPaul_Sangster@symantec.com=20
    href=3D"mailto:Paul_Sangster@symantec.com">Paul Sangster</A> </DIV>
    <DIV style=3D"FONT: 10pt arial"><B>To:</B> <A title=3Dnea@ietf.org=20
    href=3D"mailto:nea@ietf.org">nea@ietf.org</A> </DIV>
    <DIV style=3D"FONT: 10pt arial"><B>Sent:</B> Thursday, September 06, =
2007 2:31=20
    PM</DIV>
    <DIV style=3D"FONT: 10pt arial"><B>Subject:</B> [Nea] Should we =
downgrade PA=20
    security to a SHOULD, MAY or remove it?</DIV>
    <DIV><BR></DIV>
    <DIV><FONT face=3DArial size=3D2>Hopefully everyone noticed that our =
-04=20
    revision was published and is <BR>awaiting comments from the =
WG.&nbsp; As=20
    mentioned in my prior e-mail on 8/24/07,<BR>we've made a significant =
number=20
    of changes based on comments received <BR>during WGLC.&nbsp; We =
tried to=20
    address those comments but one significant<BR>question that was not=20
    addressed relates to our requirement PA-5 below:</FONT></DIV>
    <DIV>&nbsp;</DIV>
    <DIV><FONT face=3DArial size=3D2>&nbsp;"PA-5 The PA protocol MUST =
provide=20
    authentication, integrity, and =
<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;=20
    confidentiality of attributes communicated between a Posture=20
    <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Collector and Posture =
Validator.=20
    This enables end-to-end security =
<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;=20
    across a NEA deployment that might involve traversal of several=20
    <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; systems."</FONT></DIV>
    <DIV>&nbsp;</DIV>
    <DIV><FONT face=3DArial size=3D2>The question is should this =
requirement be a=20
    SHOULD, MAY, or removed <BR>entirely?&nbsp; This requirement guides =
our=20
    selection process for the PA<BR>protocol and currently would prevent =

    candidate PA protocols from being<BR>selected if they lack proper =
security=20
    protections.&nbsp; In the past<BR>we've discussed the overlapping=20
    requirements for security upon PA (MUST),<BR>PB (MAY) and PT (MUST) =
and in=20
    Prague this factored into our decision to<BR>reduce PB to a MAY =
(instead of=20
    a MUST).&nbsp; This topic is also discussed in <BR>section 8.2 and =
other=20
    parts of the spec.</FONT></DIV>
    <DIV><FONT size=3D2></FONT>&nbsp;</DIV></BLOCKQUOTE>
  <DIV><FONT size=3D2>Paul - the issue is the use model. If NEA is used =
across an=20
  Enterprise Architecture or multiple campus type topologies then the =
policy=20
  controls "MAY" want to require this end-to-end functionality. =
</FONT></DIV>
  <DIV><FONT size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT size=3D2>Likewise, if the NEA tools were used for something =
like=20
  Microsoft Genuine Advantage compliance, then it would definently be a =
MUST one=20
  would think.</FONT></DIV>
  <DIV><FONT size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT size=3D2>The real issue is that the OPERATIONS POLICIES =
(OP)=20
  &nbsp;will determine whether certain functions or criteria are put in =
place,=20
  so depending on the OP's the state of this control will changed so you =
need to=20
  be able to say MUST in some cases and MAY or OPTIONALLY in=20
others.</FONT></DIV>
  <BLOCKQUOTE=20
  style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
    <DIV><FONT size=3D2></FONT>&nbsp;</DIV>
    <DIV><FONT face=3D"Times New Roman" size=3D2></FONT>&nbsp;</DIV>
    <DIV><FONT face=3DArial size=3D2>The current PA security protections =
were=20
    included to provide for end to end <BR>(e.g. Posture Collector to =
Posture=20
    Validator) authentication, integrity and</FONT></DIV>
    <DIV><FONT face=3DArial size=3D2>confidentiality protection of the =
PA=20
    attributes<SPAN class=3D734362521-06092007>.&nbsp; These=20
    attributes&nbsp;m</SPAN>ight travel to remote <BR>systems hosting =
the=20
    posture validator that are separate from the posture <BR>transport =
server=20
    thus PT protections are no longer present.&nbsp; </FONT></DIV>
    <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV></BLOCKQUOTE>
  <DIV><FONT face=3DArial size=3D2>OK so this could be a situation =
perhaps where a=20
  Trusted Third Party was operating the NEA master PV so it needs some =
tunnel to=20
  communicate with the other components of the system. I think this is =
'doable'=20
  with the simple addition of a requirement "that all NEA components =
communicate=20
  in a secure manner" which meets Industry Security Requirements for =
those=20
  application's bases. This type of disclaimer means that the NEA Spec =
itself=20
  doesnt need to address these inter-nodal security issues =
IMHO.</FONT></DIV>
  <BLOCKQUOTE=20
  style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
    <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
    <DIV><FONT face=3DArial size=3D2>Several folks <BR>have mentioned =
that the=20
    protection of the link between the posture transport <BR>server and =
remote=20
    posture validator could be provided in other less complex <BR>ways =
and so=20
    requiring a standards based PA security protocol isn't critical =
<BR>and will=20
    just make our deliverable more difficult.&nbsp; </FONT></DIV>
    <DIV><FONT size=3D2></FONT>&nbsp;</DIV></BLOCKQUOTE>
  <DIV><FONT size=3D2>The problem here isnt NEA - its the use =
environments which=20
  NEA is to be used in and many of them already have their own security =
and=20
  audit requirements.&nbsp;</FONT></DIV>
  <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
  <BLOCKQUOTE=20
  style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
    <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
    <DIV><FONT face=3DArial size=3D2>We also would need to<BR>specify a =
solution=20
    consistent with the principals of RFC 4107 which could<BR>be=20
    challenging.</FONT></DIV>
    <DIV><FONT size=3D2></FONT>&nbsp;</DIV></BLOCKQUOTE>
  <DIV><FONT size=3D2>Understood - EAP is complex...</FONT></DIV>
  <BLOCKQUOTE=20
  style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
    <DIV><FONT size=3D2></FONT>&nbsp;</DIV>
    <DIV><FONT face=3DArial size=3D2>We would like to request that the =
working group=20
    discuss whether we should <BR>downgrade PA-5 to SHOULD, MAY or =
remove it all=20
    together.&nbsp; </FONT></DIV>
    <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV></BLOCKQUOTE>
  <DIV><FONT face=3DArial size=3D2>I think PA-5's status is dependant on =
the end use=20
  model and based therein, the existing language of its description =
needs to be=20
  changed to reflect that.</FONT></DIV>
  <BLOCKQUOTE=20
  style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
    <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
    <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
    <DIV><FONT face=3DArial size=3D2>In order<BR>to bound the discussion =
we would=20
    like to conclude the discussion by 10PM PDT<BR>on thursday, Sept =
20th (2=20
    weeks).&nbsp; This will allows the editors time to make <BR>any =
revisions to=20
    the specification and re-submit in a timely manner so we =
can<BR><SPAN=20
    class=3D734362521-06092007>product a new I-D ready for a =
</SPAN>re-start the=20
    WGLC process.</FONT></DIV>
    <DIV><FONT size=3D2></FONT>&nbsp;</DIV></BLOCKQUOTE>
  <DIV>
  <DIV><FONT face=3D"Times New Roman" size=3D2>Existing:</FONT></DIV>
  <DIV><FONT face=3D"Times New Roman" size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>&nbsp;"PA-5 The PA protocol MUST =
provide=20
  authentication, integrity, and =
<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;=20
  confidentiality of attributes communicated between a Posture=20
  <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Collector and Posture =
Validator. This=20
  enables end-to-end security <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
across a=20
  NEA deployment that might involve traversal of several=20
  <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; systems."</FONT></DIV>
  <DIV><FONT size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT size=3D2>Proposed:</FONT></DIV>
  <DIV>
  <DIV>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>&nbsp;"PA-5 The PA protocol can be =
used in=20
  conformance with other</FONT></DIV>
  <DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; tools =
and BCPs=20
  which will set at a policy level the requirements </FONT></DIV>
  <DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; for =
which=20
  components of the NEA specification are mandatory, </FONT></DIV>
  <DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; and =
which are=20
  optional. In all instances NEA systems rely on&nbsp;</FONT></DIV>
  <DIV><FONT face=3DArial =
size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;the integrity=20
  of the underlying Network System to operate on</FONT></DIV>
  <DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp;&nbsp; and so in =
these=20
  instances the PA MUST provide=20
  =
authentication,&nbsp;and&nbsp;<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;confident=
iality=20
  of attributes communicated between a=20
  Posture&nbsp;<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;NEA Components=20
  (Posture&nbsp;Collector's and Posture Validator). </FONT></DIV>
  <DIV><FONT face=3D"Times New Roman" =
size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;This=20
  Integrity and confidentiality of communications may be =
integrated</FONT></DIV>
  <DIV><FONT face=3D"Times New Roman" =
size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;into=20
  NEA implementations or those implementations may rely on =
other</FONT></DIV>
  <DIV><FONT face=3D"Times New Roman"=20
  size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;protocols and technologies to =
supply=20
  those operation's features for the</FONT></DIV>
  <DIV><FONT face=3D"Times New Roman" size=3D2>&nbsp;&nbsp;&nbsp;&nbsp; =
NEA=20
  suite.</FONT></DIV>
  <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp; This flexibility =
in NEA design=20
  and deployment enables </FONT></DIV>
  <DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp; end-to-end=20
  security&nbsp;across a NEA</FONT><FONT face=3DArial=20
  size=3D2>&nbsp;deployment&nbsp;which may </FONT></DIV>
  <DIV><FONT face=3DArial size=3D2>&nbsp;&nbsp;&nbsp; involve traversal =
of=20
  several&nbsp;heterogenous</FONT><FONT face=3DArial size=3D2> networks =
and their=20
  systems."</FONT></DIV>
  <DIV><FONT size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT size=3D2>/TSG</FONT></DIV></DIV></DIV>
  <BLOCKQUOTE=20
  style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
    <DIV><FONT face=3D"Times New Roman" size=3D2></FONT>&nbsp;</DIV>
    <DIV><FONT face=3DArial size=3D2>Paul</FONT></DIV>
    <P>
    <HR>

    <P></P>_______________________________________________<BR>Nea =
mailing=20
    =
list<BR>Nea@ietf.org<BR>https://www1.ietf.org/mailman/listinfo/nea<BR></B=
LOCKQUOTE></BLOCKQUOTE></BODY></HTML>

------_=_NextPart_001_01C7F0D5.ECF1AC00--



--===============0179671529==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea

--===============0179671529==--





From nea-bounces@ietf.org Mon Sep 10 16:10:35 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IUpaY-0007PB-9y; Mon, 10 Sep 2007 16:10:30 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IUpaW-0007Nw-SZ
	for nea-confirm+ok@megatron.ietf.org; Mon, 10 Sep 2007 16:10:28 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IUpaW-0007JO-B1
	for nea@ietf.org; Mon, 10 Sep 2007 16:10:28 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IUpaU-0002Lw-MU
	for nea@ietf.org; Mon, 10 Sep 2007 16:10:28 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id hsmhue0171s1 for <nea@ietf.org>;
	Mon, 10 Sep 2007 13:07:02 -0700 (envelope-from
	<gtomlinson@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9143)
	with ESMTP; Mon, 10 Sep 2007 13:07:01 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Mon, 10 Sep 2007 13:10:28 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Date: Mon, 10 Sep 2007 13:10:38 -0700
Message-ID: <4DF8F0A275770B4387764386ABEF4D7A1008A8@us0exb10.us.sonicwall.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acfz5qfsPz1SyEhoRu2ib5KW8a/VZA==
From: <gtomlinson@SonicWALL.com>
To: <nea@ietf.org>
X-OriginalArrivalTime: 10 Sep 2007 20:10:28.0317 (UTC)
	FILETIME=[A1A2B0D0:01C7F3E6]
X-Mlf-Version: 6.0.1.9143
X-Mlf-UniqueId: o200709102007010011925
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 10dcc25e55b9b5f7d6ded516404bdc4c
Subject: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Content-Type: multipart/mixed; boundary="===============1453852598=="
Errors-To: nea-bounces@ietf.org

This is a multi-part message in MIME format.

--===============1453852598==
Content-class: urn:content-classes:message
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C7F3E6.A09A5D57"

This is a multi-part message in MIME format.

------_=_NextPart_001_01C7F3E6.A09A5D57
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

Hello NEA WG,

Several of us at SonicWALL have recently been following the NEA WG in
general and specifically draft-ietf-nea-requirements-03 and=20
draft-ietf-nea-requirements-04.  I have the following comments
pertaining to draft-ietf-nea-requirements-04.

Page 32 Section 7.1, Requirement C-9
At a minimum, C-9 should explicitly require UTF-8 encodings be supported
in strings in order to adhere to the guidance specified in RFC 2277.  =20
Is there any reason to support character sets other than UTF-8?
Supporting only UTF-8 makes the protocol straightforward on interpreting
string=20
encodings and normalization to locale specific character sets is
generally easy to implement on most OS platforms using their intrinsic
libraries.

Page 33, Section 7.2, Requirement PA-1
The NEA model seems to imply an extensible standard attribute name
space, but PA-1 does not incorporate this as a requirement. Nor
does PA-1 specify a requirement for the standard attribute name space
schema to be uniquely identifiable from vendor attribute name=20
spaces.=20

Page 33, Section 7.2, Requirement PA-2
The NEA model implies a plurality of extensible vendor attribute name
spaces and concurrent instances of each schema, but=20
PA-2 does not  incorporate this as a requirement.  Nor does PA-2 specify
a requirement for vendor attribute name space schemas to=20
be uniquely identifiable.

Page 33, Section 7.2, Requirement PA-4
The network owner security policy may restrict giving the user complete
visibility into why an assessment failed in order to prevent=20
exposing unnecessary information to potential attackers, especially
prior to user authentication.  With this in mind, please consider=20
changing "This enables" to "This optionally enables the Posture Client,
subject to disclosure policy of the NEA Server, ".

Page 33, Section 7.12 Requirement PA-5
I am little confused over the recent comments debating downgrading
[MUST] to [SHOULD] or [MAY].  The decision to utilize end-to-end=20
cryptography is use case topology dependent.  If these use cases are
deemed to be in scope, then doesn't it make sense for this to be a=20
[MUST] requirement and amend the verbiage to read "... MUST be
optionally capable of providing ..."?


Page 34 Section 7.2
I noticed the former PA-8 requirement in draft-ietf-nea-requirements-03
was dropped; apparently in favor of the more generalized new C-9=20
requirement in draft-ietf-nea-requirements-04.  I believe there still
exists a need for a PA specific I18N string encoding requirement for
non-
displayable request and response attributes. Request attributes for end
point artifacts such as File, Directory, Process Name, etc are used in=20
common practice today in NEA equivalent deployments I am aware of.
Again, as mentioned in the comment regarding the C-9 requirement,=20
UTF-8 makes things unambiguous and at a minimum must be a supported
encoding.

Page 34, Section 7.3 Requirement PB-3
Is there any reason in PB-3 not to specify identifiers be based upon SMI
enterprise numbers?  This requirement looks much like that=20
already faced by SNMP and LDAP; and we already have the IANA machinery
to leverage here.

Page 35, Section 7.4 Requirement PT-2
In addition to mutual authentication specified in PT-2, the PT protocol
must also be capable of supporting server side only authentication=20
as an alternative to mutual authentication.  Server side only
authentication is common practice today in SSL VPNs. With this in mind,
please=20
consider changing "mutual authentication" to "mutual and server side
only authentication schemes".

Page 44, Section 9.1 Paragraph 2
In order to prevent privacy abuse by network owners, PA request
attributes must be restricted to information  known apriori  rather=20
than be open ended in order to prevent dredging for artifact
information. For example a PV looking for a domain named=20
"mydomain.com" should be allowed, whereas requesting a domain named "*"
should be disallowed by the PC.

Page 46, Section 9.2  Paragraph 2
I don't think Model 1 is practical or even possible in some cases.  This
model works only when both the PC and PV have an immutable=20
set of attributes known apriori by each.  However, it is common practice
in existing SSL VPN s to request information on the existence of files,=20
directories, processes and other artifacts that are configurable
assessment policy on the NEA Server. In order for model 1 to potentially
work,=20
the NEA client would have to send enormous amounts of information;
potentially an enumeration of its entire file name space, process name
space,=20
registry name space, etc.=20

Page 46, Section 9.2 Paragraph 4
Model 3 is used in common practice today by many of the NEA like
deployments.  At a minimum, it must be supported .  I can also see model
2=20
being used effectively by a subset of PC/PV providers.  It occurs to me,
the decision to use either model 2 or model 3 is one the PC/PV provider=20
makes and that the NEA protocols could consider supporting both models
if necessary; but Model 3 appears to be a superset of Model 2 and as=20
such so it might make sense to simply the protocol and only support
model 3. My preference is to always make things simpler if possible.


Gary Tomlinson



------_=_NextPart_001_01C7F3E6.A09A5D57
Content-Type: text/html;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Dus-ascii">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
6.5.7652.24">
<TITLE>Newcomer comments on draft-ietf-nea-requirements-04</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/rtf format -->

<P><FONT SIZE=3D2 FACE=3D"Arial">Hello NEA WG,</FONT>
</P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Several of us at SonicWALL have =
recently been following the NEA WG in general and specifically =
draft-ietf-nea-requirements-03 and </FONT></P>

<P><FONT SIZE=3D2 FACE=3D"Arial">draft-ietf-nea-requirements-04.&nbsp; I =
have the following comments pertaining to =
draft-ietf-nea-requirements-04.</FONT>
</P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Page 32 Section 7.1, Requirement =
C-9</FONT>

<BR><FONT SIZE=3D2 FACE=3D"Times New Roman">At a minimum, C-9 should =
explicitly require UTF-8 encodings be supported in strings in order to =
adhere to the guidance specified in RFC 2277.&nbsp;&nbsp; </FONT></P>

<P><FONT SIZE=3D2 FACE=3D"Times New Roman">Is there any reason to =
support character sets other than UTF-8?&nbsp; Supporting only UTF-8 =
makes the protocol straightforward on interpreting string </FONT></P>

<P><FONT SIZE=3D2 FACE=3D"Times New Roman">encodings and normalization =
to locale specific character sets is generally easy to implement on most =
OS platforms using their intrinsic libraries.</FONT></P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Page 33, Section 7.2, Requirement =
PA-1</FONT>

<BR><FONT SIZE=3D2 FACE=3D"Arial">The NEA model seems to imply an =
extensible standard attribute name space, but PA-1 does not incorporate =
this as a requirement. Nor</FONT></P>

<P><FONT SIZE=3D2 FACE=3D"Arial">does PA-1 specify a requirement for the =
standard attribute name space schema to be uniquely identifiable from =
vendor attribute name </FONT></P>

<P><FONT SIZE=3D2 FACE=3D"Arial">spaces. </FONT>
</P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Page 33, Section 7.2, Requirement =
PA-2</FONT>

<BR><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">The NEA model =
implies a plurality of extensible vendor attribute name spaces and =
concurrent instances of each schema, but </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">PA-2 does not&nbsp; =
incorporate this as a requirement.&nbsp; Nor does PA-2 specify a =
requirement for vendor attribute name space schemas to </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">be uniquely =
identifiable.</FONT>
</P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Page 33, Section 7.2, Requirement =
PA-4</FONT>

<BR><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">The network owner =
security policy may restrict giving the user complete visibility into =
why an assessment failed in order to prevent </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">exposing unnecessary =
information to potential attackers, especially prior to user =
authentication.&nbsp; With this in mind, please consider </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">changing &quot;This =
enables&quot; to &quot;This optionally enables the Posture Client, =
subject to disclosure policy of the NEA Server, &#8221;.</FONT></P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Page 33, Section 7.12 Requirement =
PA-5</FONT>

<BR><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">I am little confused =
over the recent comments debating downgrading [MUST] to [SHOULD] or =
[MAY].&nbsp; The decision to utilize end-to-end </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">cryptography is use =
case topology dependent.&nbsp; If these use cases are deemed to be in =
scope, then doesn't it make sense for this to be a </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">[MUST] requirement =
and amend the verbiage to read &quot;&#8230; MUST be optionally capable =
of providing &#8230;&quot;?</FONT>
</P>
<BR>

<P><FONT SIZE=3D2 FACE=3D"Arial">Page 34 Section 7.2</FONT>

<BR><FONT SIZE=3D2 FACE=3D"Times New Roman">I noticed the former PA-8 =
requirement in draft-ietf-nea-requirements-03 was dropped; apparently in =
favor of the more generalized new C-9 </FONT></P>

<P><FONT SIZE=3D2 FACE=3D"Times New Roman">requirement in =
draft-ietf-nea-requirements-04.&nbsp; I believe there still exists a =
need for a PA specific I18N string encoding requirement for =
non-</FONT></P>

<P><FONT SIZE=3D2 FACE=3D"Times New Roman">displayable request and =
response attributes. Request attributes for end point artifacts such as =
File, Directory, Process Name, etc are used in </FONT></P>

<P><FONT SIZE=3D2 FACE=3D"Times New Roman">common practice today in NEA =
equivalent deployments I am aware of.&nbsp; Again, as mentioned in the =
comment regarding the C-9 requirement, </FONT></P>

<P><FONT SIZE=3D2 FACE=3D"Times New Roman">UTF-8 makes things =
unambiguous and at a minimum must be a supported encoding.</FONT>
</P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Page 34, Section 7.3 Requirement =
PB-3</FONT>

<BR><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">Is there any reason =
in PB-3 not to specify identifiers be based upon SMI enterprise =
numbers?&nbsp; This requirement looks much like that </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">already faced by SNMP =
and LDAP; and we already have the IANA machinery to leverage =
here.</FONT>
</P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Page 35, Section 7.4 Requirement =
PT-2</FONT>

<BR><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">In addition to =
mutual authentication specified in PT-2, the PT protocol must also be =
capable of supporting server side only authentication </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">as an alternative to =
mutual authentication.&nbsp; Server side only authentication is common =
practice today in SSL VPNs. With this in mind, please </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">consider changing =
&#8220;mutual authentication&#8221; to &#8220;mutual and server side =
only authentication schemes&#8221;.</FONT>
</P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Page 44, Section 9.1 Paragraph 2</FONT>

<BR><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">In order to prevent =
privacy abuse by network owners, PA request attributes must be =
restricted to information&nbsp; known apriori&nbsp; rather </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">than be open ended in =
order to prevent dredging for artifact information. For example a PV =
looking for a domain named </FONT>

<BR><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">&#8220;mydomain.com&#8221; should be allowed, whereas =
requesting a domain named &#8220;*&#8221; should be disallowed by the =
PC.</FONT>
</P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Page 46, Section 9.2&nbsp; Paragraph =
2</FONT>

<BR><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">I don't think Model =
1 is practical or even possible in some cases.&nbsp; This model works =
only when both the PC and PV have an immutable </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">set of attributes =
known apriori by each.&nbsp; However, it is common practice in existing =
SSL VPN s to request information on the existence of files, </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">directories, =
processes and other artifacts that are configurable assessment policy on =
the NEA Server. In order for model 1 to potentially work, </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">the NEA client would =
have to send enormous amounts of information; potentially an enumeration =
of its entire file name space, process name space, </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">registry name space, =
etc.</FONT>=20
</P>

<P><FONT SIZE=3D2 FACE=3D"Arial">Page 46, Section 9.2 Paragraph 4</FONT>

<BR><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">Model 3 is used in =
common practice today by many of the NEA like deployments.&nbsp; At a =
minimum, it must be supported .&nbsp; I can also see model 2 </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">being used =
effectively by a subset of PC/PV providers.&nbsp; It occurs to me, the =
decision to use either model 2 or model 3 is one the PC/PV provider =
</FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">makes and that the =
NEA protocols could consider supporting both models if necessary; but =
Model 3 appears to be a superset of Model 2 and as </FONT></P>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">such so it might make =
sense to simply the protocol and only support model 3. My preference is =
to always make things simpler if possible.</FONT></P>
<BR>

<P><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">Gary Tomlinson</FONT>
</P>
<BR>

</BODY>
</HTML>
------_=_NextPart_001_01C7F3E6.A09A5D57--



--===============1453852598==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea

--===============1453852598==--





From nea-bounces@ietf.org Mon Sep 10 16:14:35 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IUpeV-0001J5-1r; Mon, 10 Sep 2007 16:14:35 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IUpeT-0001Hr-VF
	for nea-confirm+ok@megatron.ietf.org; Mon, 10 Sep 2007 16:14:33 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IUpeT-0001F1-H4
	for nea@ietf.org; Mon, 10 Sep 2007 16:14:33 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IUpeS-0004AR-L4
	for nea@ietf.org; Mon, 10 Sep 2007 16:14:33 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id hsmidq0171s5 for <nea@ietf.org>;
	Mon, 10 Sep 2007 13:11:07 -0700 (envelope-from
	<panandam@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9143)
	with ESMTP; Mon, 10 Sep 2007 13:11:07 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Mon, 10 Sep 2007 13:14:33 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Date: Mon, 10 Sep 2007 13:14:27 -0700
Message-ID: <D37A04F05A87044A8DCEEC0BF9A3AB1D07D024@us0exb10.us.sonicwall.com>
In-Reply-To: <4DF8F0A275770B4387764386ABEF4D7A1008A8@us0exb10.us.sonicwall.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acfz5qfsPz1SyEhoRu2ib5KW8a/VZAAABvKQ
References: <4DF8F0A275770B4387764386ABEF4D7A1008A8@us0exb10.us.sonicwall.com>
From: <panandam@SonicWALL.com>
To: <nea@ietf.org>
X-OriginalArrivalTime: 10 Sep 2007 20:14:33.0988 (UTC)
	FILETIME=[34111C40:01C7F3E7]
X-Mlf-Version: 6.0.1.9143
X-Mlf-UniqueId: o200709102011070013705
X-Spam-Score: 0.0 (/)
X-Scan-Signature: e9eeacd7fe925d5f7faae01ed8f85b97
Subject: [Nea] RE: Newcomer comments on draft-ietf-nea-requirements-04
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Content-Type: multipart/mixed; boundary="===============0058558594=="
Errors-To: nea-bounces@ietf.org

This is a multi-part message in MIME format.

--===============0058558594==
Content-class: urn:content-classes:message
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C7F3E7.334B6F46"

This is a multi-part message in MIME format.

------_=_NextPart_001_01C7F3E7.334B6F46
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

I have a few additional comments, following Gary's, that also pertain to
draft-ietf-nea-requirements-04.

Page 35 Section 7.4

Please consider adding a new requirement for the PT protocol to support
non-persistent network connections. This is especially important for
mobile devices that roam or have air time charges.  This has an impact
on reassessments triggered by the NEA server; in some cases (e.g. NAT
traversal), the only option may be for the NEA client to periodically
contact the NEA server to determine whether a reassessment is necessary.

Page 44 Section 9.1 Paragraph 3

As Gary stated, due to security policy, network owners may wish not to
disclose the attributes being requested. Imagine an attacker who
restricts all posture collection in order to determine what the network
operators are looking for.

Page 45 Section 9.1 Paragraph 6

This is a continuation of the thought in the comment above and concerns
this item in the NEA disclosure statement: "What information is
required". Such disclosure must be subject to  the security policy of
the network owner. What is the intent of specificity in this guidance?
Does the network owner have the latitude of providing a more general
disclosure statement?


Parvez Anandam

_____________________________________________
From: Gary Tomlinson=20
Sent: Monday, September 10, 2007 1:11 PM
To: nea@ietf.org
Subject: Newcomer comments on draft-ietf-nea-requirements-04


Hello NEA WG,

Several of us at SonicWALL have recently been following the NEA WG in
general and specifically draft-ietf-nea-requirements-03 and=20
draft-ietf-nea-requirements-04.  I have the following comments
pertaining to draft-ietf-nea-requirements-04.

Page 32 Section 7.1, Requirement C-9
At a minimum, C-9 should explicitly require UTF-8 encodings be supported
in strings in order to adhere to the guidance specified in RFC 2277.  =20
Is there any reason to support character sets other than UTF-8?
Supporting only UTF-8 makes the protocol straightforward on interpreting
string=20
encodings and normalization to locale specific character sets is
generally easy to implement on most OS platforms using their intrinsic
libraries.

Page 33, Section 7.2, Requirement PA-1
The NEA model seems to imply an extensible standard attribute name
space, but PA-1 does not incorporate this as a requirement. Nor
does PA-1 specify a requirement for the standard attribute name space
schema to be uniquely identifiable from vendor attribute name=20
spaces.=20

Page 33, Section 7.2, Requirement PA-2
The NEA model implies a plurality of extensible vendor attribute name
spaces and concurrent instances of each schema, but=20
PA-2 does not  incorporate this as a requirement.  Nor does PA-2 specify
a requirement for vendor attribute name space schemas to=20
be uniquely identifiable.

Page 33, Section 7.2, Requirement PA-4
The network owner security policy may restrict giving the user complete
visibility into why an assessment failed in order to prevent=20
exposing unnecessary information to potential attackers, especially
prior to user authentication.  With this in mind, please consider=20
changing "This enables" to "This optionally enables the Posture Client,
subject to disclosure policy of the NEA Server, ".

Page 33, Section 7.12 Requirement PA-5
I am little confused over the recent comments debating downgrading
[MUST] to [SHOULD] or [MAY].  The decision to utilize end-to-end=20
cryptography is use case topology dependent.  If these use cases are
deemed to be in scope, then doesn't it make sense for this to be a=20
[MUST] requirement and amend the verbiage to read "... MUST be
optionally capable of providing ..."?


Page 34 Section 7.2
I noticed the former PA-8 requirement in draft-ietf-nea-requirements-03
was dropped; apparently in favor of the more generalized new C-9=20
requirement in draft-ietf-nea-requirements-04.  I believe there still
exists a need for a PA specific I18N string encoding requirement for
non-
displayable request and response attributes. Request attributes for end
point artifacts such as File, Directory, Process Name, etc are used in=20
common practice today in NEA equivalent deployments I am aware of.
Again, as mentioned in the comment regarding the C-9 requirement,=20
UTF-8 makes things unambiguous and at a minimum must be a supported
encoding.

Page 34, Section 7.3 Requirement PB-3
Is there any reason in PB-3 not to specify identifiers be based upon SMI
enterprise numbers?  This requirement looks much like that=20
already faced by SNMP and LDAP; and we already have the IANA machinery
to leverage here.

Page 35, Section 7.4 Requirement PT-2
In addition to mutual authentication specified in PT-2, the PT protocol
must also be capable of supporting server side only authentication=20
as an alternative to mutual authentication.  Server side only
authentication is common practice today in SSL VPNs. With this in mind,
please=20
consider changing "mutual authentication" to "mutual and server side
only authentication schemes".

Page 44, Section 9.1 Paragraph 2
In order to prevent privacy abuse by network owners, PA request
attributes must be restricted to information  known apriori  rather=20
than be open ended in order to prevent dredging for artifact
information. For example a PV looking for a domain named=20
"mydomain.com" should be allowed, whereas requesting a domain named "*"
should be disallowed by the PC.

Page 46, Section 9.2  Paragraph 2
I don't think Model 1 is practical or even possible in some cases.  This
model works only when both the PC and PV have an immutable=20
set of attributes known apriori by each.  However, it is common practice
in existing SSL VPN s to request information on the existence of files,=20
directories, processes and other artifacts that are configurable
assessment policy on the NEA Server. In order for model 1 to potentially
work,=20
the NEA client would have to send enormous amounts of information;
potentially an enumeration of its entire file name space, process name
space,=20
registry name space, etc.=20

Page 46, Section 9.2 Paragraph 4
Model 3 is used in common practice today by many of the NEA like
deployments.  At a minimum, it must be supported .  I can also see model
2=20
being used effectively by a subset of PC/PV providers.  It occurs to me,
the decision to use either model 2 or model 3 is one the PC/PV provider=20
makes and that the NEA protocols could consider supporting both models
if necessary; but Model 3 appears to be a superset of Model 2 and as=20
such so it might make sense to simply the protocol and only support
model 3. My preference is to always make things simpler if possible.


Gary Tomlinson



------_=_NextPart_001_01C7F3E7.334B6F46
Content-Type: text/html;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Dus-ascii">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
6.5.7652.24">
<TITLE>RE: Newcomer comments on draft-ietf-nea-requirements-04</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/rtf format -->

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 =
FACE=3D"Arial">I have a</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"> <FONT SIZE=3D2 =
FACE=3D"Arial">few</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 =
FACE=3D"Arial"> additional comments, following =
Gary&#8217;s,</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"> <FONT SIZE=3D2 =
FACE=3D"Arial">that also</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"> <FONT SIZE=3D2 =
FACE=3D"Arial">pertain</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 =
FACE=3D"Arial"></FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"> <FONT SIZE=3D2 =
FACE=3D"Arial">to draft-ietf-nea-requirements-04.</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Page 35 =
Section 7.4</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT FACE=3D"Calibri">Please consider =
adding a new requirement for the PT protocol to support non-persistent =
network connections. This is especially important for mobile devices =
that roam or have air time charges. &nbsp;This has an impact on =
reassessments triggered by the NEA server; in some cases (e.g. NAT =
traversal), the only option may be for the NEA client to periodically =
contact the NEA server to determine whether a reassessment is =
necessary.</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT FACE=3D"Calibri">Page 44 Section =
9.1 Paragraph 3</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT FACE=3D"Calibri">As Gary stated, =
due to security policy, network owners may wish not to disclose the =
attributes being requested. Imagine an attacker who restricts all =
posture collection in order to determine what the network operators are =
looking for.</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT FACE=3D"Calibri">Page 45 Section =
9.1 Paragraph 6</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT FACE=3D"Calibri">This is a =
continuation of the thought in the comment above and concerns this item =
in the</FONT></SPAN><SPAN LANG=3D"en-us"> <FONT =
FACE=3D"Calibri">NEA</FONT></SPAN><SPAN LANG=3D"en-us"> <FONT =
FACE=3D"Calibri">disclosure statement: &#8220;What information is =
required&#8221;. Such disclosure</FONT></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial"> must be subject to&nbsp; =
the security policy of the network owner. What is the intent of =
specificity in this guidance?&nbsp; Does the network owner have the =
latitude</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 =
FACE=3D"Arial"> of</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 =
FACE=3D"Arial"> provid</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 =
FACE=3D"Arial">ing</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 =
FACE=3D"Arial"> a more general disclosure statement?</FONT></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Parvez =
Anandam</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 =
FACE=3D"Tahoma">_____________________________________________<BR>
</FONT></SPAN><SPAN LANG=3D"en-us"><B></B></SPAN><SPAN =
LANG=3D"en-us"><B></B></SPAN><SPAN LANG=3D"en-us"><B><FONT SIZE=3D2 =
FACE=3D"Tahoma">From:</FONT></B></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 =
FACE=3D"Tahoma"> Gary Tomlinson<BR>
</FONT></SPAN><SPAN LANG=3D"en-us"><B></B></SPAN><SPAN =
LANG=3D"en-us"><B></B></SPAN><SPAN LANG=3D"en-us"><B><FONT SIZE=3D2 =
FACE=3D"Tahoma">Sent:</FONT></B></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 =
FACE=3D"Tahoma"> Monday, September 10, 2007 1:11 PM<BR>
</FONT></SPAN><SPAN LANG=3D"en-us"><B></B></SPAN><SPAN =
LANG=3D"en-us"><B></B></SPAN><SPAN LANG=3D"en-us"><B><FONT SIZE=3D2 =
FACE=3D"Tahoma">To:</FONT></B></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 =
FACE=3D"Tahoma"> nea@ietf.org<BR>
</FONT></SPAN><SPAN LANG=3D"en-us"><B></B></SPAN><SPAN =
LANG=3D"en-us"><B></B></SPAN><SPAN LANG=3D"en-us"><B><FONT SIZE=3D2 =
FACE=3D"Tahoma">Subject:</FONT></B></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Tahoma"> Newcomer comments on =
draft-ietf-nea-requirements-04</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 =
FACE=3D"Arial">Hello NEA WG,</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Several =
of us at SonicWALL have recently been following the NEA WG in general =
and specifically draft-ietf-nea-requirements-03 and </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 =
FACE=3D"Arial">draft-ietf-nea-requirements-04.&nbsp; I have the =
following comments pertaining to =
draft-ietf-nea-requirements-04.</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Page 32 =
Section 7.1, Requirement C-9</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Times =
New Roman">At a minimum, C-9 should explicitly require UTF-8 encodings =
be supported in strings in order to adhere to the guidance specified in =
RFC 2277.&nbsp;&nbsp; </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Times New =
Roman">Is there any reason to support character sets other than =
UTF-8?&nbsp; Supporting only UTF-8 makes the protocol straightforward on =
interpreting string </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Times New =
Roman">encodings and normalization to locale specific character sets is =
generally easy to implement on most OS platforms using their intrinsic =
libraries.</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Page 33, =
Section 7.2, Requirement PA-1</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">The NEA =
model seems to imply an extensible standard attribute name space, but =
PA-1 does not incorporate this as a requirement. Nor</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">does =
PA-1 specify a requirement for the standard attribute name space schema =
to be uniquely identifiable from vendor attribute name =
</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">spaces. =
</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Page 33, =
Section 7.2, Requirement PA-2</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" =
SIZE=3D2 FACE=3D"Arial">The NEA model implies a plurality of extensible =
vendor attribute name spaces and concurrent instances of each schema, =
but </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">PA-2 does not&nbsp; incorporate this as a =
requirement.&nbsp; Nor does PA-2 specify a requirement for vendor =
attribute name space schemas to </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">be uniquely identifiable.</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Page 33, =
Section 7.2, Requirement PA-4</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" =
SIZE=3D2 FACE=3D"Arial">The network owner security policy may restrict =
giving the user complete visibility into why an assessment failed in =
order to prevent </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">exposing unnecessary information to potential attackers, =
especially prior to user authentication.&nbsp; With this in mind, please =
consider </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">changing &quot;This enables&quot; to &quot;This =
optionally enables the Posture Client, subject to disclosure policy of =
the NEA Server, &#8221;.</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Page 33, =
Section 7.12 Requirement PA-5</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" =
SIZE=3D2 FACE=3D"Arial">I am little confused over the recent comments =
debating downgrading [MUST] to [SHOULD] or [MAY].&nbsp; The decision to =
utilize end-to-end </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">cryptography is use case topology dependent.&nbsp; If =
these use cases are deemed to be in scope, then doesn't it make sense =
for this to be a </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">[MUST] requirement and amend the verbiage to read =
&quot;&#8230; MUST be optionally capable of providing =
&#8230;&quot;?</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN></P>
<BR>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Page 34 =
Section 7.2</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Times =
New Roman">I noticed the former PA-8 requirement in =
draft-ietf-nea-requirements-03 was dropped; apparently in favor of the =
more generalized new C-9 </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Times New =
Roman">requirement in draft-ietf-nea-requirements-04.&nbsp; I believe =
there still exists a need for a PA specific I18N string encoding =
requirement for non-</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Times New =
Roman">displayable request and response attributes. Request attributes =
for end point artifacts such as File, Directory, Process Name, etc are =
used in </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Times New =
Roman">common practice today in NEA equivalent deployments I am aware =
of.&nbsp; Again, as mentioned in the comment regarding the C-9 =
requirement, </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Times New =
Roman">UTF-8 makes things unambiguous and at a minimum must be a =
supported encoding.</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Page 34, =
Section 7.3 Requirement PB-3</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" =
SIZE=3D2 FACE=3D"Arial">Is there any reason in PB-3 not to specify =
identifiers be based upon SMI enterprise numbers?&nbsp; This requirement =
looks much like that </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">already faced by SNMP and LDAP; and we already have the =
IANA machinery to leverage here.</FONT></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Page 35, =
Section 7.4 Requirement PT-2</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" =
SIZE=3D2 FACE=3D"Arial">In addition to mutual authentication specified =
in PT-2, the PT protocol must also be capable of supporting server side =
only authentication </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">as an alternative to mutual authentication.&nbsp; Server =
side only authentication is common practice today in SSL VPNs. With this =
in mind, please </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">consider changing &#8220;mutual authentication&#8221; to =
&#8220;mutual and server side only authentication =
schemes&#8221;.</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Page 44, =
Section 9.1 Paragraph 2</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" =
SIZE=3D2 FACE=3D"Arial">In order to prevent privacy abuse by network =
owners, PA request attributes must be restricted to information&nbsp; =
known apriori&nbsp; rather </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">than be open ended in order to prevent dredging for =
artifact information. For example a PV looking for a domain named =
</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">&#8220;mydomain.com&#8221; should be allowed, whereas =
requesting a domain named &#8220;*&#8221; should be disallowed by the =
PC.</FONT></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Page 46, =
Section 9.2&nbsp; Paragraph 2</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" =
SIZE=3D2 FACE=3D"Arial">I don't think Model 1 is practical or even =
possible in some cases.&nbsp; This model works only when both the PC and =
PV have an immutable </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">set of attributes known apriori by each.&nbsp; However, =
it is common practice in existing SSL VPN s to request information on =
the existence of files, </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">directories, processes and other artifacts that are =
configurable assessment policy on the NEA Server. In order for model 1 =
to potentially work, </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">the NEA client would have to send enormous amounts of =
information; potentially an enumeration of its entire file name space, =
process name space, </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">registry name space, etc.</FONT></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"> =
</SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT SIZE=3D2 FACE=3D"Arial">Page 46, =
Section 9.2 Paragraph 4</FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" =
SIZE=3D2 FACE=3D"Arial">Model 3 is used in common practice today by many =
of the NEA like deployments.&nbsp; At a minimum, it must be supported =
.&nbsp; I can also see model 2 </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">being used effectively by a subset of PC/PV =
providers.&nbsp; It occurs to me, the decision to use either model 2 or =
model 3 is one the PC/PV provider </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">makes and that the NEA protocols could consider =
supporting both models if necessary; but Model 3 appears to be a =
superset of Model 2 and as </FONT></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">such so it might make sense to simply the protocol and =
only support model 3. My preference is to always make things simpler if =
possible.</FONT></SPAN></P>
<BR>

<P DIR=3DLTR><SPAN LANG=3D"en-us"><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">Gary Tomlinson</FONT></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN></P>

<P DIR=3DLTR><SPAN LANG=3D"en-us"></SPAN><SPAN =
LANG=3D"en-us"></SPAN><SPAN LANG=3D"en-us"></SPAN></P>

</BODY>
</HTML>
------_=_NextPart_001_01C7F3E7.334B6F46--



--===============0058558594==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea

--===============0058558594==--





From nea-bounces@ietf.org Mon Sep 10 17:52:44 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IUrBT-0006SK-FV; Mon, 10 Sep 2007 17:52:43 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IUrBS-0006S1-J2
	for nea-confirm+ok@megatron.ietf.org; Mon, 10 Sep 2007 17:52:42 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IUrBS-0006RY-4q
	for nea@ietf.org; Mon, 10 Sep 2007 17:52:42 -0400
Received: from elasmtp-dupuy.atl.sa.earthlink.net ([209.86.89.62])
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IUrBQ-0006qE-1r
	for nea@ietf.org; Mon, 10 Sep 2007 17:52:42 -0400
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=earthlink.net;
	b=iA1HI5C0UDHtEQJCrWOy0T+KCKmDMbrz7WZyjxpshI+VERuP2nIG/Zfi/CqSJ8Kd;
	h=Received:Message-ID:From:To:References:Subject:Date:MIME-Version:Content-Type:X-Priority:X-MSMail-Priority:X-Mailer:X-MimeOLE:X-ELNK-Trace:X-Originating-IP;
Received: from [64.125.79.23] (helo=gw)
	by elasmtp-dupuy.atl.sa.earthlink.net with asmtp (Exim 4.34)
	id 1IUrBF-0002lS-Ct; Mon, 10 Sep 2007 17:52:29 -0400
Message-ID: <021d01c7f3f4$eed1dfd0$174f7d40@home.glassey.com>
From: "todd glassey" <tglassey@earthlink.net>
To: <panandam@SonicWALL.com>,
	<nea@ietf.org>
References: <4DF8F0A275770B4387764386ABEF4D7A1008A8@us0exb10.us.sonicwall.com>
	<D37A04F05A87044A8DCEEC0BF9A3AB1D07D024@us0exb10.us.sonicwall.com>
Subject: Re: [Nea] RE: Newcomer comments on draft-ietf-nea-requirements-04
Date: Mon, 10 Sep 2007 14:52:31 -0700
MIME-Version: 1.0
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2900.3138
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.3138
X-ELNK-Trace: 01b7a7e171bdf5911aa676d7e74259b7b3291a7d08dfec79725e5de973ec17622f25338ba1056074350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c
X-Originating-IP: 64.125.79.23
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 63de1eb80dd4230367fc1d266729643d
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Content-Type: multipart/mixed; boundary="===============0587559593=="
Errors-To: nea-bounces@ietf.org

This is a multi-part message in MIME format.

--===============0587559593==
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_020E_01C7F3BA.36E6A3E0"

This is a multi-part message in MIME format.

------=_NextPart_000_020E_01C7F3BA.36E6A3E0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

RE: Newcomer comments on draft-ietf-nea-requirements-04Parvez - in =
spirit I understand what you are trying to do but I disagree and will =
enumerate why below...

---- Original Message -----=20
  From: panandam@SonicWALL.com=20
  To: nea@ietf.org=20
  Sent: Monday, September 10, 2007 1:14 PM
  Subject: [Nea] RE: Newcomer comments on draft-ietf-nea-requirements-04


  I have a few additional comments, following Gary's, that also pertain =
to draft-ietf-nea-requirements-04.

  Page 35 Section 7.4

  Please consider adding a new requirement for the PT protocol to =
support non-persistent network connections.=20

This will not be a viable thing to do in the real world no matter how =
nice it sounds I fear...

  This is especially important for mobile devices that roam or have air =
time charges.  This has an impact on reassessments triggered by the NEA =
server; in some cases (e.g. NAT traversal), the only option may be for =
the NEA client to periodically contact the NEA server to determine =
whether a reassessment is necessary.

I disagree...  Supporting Mobile Device (Nomadic Computing) and its =
Integrity for "devices that are based on non-persistant connections" may =
not be such a good idea since it will dramatically increase the overhead =
in any NEA system at a minimum and also will create a HUGE level of pain =
for people running mixed networks... For instance it would be rather =
easy to program a PDA to slam a NEA server with continuous requests for =
service after service.

To deal with these types of devices,  the NEA system will also have to =
maintain state tables for all of its local "recent" connections since =
under any reasonable policy requiring continuous certification of =
competence or NEA policy compliance means repeating the same set of =
messages again and again...=20

  Page 44 Section 9.1 Paragraph 3


  As Gary stated, due to security policy, network owners may wish not to =
disclose the attributes being requested.=20

Then why are they allowing people to plug into their networks at all =
without disclosing those policies...

  Imagine an attacker who restricts all posture collection in order to =
determine what the network operators are looking for.

But how would that system work in a NEA environment... it would be =
getting routing access to the switches right? So what is it that system =
would be doing?

Remember the key purpose of NEA is to BOTH allow one to force compliance =
with pre-existing NETWORK POLICIES in a new or recently powered-up =
system AND the continuously recertifying those systems which passed =
previously if they still continue to pass their tests.=20

To do that there needs to be some way of communicating what those =
requirements are to the client in an interactive mode.=20




  Page 45 Section 9.1 Paragraph 6

  This is a continuation of the thought in the comment above and =
concerns this item in the NEA disclosure statement: "What information is =
required". Such disclosure must be subject to  the security policy of =
the network owner. What is the intent of specificity in this guidance?  =
Does the network owner have the latitude of providing a more general =
disclosure statement?


  Parvez Anandam


  _____________________________________________
  From: Gary Tomlinson
  Sent: Monday, September 10, 2007 1:11 PM
  To: nea@ietf.org
  Subject: Newcomer comments on draft-ietf-nea-requirements-04


  Hello NEA WG,

  Several of us at SonicWALL have recently been following the NEA WG in =
general and specifically draft-ietf-nea-requirements-03 and=20

  draft-ietf-nea-requirements-04.  I have the following comments =
pertaining to draft-ietf-nea-requirements-04.

  Page 32 Section 7.1, Requirement C-9

  At a minimum, C-9 should explicitly require UTF-8 encodings be =
supported in strings in order to adhere to the guidance specified in RFC =
2277.  =20

  Is there any reason to support character sets other than UTF-8?  =
Supporting only UTF-8 makes the protocol straightforward on interpreting =
string=20

  encodings and normalization to locale specific character sets is =
generally easy to implement on most OS platforms using their intrinsic =
libraries.


  Page 33, Section 7.2, Requirement PA-1

  The NEA model seems to imply an extensible standard attribute name =
space, but PA-1 does not incorporate this as a requirement. Nor

  does PA-1 specify a requirement for the standard attribute name space =
schema to be uniquely identifiable from vendor attribute name=20

  spaces.=20

  Page 33, Section 7.2, Requirement PA-2

  The NEA model implies a plurality of extensible vendor attribute name =
spaces and concurrent instances of each schema, but=20

  PA-2 does not  incorporate this as a requirement.  Nor does PA-2 =
specify a requirement for vendor attribute name space schemas to=20

  be uniquely identifiable.


  Page 33, Section 7.2, Requirement PA-4

  The network owner security policy may restrict giving the user =
complete visibility into why an assessment failed in order to prevent=20

  exposing unnecessary information to potential attackers, especially =
prior to user authentication.  With this in mind, please consider=20

  changing "This enables" to "This optionally enables the Posture =
Client, subject to disclosure policy of the NEA Server, ".


  Page 33, Section 7.12 Requirement PA-5

  I am little confused over the recent comments debating downgrading =
[MUST] to [SHOULD] or [MAY].  The decision to utilize end-to-end=20

  cryptography is use case topology dependent.  If these use cases are =
deemed to be in scope, then doesn't it make sense for this to be a=20

  [MUST] requirement and amend the verbiage to read ". MUST be =
optionally capable of providing ."?



  Page 34 Section 7.2

  I noticed the former PA-8 requirement in =
draft-ietf-nea-requirements-03 was dropped; apparently in favor of the =
more generalized new C-9=20

  requirement in draft-ietf-nea-requirements-04.  I believe there still =
exists a need for a PA specific I18N string encoding requirement for =
non-

  displayable request and response attributes. Request attributes for =
end point artifacts such as File, Directory, Process Name, etc are used =
in=20

  common practice today in NEA equivalent deployments I am aware of.  =
Again, as mentioned in the comment regarding the C-9 requirement,=20

  UTF-8 makes things unambiguous and at a minimum must be a supported =
encoding.


  Page 34, Section 7.3 Requirement PB-3

  Is there any reason in PB-3 not to specify identifiers be based upon =
SMI enterprise numbers?  This requirement looks much like that=20

  already faced by SNMP and LDAP; and we already have the IANA machinery =
to leverage here.

  Page 35, Section 7.4 Requirement PT-2

  In addition to mutual authentication specified in PT-2, the PT =
protocol must also be capable of supporting server side only =
authentication=20

  as an alternative to mutual authentication.  Server side only =
authentication is common practice today in SSL VPNs. With this in mind, =
please=20

  consider changing "mutual authentication" to "mutual and server side =
only authentication schemes".

  Page 44, Section 9.1 Paragraph 2

  In order to prevent privacy abuse by network owners, PA request =
attributes must be restricted to information  known apriori  rather=20

  than be open ended in order to prevent dredging for artifact =
information. For example a PV looking for a domain named=20

  "mydomain.com" should be allowed, whereas requesting a domain named =
"*" should be disallowed by the PC.

  Page 46, Section 9.2  Paragraph 2

  I don't think Model 1 is practical or even possible in some cases.  =
This model works only when both the PC and PV have an immutable=20

  set of attributes known apriori by each.  However, it is common =
practice in existing SSL VPN s to request information on the existence =
of files,=20

  directories, processes and other artifacts that are configurable =
assessment policy on the NEA Server. In order for model 1 to potentially =
work,=20

  the NEA client would have to send enormous amounts of information; =
potentially an enumeration of its entire file name space, process name =
space,=20

  registry name space, etc.=20

  Page 46, Section 9.2 Paragraph 4

  Model 3 is used in common practice today by many of the NEA like =
deployments.  At a minimum, it must be supported .  I can also see model =
2=20

  being used effectively by a subset of PC/PV providers.  It occurs to =
me, the decision to use either model 2 or model 3 is one the PC/PV =
provider=20

  makes and that the NEA protocols could consider supporting both models =
if necessary; but Model 3 appears to be a superset of Model 2 and as=20

  such so it might make sense to simply the protocol and only support =
model 3. My preference is to always make things simpler if possible.



  Gary Tomlinson




-------------------------------------------------------------------------=
-----


  _______________________________________________
  Nea mailing list
  Nea@ietf.org
  https://www1.ietf.org/mailman/listinfo/nea

------=_NextPart_000_020E_01C7F3BA.36E6A3E0
Content-Type: text/html;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD><TITLE>RE: Newcomer comments on =
draft-ietf-nea-requirements-04</TITLE>
<META http-equiv=3DContent-Type content=3D"text/html; =
charset=3Diso-8859-1">
<META content=3D"MSHTML 6.00.6000.16525" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3D"Times New Roman" size=3D2>Parvez - in spirit I =
understand what=20
you are trying to do but I disagree and will enumerate why =
below...</FONT></DIV>
<DIV>&nbsp;</DIV>
<DIV>---- Original Message ----- </DIV>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <DIV=20
  style=3D"BACKGROUND: #e4e4e4; FONT: 10pt arial; font-color: =
black"><B>From:</B>=20
  <A title=3Dpanandam@SonicWALL.com=20
  href=3D"mailto:panandam@SonicWALL.com">panandam@SonicWALL.com</A> =
</DIV>
  <DIV style=3D"FONT: 10pt arial"><B>To:</B> <A title=3Dnea@ietf.org=20
  href=3D"mailto:nea@ietf.org">nea@ietf.org</A> </DIV>
  <DIV style=3D"FONT: 10pt arial"><B>Sent:</B> Monday, September 10, =
2007 1:14=20
  PM</DIV>
  <DIV style=3D"FONT: 10pt arial"><B>Subject:</B> [Nea] RE: Newcomer =
comments on=20
  draft-ietf-nea-requirements-04</DIV>
  <DIV><BR></DIV><!-- Converted from text/rtf format -->
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us><FONT face=3DArial size=3D2>I have a</FONT></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us> =
<FONT face=3DArial=20
  size=3D2>few</FONT></SPAN><SPAN lang=3Den-us></SPAN><SPAN =
lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us><FONT face=3DArial size=3D2> additional comments, =
following=20
  Gary=92s,</FONT></SPAN><SPAN lang=3Den-us></SPAN><SPAN =
lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us> <FONT face=3DArial size=3D2>that also</FONT></SPAN><SPAN =

  lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us> =
<FONT face=3DArial=20
  size=3D2>pertain</FONT></SPAN><SPAN lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us><FONT face=3DArial =
size=3D2></FONT></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us> =
<FONT face=3DArial=20
  size=3D2>to draft-ietf-nea-requirements-04.</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Page 35 =
Section=20
  7.4</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DCalibri>Please consider =
adding a new=20
  requirement for the PT protocol to support non-persistent network =
connections.=20
  </FONT></SPAN></P></BLOCKQUOTE>
<P><SPAN lang=3Den-us><FONT face=3DCalibri>This will not be a viable =
thing to do in=20
the real world no matter how nice it sounds I fear...</FONT></SPAN></P>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DCalibri>This is =
especially important=20
  for mobile devices that roam or have air time charges. &nbsp;This has =
an=20
  impact on reassessments triggered by the NEA server; in some cases =
(e.g. NAT=20
  traversal), the only option may be for the NEA client to periodically =
contact=20
  the NEA server to determine whether a reassessment is=20
  necessary.</FONT></SPAN></P></BLOCKQUOTE>
<P><SPAN lang=3Den-us><FONT face=3DCalibri>I disagree...&nbsp; =
Supporting Mobile=20
Device (Nomadic Computing) and its Integrity for "devices that are based =
on=20
non-persistant connections" may not be such a good idea since it will=20
dramatically increase the overhead in any NEA system at a minimum and =
also will=20
create a HUGE level of pain for people running mixed networks...=20
</FONT></SPAN><SPAN lang=3Den-us><FONT face=3DCalibri>For instance it =
would be=20
rather easy to program a PDA to slam a NEA server with continuous =
requests for=20
service after service.</FONT></SPAN></P>
<P><SPAN lang=3Den-us><FONT face=3DCalibri>To deal with these types of=20
devices,&nbsp; the NEA system will also have to maintain state tables =
for all of=20
its local "recent" connections since under any reasonable policy =
requiring=20
continuous certification of competence or NEA policy compliance means =
repeating=20
the same set of messages again and again... </FONT></SPAN></P>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DCalibri>Page 44 Section =
9.1 Paragraph=20
  3</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DCalibri>As Gary stated, =
due to security=20
  policy, network owners may wish not to disclose the attributes being=20
  requested. </FONT></SPAN></P></BLOCKQUOTE>
<P><SPAN lang=3Den-us><FONT face=3DCalibri>Then why are they allowing =
people to plug=20
into their networks at all without disclosing those=20
policies...</FONT></SPAN></P>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DCalibri>Imagine an =
attacker who=20
  restricts all posture collection in order to determine what the =
network=20
  operators are looking for.</FONT></SPAN></P></BLOCKQUOTE>
<P><SPAN lang=3Den-us><FONT face=3DCalibri>But how would that system =
work in a NEA=20
environment... it would be getting routing access to the switches right? =
So what=20
is it that system would be doing?</FONT></SPAN></P>
<P><SPAN lang=3Den-us><FONT face=3DCalibri>Remember the key purpose of =
NEA is to=20
BOTH allow one to force compliance with pre-existing NETWORK POLICIES in =
a new=20
or recently powered-up system AND&nbsp;the continuously recertifying =
those=20
systems which passed previously if they still continue to pass their =
tests.=20
</FONT></SPAN></P>
<P><SPAN lang=3Den-us><FONT face=3DCalibri>To do that there needs to be =
some way of=20
communicating what those requirements are to the client in an =
interactive mode.=20
</FONT></SPAN></P>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3D"Times New Roman"=20
  size=3D2></FONT></SPAN><SPAN lang=3Den-us></SPAN><SPAN =
lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us></SPAN>&nbsp;</P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DCalibri>Page 45 Section =
9.1 Paragraph=20
  6</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DCalibri>This is a =
continuation of the=20
  thought in the comment above and concerns this item in =
the</FONT></SPAN><SPAN=20
  lang=3Den-us> <FONT face=3DCalibri>NEA</FONT></SPAN><SPAN =
lang=3Den-us> <FONT=20
  face=3DCalibri>disclosure statement: =93What information is =
required=94. Such=20
  disclosure</FONT></SPAN><SPAN lang=3Den-us></SPAN><SPAN =
lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us><FONT face=3DArial size=3D2> must be subject to&nbsp; the =
security=20
  policy of the network owner. What is the intent of specificity in this =

  guidance?&nbsp; Does the network owner have the =
latitude</FONT></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us><FONT =
face=3DArial=20
  size=3D2> of</FONT></SPAN><SPAN lang=3Den-us></SPAN><SPAN =
lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us><FONT face=3DArial size=3D2> provid</FONT></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us><FONT =
face=3DArial=20
  size=3D2>ing</FONT></SPAN><SPAN lang=3Den-us></SPAN><SPAN =
lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us><FONT face=3DArial size=3D2> a more general disclosure=20
  statement?</FONT></SPAN><SPAN lang=3Den-us></SPAN><SPAN =
lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Parvez=20
  Anandam</FONT></SPAN><SPAN lang=3Den-us></SPAN><SPAN =
lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us><FONT face=3DTahoma=20
  =
size=3D2>_____________________________________________<BR></FONT></SPAN><=
SPAN=20
  lang=3Den-us><B></B></SPAN><SPAN lang=3Den-us><B></B></SPAN><SPAN=20
  lang=3Den-us><B><FONT face=3DTahoma =
size=3D2>From:</FONT></B></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us><FONT =
face=3DTahoma=20
  size=3D2> Gary Tomlinson<BR></FONT></SPAN><SPAN =
lang=3Den-us><B></B></SPAN><SPAN=20
  lang=3Den-us><B></B></SPAN><SPAN lang=3Den-us><B><FONT face=3DTahoma=20
  size=3D2>Sent:</FONT></B></SPAN><SPAN lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us><FONT face=3DTahoma size=3D2> =
Monday, September=20
  10, 2007 1:11 PM<BR></FONT></SPAN><SPAN =
lang=3Den-us><B></B></SPAN><SPAN=20
  lang=3Den-us><B></B></SPAN><SPAN lang=3Den-us><B><FONT face=3DTahoma=20
  size=3D2>To:</FONT></B></SPAN><SPAN lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us><FONT face=3DTahoma size=3D2>=20
  nea@ietf.org<BR></FONT></SPAN><SPAN lang=3Den-us><B></B></SPAN><SPAN=20
  lang=3Den-us><B></B></SPAN><SPAN lang=3Den-us><B><FONT face=3DTahoma=20
  size=3D2>Subject:</FONT></B></SPAN><SPAN lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us><FONT face=3DTahoma size=3D2> =
Newcomer comments=20
  on draft-ietf-nea-requirements-04</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us><FONT face=3DArial size=3D2>Hello NEA =
WG,</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Several of =
us at SonicWALL=20
  have recently been following the NEA WG in general and specifically=20
  draft-ietf-nea-requirements-03 and </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial=20
  size=3D2>draft-ietf-nea-requirements-04.&nbsp; I have the following =
comments=20
  pertaining to draft-ietf-nea-requirements-04.</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Page 32 =
Section 7.1,=20
  Requirement C-9</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us><FONT face=3D"Times New Roman" size=3D2>At a minimum, C-9 =
should=20
  explicitly require UTF-8 encodings be supported in strings in order to =
adhere=20
  to the guidance specified in RFC 2277.&nbsp;&nbsp; </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3D"Times New Roman" =
size=3D2>Is there any=20
  reason to support character sets other than UTF-8?&nbsp; Supporting =
only UTF-8=20
  makes the protocol straightforward on interpreting string =
</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3D"Times New Roman" =
size=3D2>encodings and=20
  normalization to locale specific character sets is generally easy to =
implement=20
  on most OS platforms using their intrinsic =
libraries.</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Page 33, =
Section 7.2,=20
  Requirement PA-1</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>The NEA =
model seems to=20
  imply an extensible standard attribute name space, but PA-1 does not=20
  incorporate this as a requirement. Nor</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>does PA-1 =
specify a=20
  requirement for the standard attribute name space schema to be =
uniquely=20
  identifiable from vendor attribute name </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>spaces. =
</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Page 33, =
Section 7.2,=20
  Requirement PA-2</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us><FONT face=3DArial color=3D#000000 size=3D2>The NEA model =
implies a=20
  plurality of extensible vendor attribute name spaces and concurrent =
instances=20
  of each schema, but </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>PA-2 does=20
  not&nbsp; incorporate this as a requirement.&nbsp; Nor does PA-2 =
specify a=20
  requirement for vendor attribute name space schemas to =
</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>be uniquely=20
  identifiable.</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Page 33, =
Section 7.2,=20
  Requirement PA-4</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us><FONT face=3DArial color=3D#000000 size=3D2>The network =
owner security=20
  policy may restrict giving the user complete visibility into why an =
assessment=20
  failed in order to prevent </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>exposing=20
  unnecessary information to potential attackers, especially prior to =
user=20
  authentication.&nbsp; With this in mind, please consider =
</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>changing=20
  "This enables" to "This optionally enables the Posture Client, subject =
to=20
  disclosure policy of the NEA Server, =94.</FONT></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =
lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Page 33, =
Section 7.12=20
  Requirement PA-5</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us><FONT face=3DArial color=3D#000000 size=3D2>I am little =
confused over the=20
  recent comments debating downgrading [MUST] to [SHOULD] or =
[MAY].&nbsp; The=20
  decision to utilize end-to-end </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>cryptography=20
  is use case topology dependent.&nbsp; If these use cases are deemed to =
be in=20
  scope, then doesn't it make sense for this to be a </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>[MUST]=20
  requirement and amend the verbiage to read "=85 MUST be optionally =
capable of=20
  providing =85"?</FONT></SPAN><SPAN lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN></P><BR>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Page 34 =
Section=20
  7.2</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us><FONT face=3D"Times New Roman" size=3D2>I noticed the =
former PA-8=20
  requirement in draft-ietf-nea-requirements-03 was dropped; apparently =
in favor=20
  of the more generalized new C-9 </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3D"Times New Roman" =
size=3D2>requirement in=20
  draft-ietf-nea-requirements-04.&nbsp; I believe there still exists a =
need for=20
  a PA specific I18N string encoding requirement for =
non-</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3D"Times New Roman" =
size=3D2>displayable=20
  request and response attributes. Request attributes for end point =
artifacts=20
  such as File, Directory, Process Name, etc are used in =
</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3D"Times New Roman" =
size=3D2>common=20
  practice today in NEA equivalent deployments I am aware of.&nbsp; =
Again, as=20
  mentioned in the comment regarding the C-9 requirement, =
</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3D"Times New Roman" =
size=3D2>UTF-8 makes=20
  things unambiguous and at a minimum must be a supported=20
  encoding.</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Page 34, =
Section 7.3=20
  Requirement PB-3</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us><FONT face=3DArial color=3D#000000 size=3D2>Is there any =
reason in PB-3=20
  not to specify identifiers be based upon SMI enterprise numbers?&nbsp; =
This=20
  requirement looks much like that </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>already=20
  faced by SNMP and LDAP; and we already have the IANA machinery to =
leverage=20
  here.</FONT></SPAN><SPAN lang=3Den-us></SPAN><SPAN =
lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Page 35, =
Section 7.4=20
  Requirement PT-2</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us><FONT face=3DArial color=3D#000000 size=3D2>In addition =
to mutual=20
  authentication specified in PT-2, the PT protocol must also be capable =
of=20
  supporting server side only authentication </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>as an=20
  alternative to mutual authentication.&nbsp; Server side only =
authentication is=20
  common practice today in SSL VPNs. With this in mind, please=20
</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>consider=20
  changing =93mutual authentication=94 to =93mutual and server side only =

  authentication schemes=94.</FONT></SPAN><SPAN =
lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Page 44, =
Section 9.1=20
  Paragraph 2</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us><FONT face=3DArial color=3D#000000 size=3D2>In order to =
prevent privacy=20
  abuse by network owners, PA request attributes must be restricted to=20
  information&nbsp; known apriori&nbsp; rather </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>than be open=20
  ended in order to prevent dredging for artifact information. For =
example a PV=20
  looking for a domain named </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000=20
  size=3D2>=93mydomain.com=94 should be allowed, whereas requesting a =
domain named =93*=94=20
  should be disallowed by the PC.</FONT></SPAN><SPAN =
lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Page 46, =
Section 9.2&nbsp;=20
  Paragraph 2</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us><FONT face=3DArial color=3D#000000 size=3D2>I don't think =
Model 1 is=20
  practical or even possible in some cases.&nbsp; This model works only =
when=20
  both the PC and PV have an immutable </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>set of=20
  attributes known apriori by each.&nbsp; However, it is common practice =
in=20
  existing SSL VPN s to request information on the existence of files,=20
  </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>directories,=20
  processes and other artifacts that are configurable assessment policy =
on the=20
  NEA Server. In order for model 1 to potentially work, =
</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>the NEA=20
  client would have to send enormous amounts of information; potentially =
an=20
  enumeration of its entire file name space, process name space,=20
  </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>registry=20
  name space, etc.</FONT></SPAN><SPAN lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us></SPAN><SPAN lang=3Den-us> </SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial size=3D2>Page 46, =
Section 9.2=20
  Paragraph 4</FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us><FONT face=3DArial color=3D#000000 size=3D2>Model 3 is =
used in common=20
  practice today by many of the NEA like deployments.&nbsp; At a =
minimum, it=20
  must be supported .&nbsp; I can also see model 2 </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>being used=20
  effectively by a subset of PC/PV providers.&nbsp; It occurs to me, the =

  decision to use either model 2 or model 3 is one the PC/PV provider=20
  </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>makes and=20
  that the NEA protocols could consider supporting both models if =
necessary; but=20
  Model 3 appears to be a superset of Model 2 and as </FONT></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>such so it=20
  might make sense to simply the protocol and only support model 3. My=20
  preference is to always make things simpler if =
possible.</FONT></SPAN></P><BR>
  <P dir=3Dltr><SPAN lang=3Den-us><FONT face=3DArial color=3D#000000 =
size=3D2>Gary=20
  Tomlinson</FONT></SPAN><SPAN lang=3Den-us></SPAN><SPAN =
lang=3Den-us></SPAN><SPAN=20
  lang=3Den-us></SPAN></P>
  <P dir=3Dltr><SPAN lang=3Den-us></SPAN><SPAN lang=3Den-us></SPAN><SPAN =

  lang=3Den-us></SPAN></P>
  <P>
  <HR>

  <P></P>_______________________________________________<BR>Nea mailing=20
  =
list<BR>Nea@ietf.org<BR>https://www1.ietf.org/mailman/listinfo/nea<BR></B=
LOCKQUOTE></BODY></HTML>

------=_NextPart_000_020E_01C7F3BA.36E6A3E0--




--===============0587559593==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea

--===============0587559593==--






From nea-bounces@ietf.org Mon Sep 10 18:08:27 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IUrQh-0008Kw-5c; Mon, 10 Sep 2007 18:08:27 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IUrQg-0008KA-0Z
	for nea-confirm+ok@megatron.ietf.org; Mon, 10 Sep 2007 18:08:26 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IUrQf-0008K2-Ka
	for nea@ietf.org; Mon, 10 Sep 2007 18:08:25 -0400
Received: from elasmtp-scoter.atl.sa.earthlink.net ([209.86.89.67])
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IUrQd-0007Gg-Ly
	for nea@ietf.org; Mon, 10 Sep 2007 18:08:25 -0400
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=earthlink.net;
	b=GDYGXDRpWYkWr7ZOTICQpdsG6HJYUdUtZcnz4KHknfJhKvqK5+gc3PpstdwDZoU0;
	h=Received:Message-ID:From:To:References:Subject:Date:MIME-Version:Content-Type:X-Priority:X-MSMail-Priority:X-Mailer:X-MimeOLE:X-ELNK-Trace:X-Originating-IP;
Received: from [64.125.79.23] (helo=gw)
	by elasmtp-scoter.atl.sa.earthlink.net with asmtp (Exim 4.34)
	id 1IUrQX-0006ig-UP; Mon, 10 Sep 2007 18:08:18 -0400
Message-ID: <022e01c7f3f7$24434ad0$174f7d40@home.glassey.com>
From: "todd glassey" <tglassey@earthlink.net>
To: <gtomlinson@SonicWALL.com>,
	<nea@ietf.org>
References: <4DF8F0A275770B4387764386ABEF4D7A1008A8@us0exb10.us.sonicwall.com>
Subject: Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Date: Mon, 10 Sep 2007 15:08:19 -0700
MIME-Version: 1.0
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2900.3138
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.3138
X-ELNK-Trace: 01b7a7e171bdf5911aa676d7e74259b7b3291a7d08dfec79c9ba97e693f9658285e011f028c4b824350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c
X-Originating-IP: 64.125.79.23
X-Spam-Score: 0.0 (/)
X-Scan-Signature: d2bb8c7db26f2d12109e0cd8e454db52
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Content-Type: multipart/mixed; boundary="===============1231200559=="
Errors-To: nea-bounces@ietf.org

This is a multi-part message in MIME format.

--===============1231200559==
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_0223_01C7F3BC.6BECC9F0"

This is a multi-part message in MIME format.

------=_NextPart_000_0223_01C7F3BC.6BECC9F0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Newcomer comments on draft-ietf-nea-requirements-04Gary
  ----- Original Message -----=20
  From: gtomlinson@SonicWALL.com=20
  To: nea@ietf.org=20
  Sent: Monday, September 10, 2007 1:10 PM
  Subject: [Nea] Newcomer comments on draft-ietf-nea-requirements-04


  Hello NEA WG,=20

  Several of us at SonicWALL have recently been following the NEA WG in =
general and specifically draft-ietf-nea-requirements-03 and=20

  draft-ietf-nea-requirements-04.  I have the following comments =
pertaining to draft-ietf-nea-requirements-04.=20

  Page 32 Section 7.1, Requirement C-9=20
  At a minimum, C-9 should explicitly require UTF-8 encodings be =
supported in strings in order to adhere to the guidance specified in RFC =
2277.  =20

You mean its its important to keep the compliance with ISO UCS...  I =
agree - the minimum standard is UCS.



  Is there any reason to support character sets other than UTF-8?  =
Supporting only UTF-8 makes the protocol straightforward on interpreting =
string=20

  encodings and normalization to locale specific character sets is =
generally easy to implement on most OS platforms using their intrinsic =
libraries.

  Page 33, Section 7.2, Requirement PA-1=20
  The NEA model seems to imply an extensible standard attribute name =
space, but PA-1 does not incorporate this as a requirement. Nor does =
PA-1 specify a requirement for the standard attribute name space schema =
to be uniquely identifiable from vendor attribute name spaces.=20

This is specific to a use model - so document that use model  and its =
requirement inside of PA-1.

  Page 33, Section 7.2, Requirement PA-2=20
  The NEA model implies a plurality of extensible vendor attribute name =
spaces and concurrent instances of each schema, but PA-2 does not  =
incorporate this as a requirement.  Nor does PA-2 specify a requirement =
for vendor attribute name space schemas to be uniquely identifiable.=20

Again - a complete set of use models would eliminate this.

  Page 33, Section 7.2, Requirement PA-4=20
  The network owner security policy may restrict giving the user =
complete visibility into why an assessment failed in order to prevent =
exposing unnecessary information to potential attackers, especially =
prior to user authentication. =20

Here then is a specific use in which the system provides hierarchical =
error or failure messages with varying detail?

  With this in mind, please consider changing "This enables" to "This =
optionally enables the Posture Client, subject to disclosure policy of =
the NEA Server, ".

  Page 33, Section 7.12 Requirement PA-5=20
  I am little confused over the recent comments debating downgrading =
[MUST] to [SHOULD] or [MAY].  The decision to utilize end-to-end =
cryptography is use case topology dependent.  If these use cases are =
deemed to be in scope, then doesn't it make sense for this to be a =
[MUST] requirement and amend the verbiage to read ". MUST be optionally =
capable of providing ."?=20

Again this is use-model specific.

  Page 34 Section 7.2=20
  I noticed the former PA-8 requirement in =
draft-ietf-nea-requirements-03 was dropped; apparently in favor of the =
more generalized new C-9 requirement in draft-ietf-nea-requirements-04.  =
I believe there still exists a need for a PA specific I18N string =
encoding requirement for non-displayable request and response =
attributes. Request attributes for end point artifacts such as File, =
Directory, Process Name, etc are used in common practice today in NEA =
equivalent deployments I am aware of.  Again, as mentioned in the =
comment regarding the C-9 requirement, UTF-8 makes things unambiguous =
and at a minimum must be a supported encoding.=20

  Page 34, Section 7.3 Requirement PB-3=20
  Is there any reason in PB-3 not to specify identifiers be based upon =
SMI enterprise numbers?  This requirement looks much like that already =
faced by SNMP and LDAP; and we already have the IANA machinery to =
leverage here.=20

  Page 35, Section 7.4 Requirement PT-2=20
  In addition to mutual authentication specified in PT-2, the PT =
protocol must also be capable of supporting server side only =
authentication as an alternative to mutual authentication.  Server side =
only authentication is common practice today in SSL VPNs. With this in =
mind, please consider changing "mutual authentication" to "mutual and =
server side only authentication schemes".=20

I suggest the following change:=20

Page 35, Section 7.4 Requirement PT-2=20
"Authentictation schemes specific to the end-use models", so that the =
type of authentication - whether Server-Side, mutual, or circuit-level =
only can be done.



  Page 44, Section 9.1 Paragraph 2=20
  In order to prevent privacy abuse by network owners, PA request =
attributes must be restricted to information  known apriori  rather than =
be open ended in order to prevent dredging for artifact information. For =
example a PV looking for a domain named "mydomain.com" should be =
allowed, whereas requesting a domain named "*" should be disallowed by =
the PC.=20

Why? - What you suggest here means that each and ever domain MUST be =
individually enumerated rather than being able to say "Any Supported or =
Defined Domain"... I can see many instances - especially those inside of =
University's and the like which may have hundreds of domains, all which =
are operated under the University's Security Persona...

  Page 46, Section 9.2  Paragraph 2=20
  I don't think Model 1 is practical or even possible in some cases.  =
This model works only when both the PC and PV have an immutable set of =
attributes known apriori by each.  However, it is common practice in =
existing SSL VPN s to request information on the existence of files, =
directories, processes and other artifacts that are configurable =
assessment policy on the NEA Server. In order for model 1 to potentially =
work, the NEA client would have to send enormous amounts of information; =
potentially an enumeration of its entire file name space, process name =
space, registry name space, etc.=20

yes it would - but what is the value of the assurance it was to provide =
if the Client side didnt present these qualification datapoints?

  Page 46, Section 9.2 Paragraph 4=20
  Model 3 is used in common practice today by many of the NEA like =
deployments.  At a minimum, it must be supported .  I can also see model =
2 being used effectively by a subset of PC/PV providers.  It occurs to =
me, the decision to use either model 2 or model 3 is one the PC/PV =
provider makes and that the NEA protocols could consider supporting both =
models if necessary; but Model 3 appears to be a superset of Model 2 and =
as such so it might make sense to simply the protocol and only support =
model 3. My preference is to always make things simpler if possible.

I agree.



  Gary Tomlinson=20





-------------------------------------------------------------------------=
-----


  _______________________________________________
  Nea mailing list
  Nea@ietf.org
  https://www1.ietf.org/mailman/listinfo/nea

------=_NextPart_000_0223_01C7F3BC.6BECC9F0
Content-Type: text/html;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD><TITLE>Newcomer comments on =
draft-ietf-nea-requirements-04</TITLE>
<META http-equiv=3DContent-Type content=3D"text/html; =
charset=3Diso-8859-1">
<META content=3D"MSHTML 6.00.6000.16525" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3D"Times New Roman" size=3D2>Gary</FONT></DIV>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <DIV style=3D"FONT: 10pt arial">----- Original Message ----- </DIV>
  <DIV=20
  style=3D"BACKGROUND: #e4e4e4; FONT: 10pt arial; font-color: =
black"><B>From:</B>=20
  <A title=3Dgtomlinson@SonicWALL.com=20
  href=3D"mailto:gtomlinson@SonicWALL.com">gtomlinson@SonicWALL.com</A> =
</DIV>
  <DIV style=3D"FONT: 10pt arial"><B>To:</B> <A title=3Dnea@ietf.org=20
  href=3D"mailto:nea@ietf.org">nea@ietf.org</A> </DIV>
  <DIV style=3D"FONT: 10pt arial"><B>Sent:</B> Monday, September 10, =
2007 1:10=20
  PM</DIV>
  <DIV style=3D"FONT: 10pt arial"><B>Subject:</B> [Nea] Newcomer =
comments on=20
  draft-ietf-nea-requirements-04</DIV>
  <DIV><BR></DIV><!-- Converted from text/rtf format -->
  <P><FONT face=3DArial size=3D2>Hello NEA WG,</FONT> </P>
  <P><FONT face=3DArial size=3D2>Several of us at SonicWALL have =
recently been=20
  following the NEA WG in general and specifically=20
  draft-ietf-nea-requirements-03 and </FONT></P>
  <P><FONT face=3DArial size=3D2>draft-ietf-nea-requirements-04.&nbsp; I =
have the=20
  following comments pertaining to =
draft-ietf-nea-requirements-04.</FONT> </P>
  <P><FONT face=3DArial size=3D2>Page 32 Section 7.1, Requirement =
C-9</FONT>=20
  <BR><FONT face=3D"Times New Roman" size=3D2>At a minimum, C-9 should =
explicitly=20
  require UTF-8 encodings be supported in strings in order to adhere to =
the=20
  guidance specified in RFC 2277.&nbsp;&nbsp; </FONT></P></BLOCKQUOTE>
<P><FONT face=3D"Times New Roman" size=3D2>You mean its its important to =
keep the=20
compliance with ISO UCS...&nbsp; I agree - the minimum standard is=20
UCS.</FONT></P>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <P><FONT face=3D"Times New Roman" size=3D2></FONT>&nbsp;</P>
  <P><FONT face=3D"Times New Roman" size=3D2>Is there any reason to =
support=20
  character sets other than UTF-8?&nbsp; Supporting only UTF-8 makes the =

  protocol straightforward on interpreting string </FONT></P>
  <P><FONT face=3D"Times New Roman" size=3D2>encodings and normalization =
to locale=20
  specific character sets is generally easy to implement on most OS =
platforms=20
  using their intrinsic libraries.</FONT></P>
  <P><FONT face=3DArial size=3D2>Page 33, Section 7.2, Requirement =
PA-1</FONT>=20
  <BR><FONT face=3DArial size=3D2>The NEA model seems to imply an =
extensible=20
  standard attribute name space, but PA-1 does not incorporate this as a =

  requirement. Nor </FONT><FONT face=3DArial size=3D2>does PA-1 specify =
a=20
  requirement for the standard attribute name space schema to be =
uniquely=20
  identifiable from vendor attribute name </FONT><FONT face=3DArial =
size=3D2>spaces.=20
  </FONT></P></BLOCKQUOTE>
<P><FONT size=3D2>This is specific to a use model - so document that use =

model&nbsp; and its requirement inside of PA-1.</FONT></P>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <P><FONT face=3DArial size=3D2>Page 33, Section 7.2, Requirement =
PA-2</FONT>=20
  <BR><FONT face=3DArial color=3D#000000 size=3D2>The NEA model implies =
a plurality of=20
  extensible vendor attribute name spaces and concurrent instances of =
each=20
  schema, but </FONT><FONT face=3DArial color=3D#000000 size=3D2>PA-2 =
does not&nbsp;=20
  incorporate this as a requirement.&nbsp; Nor does PA-2 specify a =
requirement=20
  for vendor attribute name space schemas to </FONT><FONT face=3DArial=20
  color=3D#000000 size=3D2>be uniquely identifiable.</FONT> =
</P></BLOCKQUOTE>
<P><FONT face=3D"Times New Roman" size=3D2>Again -&nbsp;a complete set =
of use models=20
would eliminate this.</FONT></P>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <P><FONT face=3DArial size=3D2>Page 33, Section 7.2, Requirement =
PA-4</FONT>=20
  <BR><FONT face=3DArial color=3D#000000 size=3D2>The network owner =
security policy=20
  may restrict giving the user complete visibility into why an =
assessment failed=20
  in order to prevent </FONT><FONT face=3DArial color=3D#000000 =
size=3D2>exposing=20
  unnecessary information to potential attackers, especially prior to =
user=20
  authentication.&nbsp; </FONT></P></BLOCKQUOTE>
<P><FONT face=3DArial color=3D#000000 size=3D2>Here then is a specific =
use in which=20
the system provides hierarchical error or failure messages with varying=20
detail?</FONT></P>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <P><FONT face=3DArial color=3D#000000 size=3D2>With this in mind, =
please consider=20
  </FONT><FONT face=3DArial color=3D#000000 size=3D2>changing "This =
enables" to "This=20
  optionally enables the Posture Client, subject to disclosure policy of =
the NEA=20
  Server, =94.</FONT></P>
  <P><FONT face=3DArial size=3D2>Page 33, Section 7.12 Requirement =
PA-5</FONT>=20
  <BR><FONT face=3DArial color=3D#000000 size=3D2>I am little confused =
over the recent=20
  comments debating downgrading [MUST] to [SHOULD] or [MAY].&nbsp; The =
decision=20
  to utilize end-to-end </FONT><FONT face=3DArial color=3D#000000=20
  size=3D2>cryptography is use case topology dependent.&nbsp; If these =
use cases=20
  are deemed to be in scope, then doesn't it make sense for this to be a =

  </FONT><FONT face=3DArial color=3D#000000 size=3D2>[MUST] requirement =
and amend the=20
  verbiage to read "=85 MUST be optionally capable of providing =
=85"?</FONT>=20
</P></BLOCKQUOTE>
<DIV><FONT face=3D"Times New Roman" size=3D2>Again this is use-model=20
specific.</FONT><BR></DIV>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <P><FONT face=3DArial size=3D2>Page 34 Section 7.2</FONT> <BR><FONT=20
  face=3D"Times New Roman" size=3D2>I noticed the former PA-8 =
requirement in=20
  draft-ietf-nea-requirements-03 was dropped; apparently in favor of the =
more=20
  generalized new C-9 </FONT><FONT face=3D"Times New Roman" =
size=3D2>requirement in=20
  draft-ietf-nea-requirements-04.&nbsp; I believe there still exists a =
need for=20
  a PA specific I18N string encoding requirement for non-</FONT><FONT=20
  face=3D"Times New Roman" size=3D2>displayable request and response =
attributes.=20
  Request attributes for end point artifacts such as File, Directory, =
Process=20
  Name, etc are used in </FONT><FONT face=3D"Times New Roman" =
size=3D2>common=20
  practice today in NEA equivalent deployments I am aware of.&nbsp; =
Again, as=20
  mentioned in the comment regarding the C-9 requirement, </FONT><FONT=20
  face=3D"Times New Roman" size=3D2>UTF-8 makes things unambiguous and =
at a minimum=20
  must be a supported encoding.</FONT> </P>
  <P><FONT face=3DArial size=3D2>Page 34, Section 7.3 Requirement =
PB-3</FONT>=20
  <BR><FONT face=3DArial color=3D#000000 size=3D2>Is there any reason in =
PB-3 not to=20
  specify identifiers be based upon SMI enterprise numbers?&nbsp; This=20
  requirement looks much like that </FONT><FONT face=3DArial =
color=3D#000000=20
  size=3D2>already faced by SNMP and LDAP; and we already have the IANA =
machinery=20
  to leverage here.</FONT> </P>
  <P><FONT face=3DArial size=3D2>Page 35, Section 7.4 Requirement =
PT-2</FONT>=20
  <BR><FONT face=3DArial color=3D#000000 size=3D2>In addition to mutual =
authentication=20
  specified in PT-2, the PT protocol must also be capable of supporting =
server=20
  side only authentication </FONT><FONT face=3DArial color=3D#000000 =
size=3D2>as an=20
  alternative to mutual authentication.&nbsp; Server side only =
authentication is=20
  common practice today in SSL VPNs. With this in mind, please =
</FONT><FONT=20
  face=3DArial color=3D#000000 size=3D2>consider changing =93mutual =
authentication=94 to=20
  =93mutual and server side only authentication schemes=94.</FONT> =
</P></BLOCKQUOTE>
<P><FONT size=3D2>I suggest the following change: </FONT></P>
<P><FONT face=3DArial size=3D2>Page 35, Section 7.4 Requirement =
PT-2</FONT>=20
<BR><FONT face=3D"Times New Roman" size=3D2>"Authentictation schemes =
specific to the=20
end-use models", so that the type of authentication - whether =
Server-Side,=20
mutual, or circuit-level only can be done.</FONT></P>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <P><FONT size=3D2></FONT>&nbsp;</P>
  <P><FONT face=3DArial size=3D2>Page 44, Section 9.1 Paragraph 2</FONT> =
<BR><FONT=20
  face=3DArial color=3D#000000 size=3D2>In order to prevent privacy =
abuse by network=20
  owners, PA request attributes must be restricted to information&nbsp; =
known=20
  apriori&nbsp; rather </FONT><FONT face=3DArial color=3D#000000 =
size=3D2>than be open=20
  ended in order to prevent dredging for artifact information. For =
example a PV=20
  looking for a domain named </FONT><FONT face=3DArial color=3D#000000=20
  size=3D2>=93mydomain.com=94 should be allowed, whereas requesting a =
domain named =93*=94=20
  should be disallowed by the PC.</FONT> </P></BLOCKQUOTE>
<P><FONT size=3D2>Why? - What you suggest here means that each and ever =
domain=20
MUST be individually enumerated rather than being able to say "Any =
Supported or=20
Defined Domain"... I can see many instances - especially those inside of =

University's and the like which may have hundreds of domains, all which =
are=20
operated under the University's Security Persona...</FONT></P>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <P><FONT face=3DArial size=3D2>Page 46, Section 9.2&nbsp; Paragraph =
2</FONT>=20
  <BR><FONT face=3DArial color=3D#000000 size=3D2>I don't think Model 1 =
is practical=20
  or even possible in some cases.&nbsp; This model works only when both =
the PC=20
  and PV have an immutable </FONT><FONT face=3DArial color=3D#000000 =
size=3D2>set of=20
  attributes known apriori by each.&nbsp; However, it is common practice =
in=20
  existing SSL VPN s to request information on the existence of files,=20
  </FONT><FONT face=3DArial color=3D#000000 size=3D2>directories, =
processes and other=20
  artifacts that are configurable assessment policy on the NEA Server. =
In order=20
  for model 1 to potentially work, </FONT><FONT face=3DArial =
color=3D#000000=20
  size=3D2>the NEA client would have to send enormous amounts of =
information;=20
  potentially an enumeration of its entire file name space, process name =
space,=20
  </FONT><FONT face=3DArial color=3D#000000 size=3D2>registry name =
space, etc.</FONT>=20
  </P></BLOCKQUOTE>
<P><FONT face=3D"Times New Roman" size=3D2>yes it would - but what is =
the value of=20
the assurance it was to provide if the Client side didnt present these=20
qualification datapoints?</FONT></P>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <P><FONT face=3DArial size=3D2>Page 46, Section 9.2 Paragraph 4</FONT> =
<BR><FONT=20
  face=3DArial color=3D#000000 size=3D2>Model 3 is used in common =
practice today by=20
  many of the NEA like deployments.&nbsp; At a minimum, it must be =
supported=20
  .&nbsp; I can also see model 2 </FONT><FONT face=3DArial =
color=3D#000000=20
  size=3D2>being used effectively by a subset of PC/PV providers.&nbsp; =
It occurs=20
  to me, the decision to use either model 2 or model 3 is one the PC/PV =
provider=20
  </FONT><FONT face=3DArial color=3D#000000 size=3D2>makes and that the =
NEA protocols=20
  could consider supporting both models if necessary; but Model 3 =
appears to be=20
  a superset of Model 2 and as </FONT><FONT face=3DArial color=3D#000000 =
size=3D2>such=20
  so it might make sense to simply the protocol and only support model =
3. My=20
  preference is to always make things simpler if =
possible.</FONT></P></BLOCKQUOTE>
<P><FONT size=3D2>I agree.</FONT></P>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px"><BR>
  <P><FONT face=3DArial color=3D#000000 size=3D2>Gary Tomlinson</FONT> =
</P><BR>
  <P>
  <HR>

  <P></P>_______________________________________________<BR>Nea mailing=20
  =
list<BR>Nea@ietf.org<BR>https://www1.ietf.org/mailman/listinfo/nea<BR></B=
LOCKQUOTE></BODY></HTML>

------=_NextPart_000_0223_01C7F3BC.6BECC9F0--




--===============1231200559==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea

--===============1231200559==--






From nea-bounces@ietf.org Mon Sep 10 19:31:18 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IUsiq-0003Oc-Hh; Mon, 10 Sep 2007 19:31:16 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IUsio-0003Nx-AL
	for nea-confirm+ok@megatron.ietf.org; Mon, 10 Sep 2007 19:31:14 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IUsio-0003Np-0J
	for nea@ietf.org; Mon, 10 Sep 2007 19:31:14 -0400
Received: from sj-iport-5.cisco.com ([171.68.10.87])
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IUsil-0008Nz-4k
	for nea@ietf.org; Mon, 10 Sep 2007 19:31:13 -0400
X-IronPort-AV: E=Sophos;i="4.20,234,1186383600"; 
	d="scan'208,217";a="175316864"
Received: from sj-dkim-2.cisco.com ([171.71.179.186])
	by sj-iport-5.cisco.com with ESMTP; 10 Sep 2007 16:31:10 -0700
Received: from sj-core-5.cisco.com (sj-core-5.cisco.com [171.71.177.238])
	by sj-dkim-2.cisco.com (8.12.11/8.12.11) with ESMTP id l8ANVAnL005283; 
	Mon, 10 Sep 2007 16:31:10 -0700
Received: from xbh-sjc-231.amer.cisco.com (xbh-sjc-231.cisco.com
	[128.107.191.100])
	by sj-core-5.cisco.com (8.12.10/8.12.6) with ESMTP id l8ANUgop004474;
	Mon, 10 Sep 2007 23:31:10 GMT
Received: from xmb-sjc-22d.amer.cisco.com ([128.107.191.68]) by
	xbh-sjc-231.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.1830); 
	Mon, 10 Sep 2007 16:31:08 -0700
Received: from 171.69.75.24 ([171.69.75.24]) by xmb-sjc-22d.amer.cisco.com
	([128.107.191.68]) with Microsoft Exchange Server HTTP-DAV ; 
	Mon, 10 Sep 2007 23:31:07 +0000
User-Agent: Microsoft-Entourage/11.3.6.070618
Date: Mon, 10 Sep 2007 16:31:08 -0700
Subject: Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
From: Kaushik Narayan <kaushik@cisco.com>
To: <gtomlinson@SonicWALL.com>, <nea@ietf.org>
Message-ID: <C30B21CC.AAD7%kaushik@cisco.com>
Thread-Topic: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acfz5qfsPz1SyEhoRu2ib5KW8a/VZAAHAHud
In-Reply-To: <4DF8F0A275770B4387764386ABEF4D7A1008A8@us0exb10.us.sonicwall.com>
Mime-version: 1.0
X-OriginalArrivalTime: 10 Sep 2007 23:31:08.0157 (UTC)
	FILETIME=[A9F07AD0:01C7F402]
DKIM-Signature: v=0.5; a=rsa-sha256; q=dns/txt; l=22956; t=1189467070;
	x=1190331070; c=relaxed/simple; s=sjdkim2002;
	h=Content-Type:From:Subject:Content-Transfer-Encoding:MIME-Version;
	d=cisco.com; i=kaushik@cisco.com;
	z=From:=20Kaushik=20Narayan=20<kaushik@cisco.com>
	|Subject:=20Re=3A=20[Nea]=20Newcomer=20comments=20on=20draft-ietf-nea-req
	uirements-04 |Sender:=20;
	bh=EoKzWp2n1klb0UlLQzYgHMjWFwAQKEcsyYJrm1YOLtQ=;
	b=k8CRyN59aJQbG1DN2f3/O5xg0Qcp6OaXpCgCg3W0SaipvtuygUjXPRPvrNuba5EpuRbU2sWx
	btBG4MNExvbkEDyXWpqgQuf5xLH/NoY40OOP5IH/sMBQSDpvAmU4vnUi;
Authentication-Results: sj-dkim-2; header.From=kaushik@cisco.com; dkim=pass (
	sig from cisco.com/sjdkim2002 verified; ); 
X-Spam-Score: -1.4 (-)
X-Scan-Signature: 3bef755d9f4ba0b7ac474570b36ffadb
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Content-Type: multipart/mixed; boundary="===============0152029570=="
Errors-To: nea-bounces@ietf.org

> This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

--===============0152029570==
Content-type: multipart/alternative;
	boundary="B_3272286668_3987761"

> This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

--B_3272286668_3987761
Content-type: text/plain;
	charset="ISO-8859-1"
Content-transfer-encoding: quoted-printable

Hi Gary,

Thanks for your comments. Please find my response inline.


On 9/10/07 1:10 PM, "gtomlinson@SonicWALL.com" <gtomlinson@SonicWALL.com>
wrote:

> Hello NEA WG,=20
>=20
> Several of us at SonicWALL have recently been following the NEA WG in gen=
eral
> and specifically draft-ietf-nea-requirements-03 and
>=20
> draft-ietf-nea-requirements-04.  I have the following comments pertaining=
 to
> draft-ietf-nea-requirements-04.
>=20
> Page 32 Section 7.1, Requirement C-9
> At a minimum, C-9 should explicitly require UTF-8 encodings be supported =
in
> strings in order to adhere to the guidance specified in RFC 2277.
>=20
> Is there any reason to support character sets other than UTF-8?  Supporti=
ng
> only UTF-8 makes the protocol straightforward on interpreting string
>=20
> encodings and normalization to locale specific character sets is generall=
y
> easy to implement on most OS platforms using their intrinsic libraries.
>=20
>=20
> <Kaushik>=20
>=20
> C-9 currently tries to capture i18n requirements for attributes that are =
both
> =B3text=B2 and =B3non-text=B2at the PA and PB layer. We did not want to get into
> specific encoding standards
> and that=B9s the reason for a very generically worded requirement
>=20
> </Kaushik>=20
>=20
>=20
>=20
> Page 33, Section 7.2, Requirement PA-1
> The NEA model seems to imply an extensible standard attribute name space,=
 but
> PA-1 does not incorporate this as a requirement. Nor
>=20
> does PA-1 specify a requirement for the standard attribute name space sch=
ema
> to be uniquely identifiable from vendor attribute name
>=20
> spaces.=20
>=20
>=20
> Page 33, Section 7.2, Requirement PA-2
> The NEA model implies a plurality of extensible vendor attribute name spa=
ces
> and concurrent instances of each schema, but
>=20
> PA-2 does not  incorporate this as a requirement.  Nor does PA-2 specify =
a
> requirement for vendor attribute name space schemas to
>=20
> be uniquely identifiable.
>=20
>=20
> <Kaushik>
>=20
> You are right. The notion of a standard and vendor specific namespaces is
> implicit in the requirements. We will edit the requirements to make the  =
need
> for separate attribute namespaces more explicit.
>=20
> </Kaushik>
>=20
>=20
>=20
> Page 33, Section 7.2, Requirement PA-4
> The network owner security policy may restrict giving the user complete
> visibility into why an assessment failed in order to prevent
>=20
> exposing unnecessary information to potential attackers, especially prior=
 to
> user authentication.  With this in mind, please consider
>=20
> changing "This enables" to "This optionally enables the Posture Client,
> subject to disclosure policy of the NEA Server, =B2.
>=20
>=20
> <Kaushik> Agreed. </Kaushik>
>=20
>=20
> Page 33, Section 7.12 Requirement PA-5
> I am little confused over the recent comments debating downgrading [MUST]=
 to
> [SHOULD] or [MAY].  The decision to utilize end-to-end
>=20
> cryptography is use case topology dependent.  If these use cases are deem=
ed to
> be in scope, then doesn't it make sense for this to be a
>=20
> [MUST] requirement and amend the verbiage to read "=8A MUST be optionally
> capable of providing =8A"?
>=20
>=20
>=20
>=20
> Page 34 Section 7.2
> I noticed the former PA-8 requirement in draft-ietf-nea-requirements-03 w=
as
> dropped; apparently in favor of the more generalized new C-9
>=20
> requirement in draft-ietf-nea-requirements-04.  I believe there still exi=
sts a
> need for a PA specific I18N string encoding requirement for non-
>=20
> displayable request and response attributes. Request attributes for end p=
oint
> artifacts such as File, Directory, Process Name, etc are used in
>=20
> common practice today in NEA equivalent deployments I am aware of.  Again=
, as
> mentioned in the comment regarding the C-9 requirement,
>=20
> UTF-8 makes things unambiguous and at a minimum must be a supported encod=
ing.
>=20
> <Kaushik> C-9 should capture this since there could be non-displayable
> attributes at the PB layer as well. </Kaushik>
>=20
>=20
>=20
>=20
> Page 34, Section 7.3 Requirement PB-3
> Is there any reason in PB-3 not to specify identifiers be based upon SMI
> enterprise numbers?  This requirement looks much like that
>=20
> already faced by SNMP and LDAP; and we already have the IANA machinery to
> leverage here.=20
>=20
>=20
> <Kaushik>=20
>=20
> Are you referring to the use of SMI for  static identifiers for collector=
s &
> validators ? I think it might be a useful suggestion
> for a potential solution, not sure if need to talk about this in the
> requirements.=20
>=20
> </Kaushik>
>=20
>=20
>=20
>=20
>=20
> Page 35, Section 7.4 Requirement PT-2
> In addition to mutual authentication specified in PT-2, the PT protocol m=
ust
> also be capable of supporting server side only authentication
>=20
> as an alternative to mutual authentication.  Server side only authenticat=
ion
> is common practice today in SSL VPNs. With this in mind, please
>=20
> consider changing =B3mutual authentication=B2 to =B3mutual and server side only
> authentication schemes=B2.
>=20
>=20
> <Kaushik>
>=20
>=20
> The requirement is a MUST for mutual authentication between NEA clients &=
 NEA
> servers. The mutual authentication could be achieved as a result
> of two independent authentication transactions (with binding for certain =
PTs),
> for e.g. For password based credentials, the NEA server will be authentic=
ated
> while a secure channel is being established (such as TLS) and NEA client =
will
> be subsequently authenticated within that secure channel.
>=20
>=20
> </Kaushik>=20
>=20
>=20
>=20
>=20
>=20
>=20
> Page 44, Section 9.1 Paragraph 2
> In order to prevent privacy abuse by network owners, PA request attribute=
s
> must be restricted to information  known apriori  rather
>=20
> than be open ended in order to prevent dredging for artifact information.=
 For
> example a PV looking for a domain named
> =B3mydomain.com=B2 should be allowed, whereas requesting a domain named =B3*=B2 s=
hould
> be disallowed by the PC.
>=20
> <Kaushik>
>=20
> I guess you are stating a more pointed use of the mechanism stated in
> paragraph 2. It might be a useful to include in that section.
>=20
> </Kaushik>=20
>=20
>=20
>=20
> Page 46, Section 9.2  Paragraph 2
> I don't think Model 1 is practical or even possible in some cases.  This =
model
> works only when both the PC and PV have an immutable
>=20
> set of attributes known apriori by each.  However, it is common practice =
in
> existing SSL VPN s to request information on the existence of files,
>=20
> directories, processes and other artifacts that are configurable assessme=
nt
> policy on the NEA Server. In order for model 1 to potentially work,
>=20
> the NEA client would have to send enormous amounts of information; potent=
ially
> an enumeration of its entire file name space, process name space,
>=20
> registry name space, etc.
>=20
> <Kaushik>
>=20
> That=B9s a fair point but Model 1 might be a useful model to assess endpoin=
ts
> that are not owned by the Enterprise, for e.g.  A contractor/guest system
> that is configured to provide a pre-set profile of attributes when challe=
nged
> by a guest/partner network.
>=20
> </Kaushik>
> =20
>=20
>=20
> Page 46, Section 9.2 Paragraph 4
> Model 3 is used in common practice today by many of the NEA like deployme=
nts.
> At a minimum, it must be supported .  I can also see model 2
>=20
> being used effectively by a subset of PC/PV providers.  It occurs to me, =
the
> decision to use either model 2 or model 3 is one the PC/PV provider
>=20
> makes and that the NEA protocols could consider supporting both models if
> necessary; but Model 3 appears to be a superset of Model 2 and as
>=20
> such so it might make sense to simply the protocol and only support model=
 3.
> My preference is to always make things simpler if possible.
>=20
>=20
>=20
> <Kaushik>=20
>=20
> Just want to point out that all the models described in section 9 are
> implementation considerations and not requirements for NEA protocol
> specification. The models explain how NEA implementations can address pri=
vacy
> concerns and it is upto to implementations to support one or more of thes=
e
> models.
>=20
> regards,
>   kaushik!
>=20
> </Kaushik>
>=20
>=20
>=20
>=20
> Gary Tomlinson=20
>=20
>=20
>=20
> _______________________________________________
> Nea mailing list
> Nea@ietf.org
> https://www1.ietf.org/mailman/listinfo/nea



--B_3272286668_3987761
Content-type: text/html;
	charset="ISO-8859-1"
Content-transfer-encoding: quoted-printable

<HTML>
<HEAD>
<TITLE>Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04</TITLE=
>
</HEAD>
<BODY>
<FONT FACE=3D"Verdana, Helvetica, Arial"><SPAN STYLE=3D'font-size:12.0px'>Hi Ga=
ry,<BR>
<BR>
Thanks for your comments. Please find my response inline.<BR>
<BR>
<BR>
On 9/10/07 1:10 PM, &quot;gtomlinson@SonicWALL.com&quot; &lt;gtomlinson@Son=
icWALL.com&gt; wrote:<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><SPAN STYLE=3D'font-size:12.0px'><FONT FACE=3D"Arial"=
>Hello NEA WG,</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"> <BR>
<BR>
</FONT><FONT FACE=3D"Arial">Several of us at SonicWALL have recently been fol=
lowing the NEA WG in general and specifically draft-ietf-nea-requirements-03=
 and <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">draft-ietf-nea-requirements-04. &nbsp;I have the =
following comments pertaining to draft-ietf-nea-requirements-04.</FONT><FONT=
 FACE=3D"Verdana, Helvetica, Arial"> <BR>
<BR>
</FONT><FONT FACE=3D"Arial">Page 32 Section 7.1, Requirement C-9</FONT><FONT =
FACE=3D"Verdana, Helvetica, Arial"> <BR>
</FONT><FONT FACE=3D"Times New Roman">At a minimum, C-9 should explicitly req=
uire UTF-8 encodings be supported in strings in order to adhere to the guida=
nce specified in RFC 2277. &nbsp;&nbsp;<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">Is there any reason to support characte=
r sets other than UTF-8? &nbsp;Supporting only UTF-8 makes the protocol stra=
ightforward on interpreting string <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">encodings and normalization to locale s=
pecific character sets is generally easy to implement on most OS platforms u=
sing their intrinsic libraries.<BR>
<BR>
<BR>
&lt;Kaushik&gt; <BR>
<BR>
C-9 currently tries to capture i18n requirements for attributes that are bo=
th &#8220;text&#8221; and &#8220;non-text&#8221;at the PA and PB layer. We d=
id not want to get into specific encoding standards<BR>
and that&#8217;s the reason for a very generically worded requirement<BR>
<BR>
&lt;/Kaushik&gt; <BR>
<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
<BR>
</FONT><FONT FACE=3D"Arial">Page 33, Section 7.2, Requirement PA-1</FONT><FON=
T FACE=3D"Verdana, Helvetica, Arial"> <BR>
</FONT><FONT FACE=3D"Arial">The NEA model seems to imply an extensible standa=
rd attribute name space, but PA-1 does not incorporate this as a requirement=
. Nor<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">does PA-1 specify a requirement for the standard =
attribute name space schema to be uniquely identifiable from vendor attribut=
e name <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">spaces. <BR>
<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Page 33, Section 7.2, Requirement PA-2</FONT><FON=
T FACE=3D"Verdana, Helvetica, Arial"> <BR>
</FONT><FONT FACE=3D"Arial">The NEA model implies a plurality of extensible v=
endor attribute name spaces and concurrent instances of each schema, but <BR=
>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">PA-2 does not &nbsp;incorporate this as a require=
ment. &nbsp;Nor does PA-2 specify a requirement for vendor attribute name sp=
ace schemas to <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">be uniquely identifiable.</FONT><FONT FACE=3D"Verda=
na, Helvetica, Arial"> <BR>
<BR>
</FONT><FONT FACE=3D"Arial"><BR>
&lt;Kaushik&gt;<BR>
<BR>
You are right. The notion of a standard and vendor specific namespaces is i=
mplicit in the requirements. We will edit the requirements to make the &nbsp=
;need for separate attribute namespaces more explicit.<BR>
<BR>
&lt;/Kaushik&gt;<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
<BR>
<BR>
</FONT><FONT FACE=3D"Arial">Page 33, Section 7.2, Requirement PA-4</FONT><FON=
T FACE=3D"Verdana, Helvetica, Arial"> <BR>
</FONT><FONT FACE=3D"Arial">The network owner security policy may restrict gi=
ving the user complete visibility into why an assessment failed in order to =
prevent <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">exposing unnecessary information to potential att=
ackers, especially prior to user authentication. &nbsp;With this in mind, pl=
ease consider <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">changing &quot;This enables&quot; to &quot;This o=
ptionally enables the Posture Client, subject to disclosure policy of the NE=
A Server, &#8221;.<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
<BR>
&lt;Kaushik&gt; Agreed. &lt;/Kaushik&gt;<BR>
<BR>
<BR>
</FONT><FONT FACE=3D"Arial">Page 33, Section 7.12 Requirement PA-5</FONT><FON=
T FACE=3D"Verdana, Helvetica, Arial"> <BR>
</FONT><FONT FACE=3D"Arial">I am little confused over the recent comments deb=
ating downgrading [MUST] to [SHOULD] or [MAY]. &nbsp;The decision to utilize=
 end-to-end <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">cryptography is use case topology dependent. &nbs=
p;If these use cases are deemed to be in scope, then doesn't it make sense f=
or this to be a <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">[MUST] requirement and amend the verbiage to read=
 &quot;&#8230; MUST be optionally capable of providing &#8230;&quot;?</FONT>=
<FONT FACE=3D"Verdana, Helvetica, Arial"> <BR>
<BR>
<BR>
<BR>
<BR>
</FONT><FONT FACE=3D"Arial">Page 34 Section 7.2</FONT><FONT FACE=3D"Verdana, He=
lvetica, Arial"> <BR>
</FONT><FONT FACE=3D"Times New Roman">I noticed the former PA-8 requirement i=
n draft-ietf-nea-requirements-03 was dropped; apparently in favor of the mor=
e generalized new C-9 <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">requirement in draft-ietf-nea-requireme=
nts-04. &nbsp;I believe there still exists a need for a PA specific I18N str=
ing encoding requirement for non-<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">displayable request and response attrib=
utes. Request attributes for end point artifacts such as File, Directory, Pr=
ocess Name, etc are used in <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">common practice today in NEA equivalent=
 deployments I am aware of. &nbsp;Again, as mentioned in the comment regardi=
ng the C-9 requirement, <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">UTF-8 makes things unambiguous and at a=
 minimum must be a supported encoding.</FONT><FONT FACE=3D"Verdana, Helvetica,=
 Arial"> <BR>
<BR>
&lt;Kaushik&gt; C-9 should capture this since there could be non-displayabl=
e attributes at the PB layer as well. &lt;/Kaushik&gt;<BR>
<BR>
<BR>
<BR>
<BR>
</FONT><FONT FACE=3D"Arial">Page 34, Section 7.3 Requirement PB-3</FONT><FONT=
 FACE=3D"Verdana, Helvetica, Arial"> <BR>
</FONT><FONT FACE=3D"Arial">Is there any reason in PB-3 not to specify identi=
fiers be based upon SMI enterprise numbers? &nbsp;This requirement looks muc=
h like that <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">already faced by SNMP and LDAP; and we already ha=
ve the IANA machinery to leverage here.</FONT><FONT FACE=3D"Verdana, Helvetica=
, Arial"> <BR>
<BR>
<BR>
&lt;Kaushik&gt; <BR>
<BR>
Are you referring to the use of SMI for &nbsp;static identifiers for collec=
tors &amp; validators ? I think it might be a useful suggestion<BR>
for a potential solution, not sure if need to talk about this in the requir=
ements. <BR>
<BR>
&lt;/Kaushik&gt;<BR>
<BR>
<BR>
<BR>
<BR>
<BR>
</FONT><FONT FACE=3D"Arial">Page 35, Section 7.4 Requirement PT-2</FONT><FONT=
 FACE=3D"Verdana, Helvetica, Arial"> <BR>
</FONT><FONT FACE=3D"Arial">In addition to mutual authentication specified in=
 PT-2, the PT protocol must also be capable of supporting server side only a=
uthentication <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">as an alternative to mutual authentication. &nbsp=
;Server side only authentication is common practice today in SSL VPNs. With =
this in mind, please <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">consider changing &#8220;mutual authentication&#8=
221; to &#8220;mutual and server side only authentication schemes&#8221;.</F=
ONT><FONT FACE=3D"Verdana, Helvetica, Arial"> <BR>
<BR>
<BR>
&lt;Kaushik&gt;<BR>
<BR>
<BR>
The requirement is a MUST for mutual authentication between NEA clients &am=
p; NEA servers. The mutual authentication could be achieved as a result<BR>
of two independent authentication transactions (with binding for certain PT=
s), for e.g. For password based credentials, the NEA server will be authenti=
cated <BR>
while a secure channel is being established (such as TLS) and NEA client wi=
ll be subsequently authenticated within that secure channel.<BR>
<BR>
<BR>
&lt;/Kaushik&gt; <BR>
<BR>
<BR>
<BR>
<BR>
<BR>
<BR>
</FONT><FONT FACE=3D"Arial">Page 44, Section 9.1 Paragraph 2</FONT><FONT FACE=
=3D"Verdana, Helvetica, Arial"> <BR>
</FONT><FONT FACE=3D"Arial">In order to prevent privacy abuse by network owne=
rs, PA request attributes must be restricted to information &nbsp;known apri=
ori &nbsp;rather <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">than be open ended in order to prevent dredging f=
or artifact information. For example a PV looking for a domain named <BR>
&#8220;mydomain.com&#8221; should be allowed, whereas requesting a domain n=
amed &#8220;*&#8221; should be disallowed by the PC.</FONT><FONT FACE=3D"Verda=
na, Helvetica, Arial"> <BR>
<BR>
&lt;Kaushik&gt;<BR>
<BR>
I guess you are stating a more pointed use of the mechanism stated in parag=
raph 2. It might be a useful to include in that section.<BR>
<BR>
&lt;/Kaushik&gt; <BR>
<BR>
<BR>
<BR>
</FONT><FONT FACE=3D"Arial">Page 46, Section 9.2 &nbsp;Paragraph 2</FONT><FON=
T FACE=3D"Verdana, Helvetica, Arial"> <BR>
</FONT><FONT FACE=3D"Arial">I don't think Model 1 is practical or even possib=
le in some cases. &nbsp;This model works only when both the PC and PV have a=
n immutable <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">set of attributes known apriori by each. &nbsp;Ho=
wever, it is common practice in existing SSL VPN s to request information on=
 the existence of files, <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">directories, processes and other artifacts that a=
re configurable assessment policy on the NEA Server. In order for model 1 to=
 potentially work, <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">the NEA client would have to send enormous amount=
s of information; potentially an enumeration of its entire file name space, =
process name space, <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">registry name space, etc.</FONT><FONT FACE=3D"Verda=
na, Helvetica, Arial"> <BR>
<BR>
&lt;Kaushik&gt;<BR>
<BR>
That&#8217;s a fair point but Model 1 might be a useful model to assess end=
points that are not owned by the Enterprise, for e.g. &nbsp;A contractor/gue=
st system<BR>
that is configured to provide a pre-set profile of attributes when challeng=
ed by a guest/partner network.<BR>
<BR>
&lt;/Kaushik&gt;<BR>
&nbsp;<BR>
<BR>
<BR>
</FONT><FONT FACE=3D"Arial">Page 46, Section 9.2 Paragraph 4</FONT><FONT FACE=
=3D"Verdana, Helvetica, Arial"> <BR>
</FONT><FONT FACE=3D"Arial">Model 3 is used in common practice today by many =
of the NEA like deployments. &nbsp;At a minimum, it must be supported . &nbs=
p;I can also see model 2 <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">being used effectively by a subset of PC/PV provi=
ders. &nbsp;It occurs to me, the decision to use either model 2 or model 3 i=
s one the PC/PV provider <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">makes and that the NEA protocols could consider s=
upporting both models if necessary; but Model 3 appears to be a superset of =
Model 2 and as <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">such so it might make sense to simply the protoco=
l and only support model 3. My preference is to always make things simpler i=
f possible.<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
<BR>
<BR>
&lt;Kaushik&gt; <BR>
<BR>
Just want to point out that all the models described in section 9 are imple=
mentation considerations and not requirements for NEA protocol specification=
. The models explain how NEA implementations can address privacy concerns an=
d it is upto to implementations to support one or more of these models.<BR>
<BR>
regards,<BR>
&nbsp;&nbsp;kaushik!<BR>
<BR>
&lt;/Kaushik&gt;<BR>
<BR>
<BR>
<BR>
<BR>
</FONT><FONT FACE=3D"Arial">Gary Tomlinson</FONT><FONT FACE=3D"Verdana, Helveti=
ca, Arial"> <BR>
<BR>
<BR>
<HR ALIGN=3DCENTER SIZE=3D"3" WIDTH=3D"95%"></FONT><FONT FACE=3D"Monaco, Courier Ne=
w">_______________________________________________<BR>
Nea mailing list<BR>
Nea@ietf.org<BR>
https://www1.ietf.org/mailman/listinfo/nea<BR>
</FONT></SPAN></BLOCKQUOTE><SPAN STYLE=3D'font-size:12.0px'><FONT FACE=3D"Monac=
o, Courier New"><BR>
</FONT></SPAN>
</BODY>
</HTML>


--B_3272286668_3987761--



--===============0152029570==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea

--===============0152029570==--





From nea-bounces@ietf.org Mon Sep 10 19:39:44 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IUsr1-0003Tz-Q0; Mon, 10 Sep 2007 19:39:43 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IUsr1-0003Pd-23
	for nea-confirm+ok@megatron.ietf.org; Mon, 10 Sep 2007 19:39:43 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IUsr0-0003Nc-Mf
	for nea@ietf.org; Mon, 10 Sep 2007 19:39:42 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IUsqy-000067-MC
	for nea@ietf.org; Mon, 10 Sep 2007 19:39:42 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id hsnaeu0171sr for <nea@ietf.org>;
	Mon, 10 Sep 2007 16:36:14 -0700 (envelope-from
	<panandam@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9143)
	with ESMTP; Mon, 10 Sep 2007 16:36:13 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Mon, 10 Sep 2007 16:39:41 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Subject: RE: [Nea] RE: Newcomer comments on draft-ietf-nea-requirements-04
Date: Mon, 10 Sep 2007 16:39:33 -0700
Message-ID: <D37A04F05A87044A8DCEEC0BF9A3AB1D07D03A@us0exb10.us.sonicwall.com>
In-Reply-To: <021d01c7f3f4$eed1dfd0$174f7d40@home.glassey.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] RE: Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acfz9Owfz+Phi6Z6QQO9tYilp4Da4wAAP68w
References: <4DF8F0A275770B4387764386ABEF4D7A1008A8@us0exb10.us.sonicwall.com>
	<D37A04F05A87044A8DCEEC0BF9A3AB1D07D024@us0exb10.us.sonicwall.com>
	<021d01c7f3f4$eed1dfd0$174f7d40@home.glassey.com>
From: <panandam@SonicWALL.com>
To: <nea@ietf.org>
X-OriginalArrivalTime: 10 Sep 2007 23:39:41.0630 (UTC)
	FILETIME=[DBFE3DE0:01C7F403]
X-Mlf-Version: 6.0.1.9143
X-Mlf-UniqueId: o200709102336130075352
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 96ace3b0d55c752b49ad5cebce8c2b2a
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Content-Type: multipart/mixed; boundary="===============0761562918=="
Errors-To: nea-bounces@ietf.org

This is a multi-part message in MIME format.

--===============0761562918==
Content-class: urn:content-classes:message
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C7F403.D9AAF09C"

This is a multi-part message in MIME format.

------_=_NextPart_001_01C7F403.D9AAF09C
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

Todd,

=20

Thanks for the reply.

=20

[PA] Comments below...

=20

From: todd glassey [mailto:tglassey@earthlink.net]=20
Sent: Monday, September 10, 2007 2:53 PM
To: Parvez Anandam; nea@ietf.org
Subject: Re: [Nea] RE: Newcomer comments on
draft-ietf-nea-requirements-04

=20

Parvez - in spirit I understand what you are trying to do but I disagree
and will enumerate why below...

=20

---- Original Message -----=20

	From: panandam@SonicWALL.com=20

	To: nea@ietf.org=20

	Sent: Monday, September 10, 2007 1:14 PM

	Subject: [Nea] RE: Newcomer comments on
draft-ietf-nea-requirements-04

	=20

	I have a few additional comments, following Gary's, that also
pertain to draft-ietf-nea-requirements-04.

	Page 35 Section 7.4

	Please consider adding a new requirement for the PT protocol to
support non-persistent network connections.=20

This will not be a viable thing to do in the real world no matter how
nice it sounds I fear...

[PA] We have empirical evidence that this is possible in the real world,
with our End Point Control solution. A classic example is a smartphone
on a cellular data network, using an SSL VPN.

	This is especially important for mobile devices that roam or
have air time charges.  This has an impact on reassessments triggered by
the NEA server; in some cases (e.g. NAT traversal), the only option may
be for the NEA client to periodically contact the NEA server to
determine whether a reassessment is necessary.

I disagree...  Supporting Mobile Device (Nomadic Computing) and its
Integrity for "devices that are based on non-persistant connections" may
not be such a good idea since it will dramatically increase the overhead
in any NEA system at a minimum and also will create a HUGE level of pain
for people running mixed networks... For instance it would be rather
easy to program a PDA to slam a NEA server with continuous requests for
service after service.

To deal with these types of devices,  the NEA system will also have to
maintain state tables for all of its local "recent" connections since
under any reasonable policy requiring continuous certification of
competence or NEA policy compliance means repeating the same set of
messages again and again...=20

[PA] Is the scope of NEA restricted to non-mobile devices, then? I hope
not. Maintaining state is common practice, for VPNs . I'm not sure there
is an alternative, if one is to support devices such as smartphones.
Such devices are increasingly being used to run SSL VPN clients; they
are part of a WAN,  not a LAN. The denial of service considerations are
important, I agree.

	Page 44 Section 9.1 Paragraph 3

	As Gary stated, due to security policy, network owners may wish
not to disclose the attributes being requested.=20

Then why are they allowing people to plug into their networks at all
without disclosing those policies...

[PA] The users may be connecting to a corporate network from home, for
example. The attributes the network owners are looking for will, of
course, be disclosed to the Posture Collectors, but that does not mean
that a (human) user should necessarily be aware of what is being
collected. This prevents a user from configuring an unauthorized laptop
to simulate an authorized one (see below). Of course, the privacy issues
increase in scope significantly once the user is out of the loop; thus
Gary's comment about preventing open-ended dredging of attributes.

	Imagine an attacker who restricts all posture collection in
order to determine what the network operators are looking for.

But how would that system work in a NEA environment... it would be
getting routing access to the switches right? So what is it that system
would be doing?

Remember the key purpose of NEA is to BOTH allow one to force compliance
with pre-existing NETWORK POLICIES in a new or recently powered-up
system AND the continuously recertifying those systems which passed
previously if they still continue to pass their tests.=20

[PA] The attack I envision goes something like this. The attacker plugs
in her laptop into a conference room jack at a competitor's. She obtains
an IP address from the central VPN gateway that everyone in the building
uses, but this does not allow her to go beyond the VPN gateway. Her
laptop is requested to perform NEA checks. As the NEA requirements are
currently written, she will be informed of all the attributes that are
being requested, say the presence of anti-virus program XYZ and the
presence of a specific Windows registry key. Knowing this critical
information, she runs program XYZ from her large attack arsenal and adds
the registry entry. Her laptop now conforms to the NEA security policies
of the VPN gateway and is indistinguishable from a corporate laptop.

To do that there needs to be some way of communicating what those
requirements are to the client in an interactive mode.=20

[PA] Certainly, there needs to be some secure way of communicating those
requirements from the Posture Validators to the Posture Collectors; the
PA, PB and PT protocols take care of that. The end-user need not know
the contents of that communication, however, for the reason I described
above.

[PA] Thanks,

[PA] Parvez Anandam

	=20

	Page 45 Section 9.1 Paragraph 6

	This is a continuation of the thought in the comment above and
concerns this item in the NEA disclosure statement: "What information is
required". Such disclosure must be subject to  the security policy of
the network owner. What is the intent of specificity in this guidance?
Does the network owner have the latitude of providing a more general
disclosure statement?

	Parvez Anandam

	_____________________________________________
	From: Gary Tomlinson
	Sent: Monday, September 10, 2007 1:11 PM
	To: nea@ietf.org
	Subject: Newcomer comments on draft-ietf-nea-requirements-04

	Hello NEA WG,

	Several of us at SonicWALL have recently been following the NEA
WG in general and specifically draft-ietf-nea-requirements-03 and=20

	draft-ietf-nea-requirements-04.  I have the following comments
pertaining to draft-ietf-nea-requirements-04.

	Page 32 Section 7.1, Requirement C-9

	At a minimum, C-9 should explicitly require UTF-8 encodings be
supported in strings in order to adhere to the guidance specified in RFC
2277.  =20

	Is there any reason to support character sets other than UTF-8?
Supporting only UTF-8 makes the protocol straightforward on interpreting
string=20

	encodings and normalization to locale specific character sets is
generally easy to implement on most OS platforms using their intrinsic
libraries.

	Page 33, Section 7.2, Requirement PA-1

	The NEA model seems to imply an extensible standard attribute
name space, but PA-1 does not incorporate this as a requirement. Nor

	does PA-1 specify a requirement for the standard attribute name
space schema to be uniquely identifiable from vendor attribute name=20

	spaces.=20

	Page 33, Section 7.2, Requirement PA-2

	The NEA model implies a plurality of extensible vendor attribute
name spaces and concurrent instances of each schema, but=20

	PA-2 does not  incorporate this as a requirement.  Nor does PA-2
specify a requirement for vendor attribute name space schemas to=20

	be uniquely identifiable.

	Page 33, Section 7.2, Requirement PA-4

	The network owner security policy may restrict giving the user
complete visibility into why an assessment failed in order to prevent=20

	exposing unnecessary information to potential attackers,
especially prior to user authentication.  With this in mind, please
consider=20

	changing "This enables" to "This optionally enables the Posture
Client, subject to disclosure policy of the NEA Server, ".

	Page 33, Section 7.12 Requirement PA-5

	I am little confused over the recent comments debating
downgrading [MUST] to [SHOULD] or [MAY].  The decision to utilize
end-to-end=20

	cryptography is use case topology dependent.  If these use cases
are deemed to be in scope, then doesn't it make sense for this to be a=20

	[MUST] requirement and amend the verbiage to read "... MUST be
optionally capable of providing ..."?

	=20

	Page 34 Section 7.2

	I noticed the former PA-8 requirement in
draft-ietf-nea-requirements-03 was dropped; apparently in favor of the
more generalized new C-9=20

	requirement in draft-ietf-nea-requirements-04.  I believe there
still exists a need for a PA specific I18N string encoding requirement
for non-

	displayable request and response attributes. Request attributes
for end point artifacts such as File, Directory, Process Name, etc are
used in=20

	common practice today in NEA equivalent deployments I am aware
of.  Again, as mentioned in the comment regarding the C-9 requirement,=20

	UTF-8 makes things unambiguous and at a minimum must be a
supported encoding.

	Page 34, Section 7.3 Requirement PB-3

	Is there any reason in PB-3 not to specify identifiers be based
upon SMI enterprise numbers?  This requirement looks much like that=20

	already faced by SNMP and LDAP; and we already have the IANA
machinery to leverage here.

	Page 35, Section 7.4 Requirement PT-2

	In addition to mutual authentication specified in PT-2, the PT
protocol must also be capable of supporting server side only
authentication=20

	as an alternative to mutual authentication.  Server side only
authentication is common practice today in SSL VPNs. With this in mind,
please=20

	consider changing "mutual authentication" to "mutual and server
side only authentication schemes".

	Page 44, Section 9.1 Paragraph 2

	In order to prevent privacy abuse by network owners, PA request
attributes must be restricted to information  known apriori  rather=20

	than be open ended in order to prevent dredging for artifact
information. For example a PV looking for a domain named=20

	"mydomain.com" should be allowed, whereas requesting a domain
named "*" should be disallowed by the PC.

	Page 46, Section 9.2  Paragraph 2

	I don't think Model 1 is practical or even possible in some
cases.  This model works only when both the PC and PV have an immutable=20

	set of attributes known apriori by each.  However, it is common
practice in existing SSL VPN s to request information on the existence
of files,=20

	directories, processes and other artifacts that are configurable
assessment policy on the NEA Server. In order for model 1 to potentially
work,=20

	the NEA client would have to send enormous amounts of
information; potentially an enumeration of its entire file name space,
process name space,=20

	registry name space, etc.=20

	Page 46, Section 9.2 Paragraph 4

	Model 3 is used in common practice today by many of the NEA like
deployments.  At a minimum, it must be supported .  I can also see model
2=20

	being used effectively by a subset of PC/PV providers.  It
occurs to me, the decision to use either model 2 or model 3 is one the
PC/PV provider=20

	makes and that the NEA protocols could consider supporting both
models if necessary; but Model 3 appears to be a superset of Model 2 and
as=20

	such so it might make sense to simply the protocol and only
support model 3. My preference is to always make things simpler if
possible.

	=20

	Gary Tomlinson

________________________________

	_______________________________________________
	Nea mailing list
	Nea@ietf.org
	https://www1.ietf.org/mailman/listinfo/nea


------_=_NextPart_001_01C7F403.D9AAF09C
Content-Type: text/html;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

<html xmlns:v=3D"urn:schemas-microsoft-com:vml" =
xmlns:o=3D"urn:schemas-microsoft-com:office:office" =
xmlns:w=3D"urn:schemas-microsoft-com:office:word" =
xmlns:m=3D"http://schemas.microsoft.com/office/2004/12/omml" =
xmlns=3D"http://www.w3.org/TR/REC-html40">

<head>
<meta http-equiv=3DContent-Type content=3D"text/html; =
charset=3Dus-ascii">
<meta name=3DGenerator content=3D"Microsoft Word 12 (filtered medium)">
<!--[if !mso]>
<style>
v\:* {behavior:url(#default#VML);}
o\:* {behavior:url(#default#VML);}
w\:* {behavior:url(#default#VML);}
.shape {behavior:url(#default#VML);}
</style>
<![endif]-->
<title>RE: Newcomer comments on draft-ietf-nea-requirements-04</title>
<style>
<!--
 /* Font Definitions */
 @font-face
	{font-family:"Cambria Math";
	panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
	{font-family:Calibri;
	panose-1:2 15 5 2 2 2 4 3 2 4;}
@font-face
	{font-family:Tahoma;
	panose-1:2 11 6 4 3 5 4 4 2 4;}
 /* Style Definitions */
 p.MsoNormal, li.MsoNormal, div.MsoNormal
	{margin:0in;
	margin-bottom:.0001pt;
	font-size:12.0pt;
	font-family:"Times New Roman","serif";}
a:link, span.MsoHyperlink
	{mso-style-priority:99;
	color:blue;
	text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
	{mso-style-priority:99;
	color:purple;
	text-decoration:underline;}
p
	{mso-style-priority:99;
	mso-margin-top-alt:auto;
	margin-right:0in;
	mso-margin-bottom-alt:auto;
	margin-left:0in;
	font-size:12.0pt;
	font-family:"Times New Roman","serif";}
span.EmailStyle18
	{mso-style-type:personal-reply;
	font-family:"Calibri","sans-serif";
	color:#1F497D;}
.MsoChpDefault
	{mso-style-type:export-only;
	font-size:10.0pt;}
@page Section1
	{size:8.5in 11.0in;
	margin:1.0in 1.0in 1.0in 1.0in;}
div.Section1
	{page:Section1;}
-->
</style>
<!--[if gte mso 9]><xml>
 <o:shapedefaults v:ext=3D"edit" spidmax=3D"1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
 <o:shapelayout v:ext=3D"edit">
  <o:idmap v:ext=3D"edit" data=3D"1" />
 </o:shapelayout></xml><![endif]-->
</head>

<body bgcolor=3Dwhite lang=3DEN-US link=3Dblue vlink=3Dpurple>

<div class=3DSection1>

<p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>Todd,<o:p></o:p></span></p>

<p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>Thanks for the reply.<o:p></o:p></span></p>

<p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'>[PA] Comments below...<o:p></o:p></span></p>

<p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";
color:#1F497D'><o:p>&nbsp;</o:p></span></p>

<div>

<div style=3D'border:none;border-top:solid #B5C4DF 1.0pt;padding:3.0pt =
0in 0in 0in'>

<p class=3DMsoNormal><b><span =
style=3D'font-size:10.0pt;font-family:"Tahoma","sans-serif"'>From:</span>=
</b><span
style=3D'font-size:10.0pt;font-family:"Tahoma","sans-serif"'> todd =
glassey
[mailto:tglassey@earthlink.net] <br>
<b>Sent:</b> Monday, September 10, 2007 2:53 PM<br>
<b>To:</b> Parvez Anandam; nea@ietf.org<br>
<b>Subject:</b> Re: [Nea] RE: Newcomer comments on
draft-ietf-nea-requirements-04<o:p></o:p></span></p>

</div>

</div>

<p class=3DMsoNormal><o:p>&nbsp;</o:p></p>

<div>

<p class=3DMsoNormal><span style=3D'font-size:10.0pt'>Parvez - in spirit =
I
understand what you are trying to do but I disagree and will enumerate =
why
below...</span><o:p></o:p></p>

</div>

<div>

<p class=3DMsoNormal>&nbsp;<o:p></o:p></p>

</div>

<div>

<p class=3DMsoNormal>---- Original Message ----- <o:p></o:p></p>

</div>

<blockquote style=3D'border:none;border-left:solid black =
1.5pt;padding:0in 0in 0in 4.0pt;
margin-left:3.75pt;margin-top:5.0pt;margin-right:0in;margin-bottom:5.0pt'=
>

<div>

<p class=3DMsoNormal style=3D'background:#E4E4E4'><b><span =
style=3D'font-size:10.0pt;
font-family:"Arial","sans-serif"'>From:</span></b><span =
style=3D'font-size:10.0pt;
font-family:"Arial","sans-serif"'> <a =
href=3D"mailto:panandam@SonicWALL.com"
title=3D"panandam@SonicWALL.com">panandam@SonicWALL.com</a> =
<o:p></o:p></span></p>

</div>

<div>

<p class=3DMsoNormal><b><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>To:</span></b=
><span
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'> <a
href=3D"mailto:nea@ietf.org" title=3D"nea@ietf.org">nea@ietf.org</a> =
<o:p></o:p></span></p>

</div>

<div>

<p class=3DMsoNormal><b><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Sent:</span><=
/b><span
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'> Monday, =
September
10, 2007 1:14 PM<o:p></o:p></span></p>

</div>

<div>

<p class=3DMsoNormal><b><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Subject:</spa=
n></b><span
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'> [Nea] RE: =
Newcomer
comments on draft-ietf-nea-requirements-04<o:p></o:p></span></p>

</div>

<div>

<p class=3DMsoNormal><o:p>&nbsp;</o:p></p>

</div>

<p><span style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>I =
have a</span>
<span style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>few =
additional
comments, following Gary&#8217;s,</span> <span =
style=3D'font-size:10.0pt;
font-family:"Arial","sans-serif"'>that also</span> <span =
style=3D'font-size:10.0pt;
font-family:"Arial","sans-serif"'>pertain</span> <span =
style=3D'font-size:10.0pt;
font-family:"Arial","sans-serif"'>to =
draft-ietf-nea-requirements-04.</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Page 35
Section 7.4</span><o:p></o:p></p>

<p><span style=3D'font-family:"Calibri","sans-serif"'>Please consider =
adding a
new requirement for the PT protocol to support non-persistent network
connections. </span><o:p></o:p></p>

</blockquote>

<p><span style=3D'font-family:"Calibri","sans-serif"'>This will not be a =
viable
thing to do in the real world no matter how nice it sounds I =
fear...<span
style=3D'color:#1F497D'><o:p></o:p></span></span></p>

<p><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>[PA]
We have empirical evidence that this is possible in the real world, with =
our
End Point Control solution. A classic example is a smartphone on a =
cellular
data network, using an SSL VPN.<o:p></o:p></span></p>

<blockquote style=3D'border:none;border-left:solid black =
1.5pt;padding:0in 0in 0in 4.0pt;
margin-left:3.75pt;margin-top:5.0pt;margin-right:0in;margin-bottom:5.0pt'=
>

<p><span style=3D'font-family:"Calibri","sans-serif"'>This is especially
important for mobile devices that roam or have air time charges. =
&nbsp;This has
an impact on reassessments triggered by the NEA server; in some cases =
(e.g. NAT
traversal), the only option may be for the NEA client to periodically =
contact the
NEA server to determine whether a reassessment is =
necessary.</span><o:p></o:p></p>

</blockquote>

<p><span style=3D'font-family:"Calibri","sans-serif"'>I =
disagree...&nbsp;
Supporting Mobile Device (Nomadic Computing) and its Integrity for
&quot;devices that are based on non-persistant connections&quot; may not =
be
such a good idea since it will dramatically increase the overhead in any =
NEA
system at a minimum and also will create a HUGE level of pain for people
running mixed networks... For instance it would be rather easy to =
program a PDA
to slam a NEA server with continuous requests for service after =
service.</span><o:p></o:p></p>

<p><span style=3D'font-family:"Calibri","sans-serif"'>To deal with these =
types of
devices,&nbsp; the NEA system will also have to maintain state tables =
for all
of its local &quot;recent&quot; connections since under any reasonable =
policy
requiring continuous certification of competence or NEA policy =
compliance means
repeating the same set of messages again and again... =
<o:p></o:p></span></p>

<p><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>[PA]
Is the scope of NEA restricted to non-mobile devices, then? I hope not. =
Maintaining
state is common practice, for VPNs . I&#8217;m not sure there is an =
alternative,
if one is to support devices such as smartphones. Such devices are =
increasingly
being used to run SSL VPN clients; they are part of a WAN, &nbsp;not a =
LAN. The
denial of service considerations are important, I =
agree.<o:p></o:p></span></p>

<blockquote style=3D'border:none;border-left:solid black =
1.5pt;padding:0in 0in 0in 4.0pt;
margin-left:3.75pt;margin-top:5.0pt;margin-right:0in;margin-bottom:5.0pt'=
>

<p><span style=3D'font-family:"Calibri","sans-serif"'>Page 44 Section =
9.1
Paragraph 3</span><o:p></o:p></p>

<p><span style=3D'font-family:"Calibri","sans-serif"'>As Gary stated, =
due to
security policy, network owners may wish not to disclose the attributes =
being
requested. </span><o:p></o:p></p>

</blockquote>

<p><span style=3D'font-family:"Calibri","sans-serif"'>Then why are they =
allowing
people to plug into their networks at all without disclosing those =
policies...<o:p></o:p></span></p>

<p><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>[PA]
The users may be connecting to a corporate network from home, for =
example. The
attributes the network owners are looking for will, of course, be =
disclosed to
the Posture Collectors, but that does not mean that a (human) user =
should necessarily
be aware of what is being collected. This prevents a user from =
configuring an
unauthorized laptop to simulate an authorized one (see below). Of =
course, the
privacy issues increase in scope significantly once the user is out of =
the loop;
thus Gary&#8217;s comment about preventing open-ended dredging of =
attributes.<o:p></o:p></span></p>

<blockquote style=3D'border:none;border-left:solid black =
1.5pt;padding:0in 0in 0in 4.0pt;
margin-left:3.75pt;margin-top:5.0pt;margin-right:0in;margin-bottom:5.0pt'=
>

<p><span style=3D'font-family:"Calibri","sans-serif"'>Imagine an =
attacker who
restricts all posture collection in order to determine what the network
operators are looking for.</span><o:p></o:p></p>

</blockquote>

<p><span style=3D'font-family:"Calibri","sans-serif"'>But how would that =
system
work in a NEA environment... it would be getting routing access to the =
switches
right? So what is it that system would be doing?</span><o:p></o:p></p>

<p><span style=3D'font-family:"Calibri","sans-serif"'>Remember the key =
purpose of
NEA is to BOTH allow one to force compliance with pre-existing NETWORK =
POLICIES
in a new or recently powered-up system AND&nbsp;the continuously =
recertifying
those systems which passed previously if they still continue to pass =
their
tests. <o:p></o:p></span></p>

<p><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>[PA]
The attack I envision goes something like this. The attacker plugs in =
her laptop
into a conference room jack at a competitor&#8217;s. She obtains an IP =
address
from the central VPN gateway that everyone in the building uses, but =
this does
not allow her to go beyond the VPN gateway. Her laptop is requested to =
perform
NEA checks. As the NEA requirements are currently written, she will be =
informed
of all the attributes that are being requested, say the presence of =
anti-virus program
XYZ and the presence of a specific Windows registry key. Knowing this =
critical
information, she runs program XYZ from her large attack arsenal and adds =
the
registry entry. Her laptop now conforms to the NEA security policies of =
the VPN
gateway and is indistinguishable from a corporate =
laptop.<o:p></o:p></span></p>

<p><span style=3D'font-family:"Calibri","sans-serif"'>To do that there =
needs to
be some way of communicating what those requirements are to the client =
in an
interactive mode. <o:p></o:p></span></p>

<p><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>[PA]
Certainly, there needs to be some secure way of communicating those =
requirements
from the Posture Validators to the Posture Collectors; the PA, PB and PT
protocols take care of that. The end-user need not know the contents of =
that
communication, however, for the reason I described =
above.<o:p></o:p></span></p>

<p><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>[PA]
Thanks,<o:p></o:p></span></p>

<p><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>[PA]
Parvez Anandam<o:p></o:p></span></p>

<blockquote style=3D'border:none;border-left:solid black =
1.5pt;padding:0in 0in 0in 4.0pt;
margin-left:3.75pt;margin-top:5.0pt;margin-right:0in;margin-bottom:5.0pt'=
>

<p>&nbsp;<o:p></o:p></p>

<p><span style=3D'font-family:"Calibri","sans-serif"'>Page 45 Section =
9.1
Paragraph 6</span><o:p></o:p></p>

<p><span style=3D'font-family:"Calibri","sans-serif"'>This is a =
continuation of
the thought in the comment above and concerns this item in the</span> =
<span
style=3D'font-family:"Calibri","sans-serif"'>NEA</span> <span =
style=3D'font-family:
"Calibri","sans-serif"'>disclosure statement: &#8220;What information is
required&#8221;. Such disclosure</span><span style=3D'font-size:10.0pt;
font-family:"Arial","sans-serif"'> must be subject to&nbsp; the security =
policy
of the network owner. What is the intent of specificity in this =
guidance?&nbsp;
Does the network owner have the latitude of providing a more general =
disclosure
statement?</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Parvez
Anandam</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Tahoma","sans-serif"'>____________=
_________________________________<br>
<b>From:</b> Gary Tomlinson<br>
<b>Sent:</b> Monday, September 10, 2007 1:11 PM<br>
<b>To:</b> nea@ietf.org<br>
<b>Subject:</b> Newcomer comments on =
draft-ietf-nea-requirements-04</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Hello NEA
WG,</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Several of
us at SonicWALL have recently been following the NEA WG in general and
specifically draft-ietf-nea-requirements-03 and </span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>draft-ietf-ne=
a-requirements-04.&nbsp;
I have the following comments pertaining to =
draft-ietf-nea-requirements-04.</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Page 32
Section 7.1, Requirement C-9</span><o:p></o:p></p>

<p><span style=3D'font-size:10.0pt'>At a minimum, C-9 should explicitly =
require
UTF-8 encodings be supported in strings in order to adhere to the =
guidance
specified in RFC 2277.&nbsp;&nbsp; </span><o:p></o:p></p>

<p><span style=3D'font-size:10.0pt'>Is there any reason to support =
character sets
other than UTF-8?&nbsp; Supporting only UTF-8 makes the protocol
straightforward on interpreting string </span><o:p></o:p></p>

<p><span style=3D'font-size:10.0pt'>encodings and normalization to =
locale
specific character sets is generally easy to implement on most OS =
platforms
using their intrinsic libraries.</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Page 33,
Section 7.2, Requirement PA-1</span><o:p></o:p></p>

<p><span style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>The =
NEA
model seems to imply an extensible standard attribute name space, but =
PA-1 does
not incorporate this as a requirement. Nor</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>does PA-1
specify a requirement for the standard attribute name space schema to be
uniquely identifiable from vendor attribute name </span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>spaces. =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Page 33,
Section 7.2, Requirement PA-2</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>T=
he
NEA model implies a plurality of extensible vendor attribute name spaces =
and
concurrent instances of each schema, but </span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>P=
A-2
does not&nbsp; incorporate this as a requirement.&nbsp; Nor does PA-2 =
specify a
requirement for vendor attribute name space schemas to =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>b=
e
uniquely identifiable.</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Page 33,
Section 7.2, Requirement PA-4</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>T=
he
network owner security policy may restrict giving the user complete =
visibility
into why an assessment failed in order to prevent </span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>e=
xposing
unnecessary information to potential attackers, especially prior to user
authentication.&nbsp; With this in mind, please consider =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>c=
hanging
&quot;This enables&quot; to &quot;This optionally enables the Posture =
Client,
subject to disclosure policy of the NEA Server, =
&#8221;.</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Page 33,
Section 7.12 Requirement PA-5</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>I=

am little confused over the recent comments debating downgrading [MUST] =
to
[SHOULD] or [MAY].&nbsp; The decision to utilize end-to-end =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>c=
ryptography
is use case topology dependent.&nbsp; If these use cases are deemed to =
be in
scope, then doesn't it make sense for this to be a =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>[=
MUST]
requirement and amend the verbiage to read &quot;&#8230; MUST be =
optionally
capable of providing &#8230;&quot;?</span><o:p></o:p></p>

<p class=3DMsoNormal><o:p>&nbsp;</o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Page 34
Section 7.2</span><o:p></o:p></p>

<p><span style=3D'font-size:10.0pt'>I noticed the former PA-8 =
requirement in
draft-ietf-nea-requirements-03 was dropped; apparently in favor of the =
more
generalized new C-9 </span><o:p></o:p></p>

<p><span style=3D'font-size:10.0pt'>requirement in
draft-ietf-nea-requirements-04.&nbsp; I believe there still exists a =
need for a
PA specific I18N string encoding requirement for =
non-</span><o:p></o:p></p>

<p><span style=3D'font-size:10.0pt'>displayable request and response =
attributes.
Request attributes for end point artifacts such as File, Directory, =
Process
Name, etc are used in </span><o:p></o:p></p>

<p><span style=3D'font-size:10.0pt'>common practice today in NEA =
equivalent
deployments I am aware of.&nbsp; Again, as mentioned in the comment =
regarding
the C-9 requirement, </span><o:p></o:p></p>

<p><span style=3D'font-size:10.0pt'>UTF-8 makes things unambiguous and =
at a
minimum must be a supported encoding.</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Page 34,
Section 7.3 Requirement PB-3</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>I=
s
there any reason in PB-3 not to specify identifiers be based upon SMI
enterprise numbers?&nbsp; This requirement looks much like that =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>a=
lready
faced by SNMP and LDAP; and we already have the IANA machinery to =
leverage
here.</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Page 35,
Section 7.4 Requirement PT-2</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>I=
n
addition to mutual authentication specified in PT-2, the PT protocol =
must also
be capable of supporting server side only authentication =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>a=
s
an alternative to mutual authentication.&nbsp; Server side only =
authentication
is common practice today in SSL VPNs. With this in mind, please =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>c=
onsider
changing &#8220;mutual authentication&#8221; to &#8220;mutual and server =
side
only authentication schemes&#8221;.</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Page 44,
Section 9.1 Paragraph 2</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>I=
n
order to prevent privacy abuse by network owners, PA request attributes =
must be
restricted to information&nbsp; known apriori&nbsp; rather =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>t=
han
be open ended in order to prevent dredging for artifact information. For
example a PV looking for a domain named </span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>&=
#8220;mydomain.com&#8221;
should be allowed, whereas requesting a domain named &#8220;*&#8221; =
should be
disallowed by the PC.</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Page 46,
Section 9.2&nbsp; Paragraph 2</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>I=

don't think Model 1 is practical or even possible in some cases.&nbsp; =
This
model works only when both the PC and PV have an immutable =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>s=
et
of attributes known apriori by each.&nbsp; However, it is common =
practice in
existing SSL VPN s to request information on the existence of files, =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>d=
irectories,
processes and other artifacts that are configurable assessment policy on =
the
NEA Server. In order for model 1 to potentially work, =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>t=
he
NEA client would have to send enormous amounts of information; =
potentially an
enumeration of its entire file name space, process name space, =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>r=
egistry
name space, etc.</span> <o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif"'>Page 46,
Section 9.2 Paragraph 4</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>M=
odel
3 is used in common practice today by many of the NEA like =
deployments.&nbsp;
At a minimum, it must be supported .&nbsp; I can also see model 2 =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>b=
eing
used effectively by a subset of PC/PV providers.&nbsp; It occurs to me, =
the
decision to use either model 2 or model 3 is one the PC/PV provider =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>m=
akes
and that the NEA protocols could consider supporting both models if =
necessary;
but Model 3 appears to be a superset of Model 2 and as =
</span><o:p></o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>s=
uch
so it might make sense to simply the protocol and only support model 3. =
My
preference is to always make things simpler if =
possible.</span><o:p></o:p></p>

<p class=3DMsoNormal><o:p>&nbsp;</o:p></p>

<p><span =
style=3D'font-size:10.0pt;font-family:"Arial","sans-serif";color:black'>G=
ary
Tomlinson</span><o:p></o:p></p>

<div class=3DMsoNormal align=3Dcenter style=3D'text-align:center'>

<hr size=3D2 width=3D"100%" align=3Dcenter>

</div>

<p class=3DMsoNormal>_______________________________________________<br>
Nea mailing list<br>
Nea@ietf.org<br>
https://www1.ietf.org/mailman/listinfo/nea<o:p></o:p></p>

</blockquote>

</div>

</body>

</html>

------_=_NextPart_001_01C7F403.D9AAF09C--



--===============0761562918==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea

--===============0761562918==--





From nea-bounces@ietf.org Mon Sep 10 21:26:47 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IUuWb-0005Zd-TU; Mon, 10 Sep 2007 21:26:45 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IUuWY-0005Yw-97
	for nea-confirm+ok@megatron.ietf.org; Mon, 10 Sep 2007 21:26:42 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IUuWX-0005Yj-SW
	for nea@ietf.org; Mon, 10 Sep 2007 21:26:41 -0400
Received: from sj-iport-6.cisco.com ([171.71.176.117])
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IUuWW-0003Qj-S1
	for nea@ietf.org; Mon, 10 Sep 2007 21:26:41 -0400
X-IronPort-AV: E=Sophos;i="4.20,234,1186383600"; 
	d="scan'208,217";a="215659774"
Received: from sj-dkim-4.cisco.com ([171.71.179.196])
	by sj-iport-6.cisco.com with ESMTP; 10 Sep 2007 18:26:40 -0700
Received: from sj-core-3.cisco.com (sj-core-3.cisco.com [171.68.223.137])
	by sj-dkim-4.cisco.com (8.12.11/8.12.11) with ESMTP id l8B1Qdp7017046; 
	Mon, 10 Sep 2007 18:26:39 -0700
Received: from xbh-sjc-231.amer.cisco.com (xbh-sjc-231.cisco.com
	[128.107.191.100])
	by sj-core-3.cisco.com (8.12.10/8.12.6) with ESMTP id l8B1Qd32020814;
	Tue, 11 Sep 2007 01:26:39 GMT
Received: from xmb-sjc-22d.amer.cisco.com ([128.107.191.68]) by
	xbh-sjc-231.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.1830); 
	Mon, 10 Sep 2007 18:26:39 -0700
Received: from 171.69.75.24 ([171.69.75.24]) by xmb-sjc-22d.amer.cisco.com
	([128.107.191.68]) with Microsoft Exchange Server HTTP-DAV ; 
	Tue, 11 Sep 2007 01:26:39 +0000
User-Agent: Microsoft-Entourage/11.3.6.070618
Date: Mon, 10 Sep 2007 18:26:39 -0700
Subject: Re: [Nea] RE: Newcomer comments on draft-ietf-nea-requirements-04
From: Kaushik Narayan <kaushik@cisco.com>
To: <panandam@SonicWALL.com>, <nea@ietf.org>
Message-ID: <C30B3CDF.AAED%kaushik@cisco.com>
Thread-Topic: [Nea] RE: Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acfz5qfsPz1SyEhoRu2ib5KW8a/VZAAABvKQAAsCVYc=
In-Reply-To: <D37A04F05A87044A8DCEEC0BF9A3AB1D07D024@us0exb10.us.sonicwall.com>
Mime-version: 1.0
X-OriginalArrivalTime: 11 Sep 2007 01:26:39.0449 (UTC)
	FILETIME=[CD4FB490:01C7F412]
DKIM-Signature: v=0.5; a=rsa-sha256; q=dns/txt; l=22202; t=1189473999;
	x=1190337999; c=relaxed/simple; s=sjdkim4002;
	h=Content-Type:From:Subject:Content-Transfer-Encoding:MIME-Version;
	d=cisco.com; i=kaushik@cisco.com;
	z=From:=20Kaushik=20Narayan=20<kaushik@cisco.com>
	|Subject:=20Re=3A=20[Nea]=20RE=3A=20Newcomer=20comments=20on=20draft-ietf
	-nea-requirements-04 |Sender:=20;
	bh=VCjN7xsobSurHdjsgAAZu8Zz80YlGfeofGmSwlVdYdI=;
	b=EbtjRbIRWouPoqfKhO+gR6PqrUduWN4AVVgdYe8Hf26fxARGLBRf01cpMYQZqqsEHSAc3I67
	x+tz2TmVk/zqnO8aNJk3V8vGpwcvuE2O9ec7K/t41ArqC3dG70if9dJE;
Authentication-Results: sj-dkim-4; header.From=kaushik@cisco.com; dkim=pass (
	sig from cisco.com/sjdkim4002 verified; ); 
X-Spam-Score: 0.0 (/)
X-Scan-Signature: c8611c7316981838cbe4195d07ac7fdb
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Content-Type: multipart/mixed; boundary="===============1701023730=="
Errors-To: nea-bounces@ietf.org

> This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

--===============1701023730==
Content-type: multipart/alternative;
	boundary="B_3272293600_4380697"

> This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

--B_3272293600_4380697
Content-type: text/plain;
	charset="ISO-8859-1"
Content-transfer-encoding: quoted-printable

Hi Parvez,

Please find my reply inline.

> <snipped>
>=20
> I have a few additional comments, following Gary=B9s, that also pertain to
> draft-ietf-nea-requirements-04.
>=20
> Page 35 Section 7.4
>=20
> Please consider adding a new requirement for the PT protocol to support
> non-persistent network connections. This is especially important for mobi=
le
> devices that roam or have air time charges.  This has an impact on
> reassessments triggered by the NEA server; in some cases (e.g. NAT traver=
sal),
> the only option may be for the NEA client to periodically contact the NEA
> server to determine whether a reassessment is necessary.
>=20
>=20
>=20
> <Kaushik>=20
>=20
> I am not sure about what you mean by =B3persistent network connection=B2, i.e=
.
> What does that mean for a PT that runs over a network access protocol suc=
h as
> 802.1x or IKEv2. In any case the requirements to support client & server
> triggered assessment/re-assessment is a SHOULD and not a MUST.
>=20
>=20
> regards,
>   kaushik
>=20
>=20
> </Kaushik>
>=20
>=20
> Page 44 Section 9.1 Paragraph 3
>=20
> As Gary stated, due to security policy, network owners may wish not to
> disclose the attributes being requested. Imagine an attacker who restrict=
s all
> posture collection in order to determine what the network operators are
> looking for.
>=20
> Page 45 Section 9.1 Paragraph 6
>=20
> This is a continuation of the thought in the comment above and concerns t=
his
> item in the NEA disclosure statement: =B3What information is required=B2. Suc=
h
> disclosure must be subject to  the security policy of the network owner. =
What
> is the intent of specificity in this guidance?  Does the network owner ha=
ve
> the latitude of providing a more general disclosure statement?
>=20
> Parvez Anandam
>=20
> _____________________________________________
> From: Gary Tomlinson
> Sent: Monday, September 10, 2007 1:11 PM
> To: nea@ietf.org
> Subject: Newcomer comments on draft-ietf-nea-requirements-04
>=20
> Hello NEA WG,
>=20
> Several of us at SonicWALL have recently been following the NEA WG in gen=
eral
> and specifically draft-ietf-nea-requirements-03 and
>=20
> draft-ietf-nea-requirements-04.  I have the following comments pertaining=
 to
> draft-ietf-nea-requirements-04.
>=20
> Page 32 Section 7.1, Requirement C-9
>=20
> At a minimum, C-9 should explicitly require UTF-8 encodings be supported =
in
> strings in order to adhere to the guidance specified in RFC 2277.
>=20
> Is there any reason to support character sets other than UTF-8?  Supporti=
ng
> only UTF-8 makes the protocol straightforward on interpreting string
>=20
> encodings and normalization to locale specific character sets is generall=
y
> easy to implement on most OS platforms using their intrinsic libraries.
>=20
> Page 33, Section 7.2, Requirement PA-1
>=20
> The NEA model seems to imply an extensible standard attribute name space,=
 but
> PA-1 does not incorporate this as a requirement. Nor
>=20
> does PA-1 specify a requirement for the standard attribute name space sch=
ema
> to be uniquely identifiable from vendor attribute name
>=20
> spaces.=20
>=20
> Page 33, Section 7.2, Requirement PA-2
>=20
> The NEA model implies a plurality of extensible vendor attribute name spa=
ces
> and concurrent instances of each schema, but
>=20
> PA-2 does not  incorporate this as a requirement.  Nor does PA-2 specify =
a
> requirement for vendor attribute name space schemas to
>=20
> be uniquely identifiable.
>=20
> Page 33, Section 7.2, Requirement PA-4
>=20
> The network owner security policy may restrict giving the user complete
> visibility into why an assessment failed in order to prevent
>=20
> exposing unnecessary information to potential attackers, especially prior=
 to
> user authentication.  With this in mind, please consider
>=20
> changing "This enables" to "This optionally enables the Posture Client,
> subject to disclosure policy of the NEA Server, =B2.
>=20
> Page 33, Section 7.12 Requirement PA-5
>=20
> I am little confused over the recent comments debating downgrading [MUST]=
 to
> [SHOULD] or [MAY].  The decision to utilize end-to-end
>=20
> cryptography is use case topology dependent.  If these use cases are deem=
ed to
> be in scope, then doesn't it make sense for this to be a
>=20
> [MUST] requirement and amend the verbiage to read "=8A MUST be optionally
> capable of providing =8A"?
>=20
>=20
> Page 34 Section 7.2
>=20
> I noticed the former PA-8 requirement in draft-ietf-nea-requirements-03 w=
as
> dropped; apparently in favor of the more generalized new C-9
>=20
> requirement in draft-ietf-nea-requirements-04.  I believe there still exi=
sts a
> need for a PA specific I18N string encoding requirement for non-
>=20
> displayable request and response attributes. Request attributes for end p=
oint
> artifacts such as File, Directory, Process Name, etc are used in
>=20
> common practice today in NEA equivalent deployments I am aware of.  Again=
, as
> mentioned in the comment regarding the C-9 requirement,
>=20
> UTF-8 makes things unambiguous and at a minimum must be a supported encod=
ing.
>=20
> Page 34, Section 7.3 Requirement PB-3
>=20
> Is there any reason in PB-3 not to specify identifiers be based upon SMI
> enterprise numbers?  This requirement looks much like that
>=20
> already faced by SNMP and LDAP; and we already have the IANA machinery to
> leverage here.
>=20
> Page 35, Section 7.4 Requirement PT-2
>=20
> In addition to mutual authentication specified in PT-2, the PT protocol m=
ust
> also be capable of supporting server side only authentication
>=20
> as an alternative to mutual authentication.  Server side only authenticat=
ion
> is common practice today in SSL VPNs. With this in mind, please
>=20
> consider changing =B3mutual authentication=B2 to =B3mutual and server side only
> authentication schemes=B2.
>=20
> Page 44, Section 9.1 Paragraph 2
>=20
> In order to prevent privacy abuse by network owners, PA request attribute=
s
> must be restricted to information  known apriori  rather
>=20
> than be open ended in order to prevent dredging for artifact information.=
 For
> example a PV looking for a domain named
>=20
> =B3mydomain.com=B2 should be allowed, whereas requesting a domain named =B3*=B2 s=
hould
> be disallowed by the PC.
>=20
> Page 46, Section 9.2  Paragraph 2
>=20
> I don't think Model 1 is practical or even possible in some cases.  This =
model
> works only when both the PC and PV have an immutable
>=20
> set of attributes known apriori by each.  However, it is common practice =
in
> existing SSL VPN s to request information on the existence of files,
>=20
> directories, processes and other artifacts that are configurable assessme=
nt
> policy on the NEA Server. In order for model 1 to potentially work,
>=20
> the NEA client would have to send enormous amounts of information; potent=
ially
> an enumeration of its entire file name space, process name space,
>=20
> registry name space, etc.
>=20
> Page 46, Section 9.2 Paragraph 4
>=20
> Model 3 is used in common practice today by many of the NEA like deployme=
nts.
> At a minimum, it must be supported .  I can also see model 2
>=20
> being used effectively by a subset of PC/PV providers.  It occurs to me, =
the
> decision to use either model 2 or model 3 is one the PC/PV provider
>=20
> makes and that the NEA protocols could consider supporting both models if
> necessary; but Model 3 appears to be a superset of Model 2 and as
>=20
> such so it might make sense to simply the protocol and only support model=
 3.
> My preference is to always make things simpler if possible.
>=20
>=20
> Gary Tomlinson
>=20
>=20
>=20
> _______________________________________________
> Nea mailing list
> Nea@ietf.org
> https://www1.ietf.org/mailman/listinfo/nea



--B_3272293600_4380697
Content-type: text/html;
	charset="ISO-8859-1"
Content-transfer-encoding: quoted-printable

<HTML>
<HEAD>
<TITLE>Re: [Nea] RE: Newcomer comments on draft-ietf-nea-requirements-04</T=
ITLE>
</HEAD>
<BODY>
<FONT FACE=3D"Verdana, Helvetica, Arial"><SPAN STYLE=3D'font-size:12.0px'>Hi Pa=
rvez,<BR>
<BR>
Please find my reply inline.<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPAN STYL=
E=3D'font-size:12.0px'>&lt;snipped&gt;<BR>
<BR>
</SPAN></FONT><SPAN STYLE=3D'font-size:12.0px'><FONT FACE=3D"Arial">I have a</F=
ONT><FONT FACE=3D"Verdana, Helvetica, Arial"> </FONT><FONT FACE=3D"Arial">few ad=
ditional comments, following Gary&#8217;s,</FONT><FONT FACE=3D"Verdana, Helvet=
ica, Arial"> </FONT><FONT FACE=3D"Arial">that also</FONT><FONT FACE=3D"Verdana, =
Helvetica, Arial"> </FONT><FONT FACE=3D"Arial">pertain</FONT><FONT FACE=3D"Verda=
na, Helvetica, Arial"> </FONT><FONT FACE=3D"Arial">to draft-ietf-nea-requireme=
nts-04.<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Page 35 Section 7.4<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
Please consider adding a new requirement for the PT protocol to support non=
-persistent network connections. This is especially important for mobile dev=
ices that roam or have air time charges. &nbsp;This has an impact on reasses=
sments triggered by the NEA server; in some cases (e.g. NAT traversal), the =
only option may be for the NEA client to periodically contact the NEA server=
 to determine whether a reassessment is necessary.<BR>
<BR>
<BR>
<BR>
&lt;Kaushik&gt; <BR>
<BR>
I am not sure about what you mean by &#8220;persistent network connection&#=
8221;, i.e. What does that mean for a PT that runs over a network access pro=
tocol such as 802.1x or IKEv2. In any case the requirements to support clien=
t &amp; server triggered assessment/re-assessment is a SHOULD and not a MUST=
.<BR>
<BR>
<BR>
regards,<BR>
&nbsp;&nbsp;kaushik<BR>
<BR>
<BR>
&lt;/Kaushik&gt;<BR>
<BR>
<BR>
Page 44 Section 9.1 Paragraph 3<BR>
<BR>
As Gary stated, due to security policy, network owners may wish not to disc=
lose the attributes being requested. Imagine an attacker who restricts all p=
osture collection in order to determine what the network operators are looki=
ng for.<BR>
<BR>
Page 45 Section 9.1 Paragraph 6<BR>
<BR>
This is a continuation of the thought in the comment above and concerns thi=
s item in the NEA disclosure statement: &#8220;What information is required&=
#8221;. Such disclosure</FONT><FONT FACE=3D"Arial"> must be subject to &nbsp;t=
he security policy of the network owner. What is the intent of specificity i=
n this guidance? &nbsp;Does the network owner have the latitude of providing=
 a more general disclosure statement?<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Parvez Anandam<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Tahoma">_____________________________________________<BR=
>
<B>From:</B> Gary Tomlinson<BR>
<B>Sent:</B> Monday, September 10, 2007 1:11 PM<BR>
<B>To:</B> nea@ietf.org<BR>
<B>Subject:</B> Newcomer comments on draft-ietf-nea-requirements-04<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Hello NEA WG,<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Several of us at SonicWALL have recently been fol=
lowing the NEA WG in general and specifically draft-ietf-nea-requirements-03=
 and <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">draft-ietf-nea-requirements-04. &nbsp;I have the =
following comments pertaining to draft-ietf-nea-requirements-04.<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Page 32 Section 7.1, Requirement C-9<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">At a minimum, C-9 should explicitly req=
uire UTF-8 encodings be supported in strings in order to adhere to the guida=
nce specified in RFC 2277. &nbsp;&nbsp;<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">Is there any reason to support characte=
r sets other than UTF-8? &nbsp;Supporting only UTF-8 makes the protocol stra=
ightforward on interpreting string <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">encodings and normalization to locale s=
pecific character sets is generally easy to implement on most OS platforms u=
sing their intrinsic libraries.<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Page 33, Section 7.2, Requirement PA-1<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">The NEA model seems to imply an extensible standa=
rd attribute name space, but PA-1 does not incorporate this as a requirement=
. Nor<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">does PA-1 specify a requirement for the standard =
attribute name space schema to be uniquely identifiable from vendor attribut=
e name <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">spaces. <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Page 33, Section 7.2, Requirement PA-2<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">The NEA model implies a plurality of extensible v=
endor attribute name spaces and concurrent instances of each schema, but <BR=
>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">PA-2 does not &nbsp;incorporate this as a require=
ment. &nbsp;Nor does PA-2 specify a requirement for vendor attribute name sp=
ace schemas to <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">be uniquely identifiable.<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Page 33, Section 7.2, Requirement PA-4<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">The network owner security policy may restrict gi=
ving the user complete visibility into why an assessment failed in order to =
prevent <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">exposing unnecessary information to potential att=
ackers, especially prior to user authentication. &nbsp;With this in mind, pl=
ease consider <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">changing &quot;This enables&quot; to &quot;This o=
ptionally enables the Posture Client, subject to disclosure policy of the NE=
A Server, &#8221;.<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Page 33, Section 7.12 Requirement PA-5<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">I am little confused over the recent comments deb=
ating downgrading [MUST] to [SHOULD] or [MAY]. &nbsp;The decision to utilize=
 end-to-end <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">cryptography is use case topology dependent. &nbs=
p;If these use cases are deemed to be in scope, then doesn't it make sense f=
or this to be a <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">[MUST] requirement and amend the verbiage to read=
 &quot;&#8230; MUST be optionally capable of providing &#8230;&quot;?<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
<BR>
</FONT><FONT FACE=3D"Arial">Page 34 Section 7.2<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">I noticed the former PA-8 requirement i=
n draft-ietf-nea-requirements-03 was dropped; apparently in favor of the mor=
e generalized new C-9 <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">requirement in draft-ietf-nea-requireme=
nts-04. &nbsp;I believe there still exists a need for a PA specific I18N str=
ing encoding requirement for non-<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">displayable request and response attrib=
utes. Request attributes for end point artifacts such as File, Directory, Pr=
ocess Name, etc are used in <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">common practice today in NEA equivalent=
 deployments I am aware of. &nbsp;Again, as mentioned in the comment regardi=
ng the C-9 requirement, <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Times New Roman">UTF-8 makes things unambiguous and at a=
 minimum must be a supported encoding.<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Page 34, Section 7.3 Requirement PB-3<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Is there any reason in PB-3 not to specify identi=
fiers be based upon SMI enterprise numbers? &nbsp;This requirement looks muc=
h like that <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">already faced by SNMP and LDAP; and we already ha=
ve the IANA machinery to leverage here.<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Page 35, Section 7.4 Requirement PT-2<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">In addition to mutual authentication specified in=
 PT-2, the PT protocol must also be capable of supporting server side only a=
uthentication <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">as an alternative to mutual authentication. &nbsp=
;Server side only authentication is common practice today in SSL VPNs. With =
this in mind, please <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">consider changing &#8220;mutual authentication&#8=
221; to &#8220;mutual and server side only authentication schemes&#8221;.<BR=
>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Page 44, Section 9.1 Paragraph 2<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">In order to prevent privacy abuse by network owne=
rs, PA request attributes must be restricted to information &nbsp;known apri=
ori &nbsp;rather <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">than be open ended in order to prevent dredging f=
or artifact information. For example a PV looking for a domain named <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">&#8220;mydomain.com&#8221; should be allowed, whe=
reas requesting a domain named &#8220;*&#8221; should be disallowed by the P=
C.<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Page 46, Section 9.2 &nbsp;Paragraph 2<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">I don't think Model 1 is practical or even possib=
le in some cases. &nbsp;This model works only when both the PC and PV have a=
n immutable <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">set of attributes known apriori by each. &nbsp;Ho=
wever, it is common practice in existing SSL VPN s to request information on=
 the existence of files, <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">directories, processes and other artifacts that a=
re configurable assessment policy on the NEA Server. In order for model 1 to=
 potentially work, <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">the NEA client would have to send enormous amount=
s of information; potentially an enumeration of its entire file name space, =
process name space, <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">registry name space, etc.</FONT><FONT FACE=3D"Verda=
na, Helvetica, Arial"> <BR>
<BR>
</FONT><FONT FACE=3D"Arial">Page 46, Section 9.2 Paragraph 4<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">Model 3 is used in common practice today by many =
of the NEA like deployments. &nbsp;At a minimum, it must be supported . &nbs=
p;I can also see model 2 <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">being used effectively by a subset of PC/PV provi=
ders. &nbsp;It occurs to me, the decision to use either model 2 or model 3 i=
s one the PC/PV provider <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">makes and that the NEA protocols could consider s=
upporting both models if necessary; but Model 3 appears to be a superset of =
Model 2 and as <BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
</FONT><FONT FACE=3D"Arial">such so it might make sense to simply the protoco=
l and only support model 3. My preference is to always make things simpler i=
f possible.<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
<BR>
</FONT><FONT FACE=3D"Arial">Gary Tomlinson<BR>
</FONT><FONT FACE=3D"Verdana, Helvetica, Arial"><BR>
<BR>
<HR ALIGN=3DCENTER SIZE=3D"3" WIDTH=3D"95%"></FONT><FONT FACE=3D"Monaco, Courier Ne=
w">_______________________________________________<BR>
Nea mailing list<BR>
Nea@ietf.org<BR>
https://www1.ietf.org/mailman/listinfo/nea<BR>
</FONT></SPAN></BLOCKQUOTE><SPAN STYLE=3D'font-size:12.0px'><FONT FACE=3D"Monac=
o, Courier New"><BR>
</FONT></SPAN>
</BODY>
</HTML>


--B_3272293600_4380697--



--===============1701023730==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea

--===============1701023730==--





From nea-bounces@ietf.org Tue Sep 11 14:57:05 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVAv1-0002fg-IW; Tue, 11 Sep 2007 14:57:03 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVAv0-0002fa-DH
	for nea-confirm+ok@megatron.ietf.org; Tue, 11 Sep 2007 14:57:02 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVAv0-0002fR-1n
	for nea@ietf.org; Tue, 11 Sep 2007 14:57:02 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IVAuy-0006Bw-Uk
	for nea@ietf.org; Tue, 11 Sep 2007 14:57:01 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id hsri260171sr for <nea@ietf.org>;
	Tue, 11 Sep 2007 11:53:22 -0700 (envelope-from
	<gtomlinson@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9143)
	with ESMTP; Tue, 11 Sep 2007 11:53:22 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Tue, 11 Sep 2007 11:57:03 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Date: Tue, 11 Sep 2007 11:56:50 -0700
Message-ID: <4DF8F0A275770B4387764386ABEF4D7A1009ED@us0exb10.us.sonicwall.com>
In-Reply-To: <022e01c7f3f7$24434ad0$174f7d40@home.glassey.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acfz9x48SlmndTlcSzKBm3+RbhwB3gArUexQ
From: <gtomlinson@SonicWALL.com>
To: <tglassey@earthlink.net>,
	<nea@ietf.org>
X-OriginalArrivalTime: 11 Sep 2007 18:57:03.0190 (UTC)
	FILETIME=[8A636760:01C7F4A5]
X-Mlf-Version: 6.0.1.9143
X-Mlf-UniqueId: o200709111853220084911
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 311e798ce51dbeacf5cdfcc8e9fda21b
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hello Todd,

Thanks for you comments. My comments are inline with the appropriate
Snippet.=20

>> -----Original Message-----
>> From: todd glassey [mailto:tglassey@earthlink.net]
>> Sent: Monday, September 10, 2007 3:08 PM
>> To: Gary Tomlinson; nea@ietf.org
>> Subject: Re: [Nea] Newcomer comments on=20
>> draft-ietf-nea-requirements-04
>>=20
>> Gary
>>=20
>> 	----- Original Message -----=20
>> 	From: gtomlinson@SonicWALL.com=20
>> 	To: nea@ietf.org=20
>> 	Sent: Monday, September 10, 2007 1:10 PM
>> 	Subject: [Nea] Newcomer comments on
>> draft-ietf-nea-requirements-04
>>=20

[omitted]

>>=20
>> 	Page 32 Section 7.1, Requirement C-9=20
>> 	At a minimum, C-9 should explicitly require UTF-8=20
>> encodings be supported in strings in order to adhere to the=20
>> guidance specified in RFC 2277.  =20
>>=20
> You mean its its important to keep the compliance with ISO=20
> UCS...  I agree - the minimum standard is UCS.
=20
Actually, I literally meant UTF-8 for the following reasons; a) it is
the encoding
specified in RFC 2277; b) it represents the ISO Universal Character Set
(UCS);
and c) is an encoding standard in addition to a character set.

[omitted]

>>=20
>> 	Page 33, Section 7.2, Requirement PA-1=20
>> 	The NEA model seems to imply an extensible standard=20
>> attribute name space, but PA-1 does not incorporate this as a=20
>> requirement. Nor does PA-1 specify a requirement for the=20
>> standard attribute name space schema to be uniquely=20
>> identifiable from vendor attribute name spaces.=20
>>=20
> This is specific to a use model - so document that use model =20
> and its requirement inside of PA-1.
>=20

The use case I had in mind is one in which there are multiple PC's
and PV's involved in the assessment of an endpoint. I assumed this=20
was implied in the model as required in section 7.3. The=20
reasoning behind the need for a name space identifier is=20
to distinguish the standard attributes from the vendor attributes;
although
it may be the case the WG is relying on PB-3 to implicitly serve this=20
purpose.  If so, that should be stated in PA-1. The reasoning behind an=20
extensible name space schema is two fold; one the schema needs to be
public for=20
any PV/PC combo to interoperate, and two extensibility allows the
protocol
to adapt over time.

> 	Page 33, Section 7.2, Requirement PA-2=20
> 	The NEA model implies a plurality of extensible vendor=20
> attribute name spaces and concurrent instances of each=20
> schema, but PA-2 does not  incorporate this as a requirement.=20
>  Nor does PA-2 specify a requirement for vendor attribute=20
> name space schemas to be uniquely identifiable.=20
>=20
> Again - a complete set of use models would eliminate this.

I'm not sure I can describe a complete set of use cases; although again
I assumed
the same use case as in PA-1 and thus the need to distinguish the vendor
specific=20
attributes from each other and from the standard attributes as well.
Again, if the WG=20
assumed PB-3 will provide this distinction this should be stated in
PA-2.  As to the
extensible vendor attribute name space I was thinking SNMP MIB OID model
in order to=20
enable the possibility of interoperability between PV/PC implementations
from different=20
vendors.  However, if this capability is out of scope and the WG expects
the standard=20
attribute set to serve this purpose, I'll retract this part of my PA-2
comment.

>>=20
>> 	Page 33, Section 7.2, Requirement PA-4=20
>> 	The network owner security policy may restrict giving=20
>> the user complete visibility into why an assessment failed in=20
>> order to prevent exposing unnecessary information to=20
>> potential attackers, especially prior to user authentication. =20
>>=20
> Here then is a specific use in which the system provides=20
> hierarchical error or failure messages with varying detail?

Yes, although I think the reasons are outside the scope of the=20
protocol itself.  I'm not sure how to phrase the guidance here=20
regarding how to provide useful information to the user as it is=20
at the discretion of the network operator's policy.  It seems fine to
me to have errors or failures which can be cryptic to users in policies=20
which don't want to provide such information to untrusted parties.

>=20
>> 	With this in mind, please consider changing "This=20
>> enables" to "This optionally enables the Posture Client,=20
>> subject to disclosure policy of the NEA Server, ".
>>=20
>> 	Page 33, Section 7.12 Requirement PA-5=20
>> 	I am little confused over the recent comments debating=20
>> downgrading [MUST] to [SHOULD] or [MAY].  The decision to=20
>> utilize end-to-end cryptography is use case topology=20
>> dependent.  If these use cases are deemed to be in scope,=20
>> then doesn't it make sense for this to be a [MUST]=20
>> requirement and amend the verbiage to read "... MUST be=20
>> optionally capable of providing ..."?=20
>>=20
> Again this is use-model specific.

Yes, many of these are use case specific, however the PA protocol
needs to enable use cases that are in scope; which I inferred from
Section 8.2.

>>=20

[omitted]

>>
>>	Page 35, Section 7.4 Requirement PT-2=20
>>	In addition to mutual authentication specified in PT-2,=20
>> the PT protocol must also be capable of supporting server=20
>> side only authentication as an alternative to mutual=20
>> authentication.  Server side only authentication is common=20
>> practice today in SSL VPNs. With this in mind, please=20
>> consider changing "mutual authentication" to "mutual and=20
>> server side only authentication schemes".=20
>=20
> I suggest the following change:=20
>=20
> Page 35, Section 7.4 Requirement PT-2
> "Authentictation schemes specific to the end-use models", so=20
> that the type of authentication - whether Server-Side,=20
> mutual, or circuit-level only can be done.
>=20

Yes, this is more of the intent I'm thinking as well.=20

> 	=20
>=20
>> 	Page 44, Section 9.1 Paragraph 2=20
>> 	In order to prevent privacy abuse by network owners, PA=20
>> request attributes must be restricted to information  known=20
>> apriori  rather than be open ended in order to prevent=20
>> dredging for artifact information. For example a PV looking=20
>> for a domain named "mydomain.com" should be allowed, whereas=20
>> requesting a domain named "*" should be disallowed by the PC.=20
>>=20
> Why? - What you suggest here means that each and ever domain=20
> MUST be individually enumerated rather than being able to say=20
> "Any Supported or Defined Domain"... I can see many instances=20
> - especially those inside of University's and the like which=20
> may have hundreds of domains, all which are operated under=20
> the University's Security Persona...
>=20

Good point, although I think my comment still stands in this case.
Asking are you a member of "myuniversity.edu" isn't open ended
in my mind since this is explicitly scoped as a specific group query.
My comment was to provide guidance to PC & PV developers to exercise
good judgment in maintaining user privacy in their disclosures via the
PA protocol exchanges.

>> 	Page 46, Section 9.2  Paragraph 2=20
>> 	I don't think Model 1 is practical or even possible in=20
>> some cases.  This model works only when both the PC and PV=20
>> have an immutable set of attributes known apriori by each. =20
>> However, it is common practice in existing SSL VPN s to=20
>> request information on the existence of files, directories,=20
>> processes and other artifacts that are configurable=20
>> assessment policy on the NEA Server. In order for model 1 to=20
>> potentially work, the NEA client would have to send enormous=20
>> amounts of information; potentially an enumeration of its=20
>> entire file name space, process name space, registry name space, etc.

>=20
> yes it would - but what is the value of the assurance it was=20
> to provide if the Client side didnt present these=20
> qualification datapoints?
>=20

I'm not following you on this one.  Can you provide more clarity on
this comment?

Gary


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Tue Sep 11 16:08:41 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVC2J-0002Pr-TO; Tue, 11 Sep 2007 16:08:39 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVC2G-0002PK-3P
	for nea-confirm+ok@megatron.ietf.org; Tue, 11 Sep 2007 16:08:36 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVC2F-0002PA-Ns
	for nea@ietf.org; Tue, 11 Sep 2007 16:08:35 -0400
Received: from elasmtp-spurfowl.atl.sa.earthlink.net ([209.86.89.66])
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IVC2E-0008AB-Ge
	for nea@ietf.org; Tue, 11 Sep 2007 16:08:35 -0400
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=earthlink.net;
	b=jjpuM81I8pFSOXVF6GoXox10r/G6ovdtUzNAYT+4AnLYJsc/P5enXn/NpwcYy9Ts;
	h=Received:Message-ID:From:To:References:Subject:Date:MIME-Version:Content-Type:Content-Transfer-Encoding:X-Priority:X-MSMail-Priority:X-Mailer:X-MimeOLE:X-ELNK-Trace:X-Originating-IP;
Received: from [64.125.79.23] (helo=gw)
	by elasmtp-spurfowl.atl.sa.earthlink.net with asmtp (Exim 4.34)
	id 1IVC2B-0007ul-SK; Tue, 11 Sep 2007 16:08:32 -0400
Message-ID: <01a101c7f4af$8d5bb440$174f7d40@home.glassey.com>
From: "todd glassey" <tglassey@earthlink.net>
To: <gtomlinson@SonicWALL.com>,
	<nea@ietf.org>
References: <4DF8F0A275770B4387764386ABEF4D7A1009ED@us0exb10.us.sonicwall.com>
Subject: Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Date: Tue, 11 Sep 2007 13:04:22 -0700
MIME-Version: 1.0
Content-Type: text/plain; format=flowed; charset="iso-8859-1";
	reply-type=original
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2900.3138
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.3138
X-ELNK-Trace: 01b7a7e171bdf5911aa676d7e74259b7b3291a7d08dfec79de9faa8987c954015056e3c60b0d3e7b350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c
X-Originating-IP: 64.125.79.23
X-Spam-Score: 0.0 (/)
X-Scan-Signature: e178fd6cb61ffb6940cd878e7fea8606
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Thanks Gary - response inline :-)
----- Original Message ----- 
From: <gtomlinson@SonicWALL.com>
To: <tglassey@earthlink.net>; <nea@ietf.org>
Sent: Tuesday, September 11, 2007 11:56 AM
Subject: RE: [Nea] Newcomer comments on draft-ietf-nea-requirements-04


Hello Todd,

Thanks for you comments. My comments are inline with the appropriate
Snippet.

>> -----Original Message-----
>> From: todd glassey [mailto:tglassey@earthlink.net]
>> Sent: Monday, September 10, 2007 3:08 PM
>> To: Gary Tomlinson; nea@ietf.org
>> Subject: Re: [Nea] Newcomer comments on
>> draft-ietf-nea-requirements-04
>>
>> Gary
>>
>> ----- Original Message ----- 
>> From: gtomlinson@SonicWALL.com
>> To: nea@ietf.org
>> Sent: Monday, September 10, 2007 1:10 PM
>> Subject: [Nea] Newcomer comments on
>> draft-ietf-nea-requirements-04
>>

[omitted]

>>
>> Page 32 Section 7.1, Requirement C-9
>> At a minimum, C-9 should explicitly require UTF-8
>> encodings be supported in strings in order to adhere to the
>> guidance specified in RFC 2277.
>>
> You mean its its important to keep the compliance with ISO
> UCS...  I agree - the minimum standard is UCS.

Actually, I literally meant UTF-8 for the following reasons; a) it is
the encoding
specified in RFC 2277; b) it represents the ISO Universal Character Set
(UCS);
and c) is an encoding standard in addition to a character set.

[omitted]

TSG: COOL

>>
>> Page 33, Section 7.2, Requirement PA-1
>> The NEA model seems to imply an extensible standard
>> attribute name space, but PA-1 does not incorporate this as a
>> requirement. Nor does PA-1 specify a requirement for the
>> standard attribute name space schema to be uniquely
>> identifiable from vendor attribute name spaces.
>>
> This is specific to a use model - so document that use model
> and its requirement inside of PA-1.
>

> The use case I had in mind is one in which there are multiple PC's
and PV's involved in the assessment of an endpoint.


> I assumed this was implied in the model as required in section 7.3. The 
> reasoning behind the need for a name space identifier is to distinguish 
> the standard attributes from the vendor attributes; althoughit may be the 
> case the WG is relying on PB-3 to implicitly serve this purpose.  If so, 
> that should be stated in PA-1. The reasoning behind an extensible name 
> space schema is two fold; one the schema needs to be  public for any PV/PC 
> combo to interoperate, and two extensibility allows the
protocol to adapt over time.

> Page 33, Section 7.2, Requirement PA-2
> The NEA model implies a plurality of extensible vendor
> attribute name spaces and concurrent instances of each
> schema, but PA-2 does not  incorporate this as a requirement.
>  Nor does PA-2 specify a requirement for vendor attribute
> name space schemas to be uniquely identifiable.
>
> Again - a complete set of use models would eliminate this.

I'm not sure I can describe a complete set of use cases; although again I 
assumed the same use case as in PA-1 and thus the need to distinguish the 
vendor specific  attributes from each other and from the standard attributes 
as well.

TSG: Gary - You nor anyone else here as individuals needs to docuemtn all of 
the use models or cases - the group collectively does need to document say 
about 5 or more solid use cases and the controls for each one.

TSG: It would also would be a good idea to map COBIT and ISO17799 use cases 
on top of the protocol as well so it is easier to absorb into Audit 
Regulatred environments.


Again, if the WG assumed PB-3 will provide this distinction this should be 
stated in PA-2.  As to the extensible vendor attribute name space I was 
thinking SNMP MIB OID model in order to  enable the possibility of 
interoperability between PV/PC implementations from different  vendors.


TSG: Gary - How will that fit into the control framework of SNMP then? Can 
we use SNMP to talk to the NEA Agent PC/PV systems?

However, if this capability is out of scope and the WG expects the standard 
attribute set to serve this purpose, I'll retract this part of my PA-2 
comment.

>>
>> Page 33, Section 7.2, Requirement PA-4
>> The network owner security policy may restrict giving
>> the user complete visibility into why an assessment failed in
>> order to prevent exposing unnecessary information to
>> potential attackers, especially prior to user authentication.
>>
> Here then is a specific use in which the system provides
> hierarchical error or failure messages with varying detail?

Yes, although I think the reasons are outside the scope of the  protocol 
itself.  I'm not sure how to phrase the guidance here  regarding how to 
provide useful information to the user as it is  at the discretion of the 
network operator's policy.

TSG: Then perhaps... the scope of the protocol definition may need to be 
adjusted then since these control systems are key to a successful deployment 
of NEA IMHO.

> It seems fine to me to have errors or failures which can be cryptic to 
> users in policies  which don't want to provide such information to 
> untrusted parties.

>
>> With this in mind, please consider changing "This
>> enables" to "This optionally enables the Posture Client,
>> subject to disclosure policy of the NEA Server, ".
>>
>> Page 33, Section 7.12 Requirement PA-5
>> I am little confused over the recent comments debating
>> downgrading [MUST] to [SHOULD] or [MAY].  The decision to
>> utilize end-to-end cryptography is use case topology
>> dependent.  If these use cases are deemed to be in scope,
>> then doesn't it make sense for this to be a [MUST]
>> requirement and amend the verbiage to read "... MUST be
>> optionally capable of providing ..."?
>>
> Again this is use-model specific.

Yes, many of these are use case specific, however the PA protocol
needs to enable use cases that are in scope; which I inferred from
Section 8.2.

TSG: This is one of the things we need to complete. The Inferring of Use 
Models with NEA is a no-no IMHO. The uses of NEA need to be tight and very 
very secure, meaning that they are well defined and controlled so there is 
little if any wiggle room in them to allow developers to improperly 
implement or operator's to improperly configure NEA systems (again IMHO).

>>

[omitted]

>>
>> Page 35, Section 7.4 Requirement PT-2
>> In addition to mutual authentication specified in PT-2,
>> the PT protocol must also be capable of supporting server
>> side only authentication as an alternative to mutual
>> authentication.  Server side only authentication is common
>> practice today in SSL VPNs. With this in mind, please
>> consider changing "mutual authentication" to "mutual and
>> server side only authentication schemes".
>
> I suggest the following change:
>
> Page 35, Section 7.4 Requirement PT-2
> "Authentictation schemes specific to the end-use models", so
> that the type of authentication - whether Server-Side,
> mutual, or circuit-level only can be done.
>

Yes, this is more of the intent I'm thinking as well.

>
>
>> Page 44, Section 9.1 Paragraph 2
>> In order to prevent privacy abuse by network owners, PA
>> request attributes must be restricted to information  known
>> apriori  rather than be open ended in order to prevent
>> dredging for artifact information. For example a PV looking
>> for a domain named "mydomain.com" should be allowed, whereas
>> requesting a domain named "*" should be disallowed by the PC.
>>
> Why? - What you suggest here means that each and every domain
> MUST be individually enumerated rather than being able to say
> "Any Supported or Defined Domain"... I can see many instances
> - especially those inside of University's and the like which
> may have hundreds of domains, all which are operated under
> the University's Security Persona...
>

Good point, although I think my comment still stands in this case.

Asking are you a member of "myuniversity.edu" isn't open ended
in my mind since this is explicitly scoped as a specific group query.

TSG: OK but in this case the system is one that already is qualified or has 
had something to point it at MYUniversity.COM. What happens when you pull a 
fresh machine from the box and plug it into the network. How is it going to 
get its basic info - like its IP address for instance, and what will setup 
the server. Seems to me that in that use-case, the hosting network would 
want the NEA assurance process to work at the DHCP level or have I missed 
something?

TSG: Other cases are where a previously known system was turned off for a 
period or removed from the local network so its NEA Certification Status is 
expired or temporarilly set aside...

My comment was to provide guidance to PC & PV developers to exercise
good judgment in maintaining user privacy in their disclosures via the
PA protocol exchanges.

TSG: arent those specific to the policy controls of the hosting org? - I 
understand the need to build a baseline but what features make that up?

>> Page 46, Section 9.2  Paragraph 2
>> I don't think Model 1 is practical or even possible in
>> some cases.  This model works only when both the PC and PV
>> have an immutable set of attributes known apriori by each.
>> However, it is common practice in existing SSL VPN s to
>> request information on the existence of files, directories,
>> processes and other artifacts that are configurable
>> assessment policy on the NEA Server. In order for model 1 to
>> potentially work, the NEA client would have to send enormous
>> amounts of information; potentially an enumeration of its
>> entire file name space, process name space, registry name space, etc.

>
> yes it would - but what is the value of the assurance it was
> to provide if the Client side didnt present these
> qualification datapoints?
>

I'm not following you on this one.  Can you provide more clarity on
this comment?

TSG: Sure - if NEA is to provide a stateful assurance of complaince with 
specific policies, then the scope of what triggers or forms those policies 
on the Device Under Test (aka a "DUT" in the Chip Test Systems world), then 
we need to be able to force both flattened and hierarchical compliance. 
Flattened compliance is a single set of tests wherein the client sends the 
querying system a notice saying it is in compliance with POLICY XYZ... and 
in the hierarchical mode the Client responds to remote test questions of 
compliance and the Statement of Compliance with the sum of those policies is 
generated by the evaluating system or machine.

TSG: Does that give you enough detail?

T

Gary 



_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Tue Sep 11 20:20:08 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVFxe-0005Sf-Jg; Tue, 11 Sep 2007 20:20:06 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVFxd-0005Qu-Pr
	for nea-confirm+ok@megatron.ietf.org; Tue, 11 Sep 2007 20:20:05 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVFxd-0005PI-D9
	for nea@ietf.org; Tue, 11 Sep 2007 20:20:05 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IVFxb-0002iM-VR
	for nea@ietf.org; Tue, 11 Sep 2007 20:20:05 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id hssnte0171se for <nea@ietf.org>;
	Tue, 11 Sep 2007 17:16:22 -0700 (envelope-from
	<gtomlinson@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9143)
	with ESMTP; Tue, 11 Sep 2007 17:16:22 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Tue, 11 Sep 2007 17:20:06 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Date: Tue, 11 Sep 2007 17:17:12 -0700
Message-ID: <4DF8F0A275770B4387764386ABEF4D7A112236@us0exb10.us.sonicwall.com>
In-Reply-To: <C30B21CC.AAD7%kaushik@cisco.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acfz5qfsPz1SyEhoRu2ib5KW8a/VZAAHAHudACY6VtA=
From: <gtomlinson@SonicWALL.com>
To: <kaushik@cisco.com>,
	<nea@ietf.org>
X-OriginalArrivalTime: 12 Sep 2007 00:20:06.0994 (UTC)
	FILETIME=[AC092B20:01C7F4D2]
X-Mlf-Version: 6.0.1.9143
X-Mlf-UniqueId: o200709120016220030587
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 24d000849df6f171c5ec1cca2ea21b82
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hello Kaushik,

Thanks for the clarifications. Please find my response inline.=20

> -----Original Message-----
> From: Kaushik Narayan [mailto:kaushik@cisco.com]=20
> Sent: Monday, September 10, 2007 4:31 PM
> To: Gary Tomlinson; nea@ietf.org
> Subject: Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
>=20
> Hi Gary,
>=20
> Thanks for your comments. Please find my response inline.
>=20
>=20
> On 9/10/07 1:10 PM, "gtomlinson@SonicWALL.com"=20
> <gtomlinson@SonicWALL.com> wrote:
>=20
>=20
> =09
> 	Page 32 Section 7.1, Requirement C-9=20
> 	At a minimum, C-9 should explicitly require UTF-8=20
> encodings be supported in strings in order to adhere to the=20
> guidance specified in RFC 2277.  =20
> =09
> 	Is there any reason to support character sets other=20
> than UTF-8?  Supporting only UTF-8 makes the protocol=20
> straightforward on interpreting string=20
> =09
> 	encodings and normalization to locale specific=20
> character sets is generally easy to implement on most OS=20
> platforms using their intrinsic libraries.
> =09
> =09
> 	<Kaushik>=20
> =09
> 	C-9 currently tries to capture i18n requirements for=20
> attributes that are both "text" and "non-text"at the PA and=20
> PB layer. We did not want to get into specific encoding standards
> 	and that's the reason for a very generically worded requirement
> =09
> 	</Kaushik>=20
> =09

<Gary>
I'm not sure how "non-text" fits into i18n?  Did you mean
"display-strings" and
"non-display-strings"?  If so, why not change the phrasing of C-9 to
"For all string
attributes, the...."? Since this is a common requirement for PA & PB,
why did you make
the requirement so generically?  Does the WG plan to provide this much
leeway to the
PA and PB protocol design teams? If so, I'll retract my UTF-8 comments
here and look to
apply them to the forthcoming PA & PB protocol drafts.
</Gary>

> =09
> 	Page 34 Section 7.2=20
> 	I noticed the former PA-8 requirement in=20
> draft-ietf-nea-requirements-03 was dropped; apparently in=20
> favor of the more generalized new C-9=20
> =09
> 	requirement in draft-ietf-nea-requirements-04.  I=20
> believe there still exists a need for a PA specific I18N=20
> string encoding requirement for non-
> =09
> 	displayable request and response attributes. Request=20
> attributes for end point artifacts such as File, Directory,=20
> Process Name, etc are used in=20
> =09
> 	common practice today in NEA equivalent deployments I=20
> am aware of.  Again, as mentioned in the comment regarding=20
> the C-9 requirement,=20
> =09
> 	UTF-8 makes things unambiguous and at a minimum must be=20
> a supported encoding.=20
> =09
> 	<Kaushik> C-9 should capture this since there could be=20
> non-displayable attributes at the PB layer as well. </Kaushik>
> =09
>=20

<Gary>
Understood.  Please disregard this comment.
</Gary>
=09
> =09
> =09
> 	Page 34, Section 7.3 Requirement PB-3=20
> 	Is there any reason in PB-3 not to specify identifiers=20
> be based upon SMI enterprise numbers?  This requirement looks=20
> much like that=20
> =09
> 	already faced by SNMP and LDAP; and we already have the=20
> IANA machinery to leverage here.=20
> =09
> =09
> 	<Kaushik>=20
> =09
> 	Are you referring to the use of SMI for  static=20
> identifiers for collectors & validators ? I think it might be=20
> a useful suggestion
> 	for a potential solution, not sure if need to talk=20
> about this in the requirements.=20
> =09
> 	</Kaushik>
> =09
>=20

<Gary>
Yes, I am referring to SMI for PV & PC identifiers.  I tend to want
tight
requirements and made the leap of incorporating an already existing IETF
standard
in here; but that may be beyond what the WG wants at this stage.  At a
minimum, I=20
do think PB-3 needs to specify identifiers must be globally unique
within the NEA=20
PB name space.
</Gary>
=09
> =09
> =09
> =09
> 	Page 35, Section 7.4 Requirement PT-2=20
> 	In addition to mutual authentication specified in PT-2,=20
> the PT protocol must also be capable of supporting server=20
> side only authentication=20
> =09
> 	as an alternative to mutual authentication.  Server=20
> side only authentication is common practice today in SSL=20
> VPNs. With this in mind, please=20
> =09
> 	consider changing "mutual authentication" to "mutual=20
> and server side only authentication schemes".=20
> =09
> =09
> 	<Kaushik>
> =09
> =09
> 	The requirement is a MUST for mutual authentication=20
> between NEA clients & NEA servers. The mutual authentication=20
> could be achieved as a result
> 	of two independent authentication transactions (with=20
> binding for certain PTs), for e.g. For password based=20
> credentials, the NEA server will be authenticated=20
> 	while a secure channel is being established (such as=20
> TLS) and NEA client will be subsequently authenticated within=20
> that secure channel.
> =09
> =09
> 	</Kaushik>=20
>=20

<Gary>
Understood on the multi-layer approach. However, this requirement is=20
beyond the scope of the PT protocol and I think a new common C-XX=20
requirement is necessary that states the NEA Client and NEA Server=20
must provide  a secure channel via mutual authentication,
confidentiality=20
and integrity through one or more of the PA, PB and PT protocols
collectively. =20
As to my comment on PT-2, I believe that still stands as it is required
for
many if not most SSL VPN TLS connections.
</Gary>
=09
> =09
> =09
> =09
> =09
> =09
> 	Page 44, Section 9.1 Paragraph 2=20
> 	In order to prevent privacy abuse by network owners, PA=20
> request attributes must be restricted to information  known=20
> apriori  rather=20
> =09
> 	than be open ended in order to prevent dredging for=20
> artifact information. For example a PV looking for a domain named=20
> 	"mydomain.com" should be allowed, whereas requesting a=20
> domain named "*" should be disallowed by the PC.=20
> =09
> 	<Kaushik>
> =09
> 	I guess you are stating a more pointed use of the=20
> mechanism stated in paragraph 2. It might be a useful to=20
> include in that section.
> =09
> 	</Kaushik>=20
> =09
>=20

<Gary>
Yes, that is correct. What I am concerned about is providing guidance
on user disclosure policies.  If the administrator of the user's machine
wants to allow open ended dredging, then so be it, but it should be
explicitly
allowed and default to denied as many user's won't possess the skills to

protect themselves properly.
</Gary>

=09
> =09
> 	Page 46, Section 9.2  Paragraph 2=20
> 	I don't think Model 1 is practical or even possible in=20
> some cases.  This model works only when both the PC and PV=20
> have an immutable=20
> =09
> 	set of attributes known apriori by each.  However, it=20
> is common practice in existing SSL VPN s to request=20
> information on the existence of files,=20
> =09
> 	directories, processes and other artifacts that are=20
> configurable assessment policy on the NEA Server. In order=20
> for model 1 to potentially work,=20
> =09
> 	the NEA client would have to send enormous amounts of=20
> information; potentially an enumeration of its entire file=20
> name space, process name space,=20
> =09
> 	registry name space, etc.=20
> =09
> 	<Kaushik>
> =09
> 	That's a fair point but Model 1 might be a useful model=20
> to assess endpoints that are not owned by the Enterprise, for=20
> e.g.  A contractor/guest system
> 	that is configured to provide a pre-set profile of=20
> attributes when challenged by a guest/partner network.
> =09
> 	</Kaushik>
> 	=20
>=20

<Gary>
Yes.  I'll retract this comment in light of the clarification
you made on Model 3.  I miss interpreted the context to construe
only one model would emerge as most feasible for the reference
architecture.
</Gary>

=09
> =09
> 	Page 46, Section 9.2 Paragraph 4=20
> 	Model 3 is used in common practice today by many of the=20
> NEA like deployments.  At a minimum, it must be supported . =20
> I can also see model 2=20
> =09
> 	being used effectively by a subset of PC/PV providers. =20
> It occurs to me, the decision to use either model 2 or model=20
> 3 is one the PC/PV provider=20
> =09
> 	makes and that the NEA protocols could consider=20
> supporting both models if necessary; but Model 3 appears to=20
> be a superset of Model 2 and as=20
> =09
> 	such so it might make sense to simply the protocol and=20
> only support model 3. My preference is to always make things=20
> simpler if possible.
> =09
> =09
> =09
> 	<Kaushik>=20
> =09
> 	Just want to point out that all the models described in=20
> section 9 are implementation considerations and not=20
> requirements for NEA protocol specification. The models=20
> explain how NEA implementations can address privacy concerns=20
> and it is upto to implementations to support one or more of=20
> these models.
> =09
> 	regards,
> 	  kaushik!
> =09
> 	</Kaushik>
> =09
>=20

<Gary>
Understood, although I don't think section 9.2 paragraph 1 clearly=20
states this.  I suggest rephrasing this paragraph to explicitly state
the NEA reference architecture and its protocols can be used with any
and
all of the 3 NEA Client to NEA Server interaction models; and that the
selection of an interaction model by a PC/PV implementer is use case=20
dependent. Once this context is established, I think it is fine to
switch=20
the context to protocol consideration guidance for the forthcoming PB
and=20
PA design teams.

Regards,
Gary
</Gary>
=09


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Tue Sep 11 22:34:57 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVI45-0001Jn-CC; Tue, 11 Sep 2007 22:34:53 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVI44-0001HL-2y
	for nea-confirm+ok@megatron.ietf.org; Tue, 11 Sep 2007 22:34:52 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVI43-0001GV-NC
	for nea@ietf.org; Tue, 11 Sep 2007 22:34:51 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IVI42-0007e4-LB
	for nea@ietf.org; Tue, 11 Sep 2007 22:34:51 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id hst7mo0171s2 for <nea@ietf.org>;
	Tue, 11 Sep 2007 19:30:38 -0700 (envelope-from
	<panandam@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9143)
	with ESMTP; Tue, 11 Sep 2007 19:30:38 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Tue, 11 Sep 2007 19:34:23 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] RE: Newcomer comments on draft-ietf-nea-requirements-04
Date: Tue, 11 Sep 2007 19:34:22 -0700
Message-ID: <D37A04F05A87044A8DCEEC0BF9A3AB1D039DC2@us0exb10.us.sonicwall.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] RE: Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acfz5qfsPz1SyEhoRu2ib5KW8a/VZAAABvKQAAsCVYcANFYU5w==
References: <C30B3CDF.AAED%kaushik@cisco.com>
From: <panandam@SonicWALL.com>
To: <nea@ietf.org>
X-OriginalArrivalTime: 12 Sep 2007 02:34:23.0113 (UTC)
	FILETIME=[6DDB5F90:01C7F4E5]
X-Mlf-Version: 6.0.1.9143
X-Mlf-UniqueId: o200709120230380047856
X-Spam-Score: 0.0 (/)
X-Scan-Signature: ce732c7d36989a1bd55104ba259c40a1
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hi Kaushik,
=20
Ah yes, SHOULD. Good, thanks!
=20
Regards,
Parvez

________________________________

From: Kaushik Narayan [mailto:kaushik@cisco.com]
Sent: Mon 9/10/2007 6:26 PM
To: Parvez Anandam; nea@ietf.org
Subject: Re: [Nea] RE: Newcomer comments on =
draft-ietf-nea-requirements-04


Hi Parvez,

Please find my reply inline.



	<snipped>
=09
	I have a few additional comments, following Gary's, that also pertain =
to draft-ietf-nea-requirements-04.
=09
	Page 35 Section 7.4
=09
	Please consider adding a new requirement for the PT protocol to support =
non-persistent network connections. This is especially important for =
mobile devices that roam or have air time charges.  This has an impact =
on reassessments triggered by the NEA server; in some cases (e.g. NAT =
traversal), the only option may be for the NEA client to periodically =
contact the NEA server to determine whether a reassessment is necessary.
=09
=09
=09
	<Kaushik>=20
=09
	I am not sure about what you mean by "persistent network connection", =
i.e. What does that mean for a PT that runs over a network access =
protocol such as 802.1x or IKEv2. In any case the requirements to =
support client & server triggered assessment/re-assessment is a SHOULD =
and not a MUST.
=09
=09
	regards,
	  kaushik
=09
=09
	</Kaushik>
=09
=09
	Page 44 Section 9.1 Paragraph 3
=09
	As Gary stated, due to security policy, network owners may wish not to =
disclose the attributes being requested. Imagine an attacker who =
restricts all posture collection in order to determine what the network =
operators are looking for.
=09
	Page 45 Section 9.1 Paragraph 6
=09
	This is a continuation of the thought in the comment above and concerns =
this item in the NEA disclosure statement: "What information is =
required". Such disclosure must be subject to  the security policy of =
the network owner. What is the intent of specificity in this guidance?  =
Does the network owner have the latitude of providing a more general =
disclosure statement?
=09
	Parvez Anandam
=09
	_____________________________________________
	From: Gary Tomlinson
	Sent: Monday, September 10, 2007 1:11 PM
	To: nea@ietf.org
	Subject: Newcomer comments on draft-ietf-nea-requirements-04
=09
	Hello NEA WG,
=09
	Several of us at SonicWALL have recently been following the NEA WG in =
general and specifically draft-ietf-nea-requirements-03 and=20
=09
	draft-ietf-nea-requirements-04.  I have the following comments =
pertaining to draft-ietf-nea-requirements-04.
=09
	Page 32 Section 7.1, Requirement C-9
=09
	At a minimum, C-9 should explicitly require UTF-8 encodings be =
supported in strings in order to adhere to the guidance specified in RFC =
2277.  =20
=09
	Is there any reason to support character sets other than UTF-8?  =
Supporting only UTF-8 makes the protocol straightforward on interpreting =
string=20
=09
	encodings and normalization to locale specific character sets is =
generally easy to implement on most OS platforms using their intrinsic =
libraries.
=09
	Page 33, Section 7.2, Requirement PA-1
=09
	The NEA model seems to imply an extensible standard attribute name =
space, but PA-1 does not incorporate this as a requirement. Nor
=09
	does PA-1 specify a requirement for the standard attribute name space =
schema to be uniquely identifiable from vendor attribute name=20
=09
	spaces.=20
=09
	Page 33, Section 7.2, Requirement PA-2
=09
	The NEA model implies a plurality of extensible vendor attribute name =
spaces and concurrent instances of each schema, but=20
=09
	PA-2 does not  incorporate this as a requirement.  Nor does PA-2 =
specify a requirement for vendor attribute name space schemas to=20
=09
	be uniquely identifiable.
=09
	Page 33, Section 7.2, Requirement PA-4
=09
	The network owner security policy may restrict giving the user complete =
visibility into why an assessment failed in order to prevent=20
=09
	exposing unnecessary information to potential attackers, especially =
prior to user authentication.  With this in mind, please consider=20
=09
	changing "This enables" to "This optionally enables the Posture Client, =
subject to disclosure policy of the NEA Server, ".
=09
	Page 33, Section 7.12 Requirement PA-5
=09
	I am little confused over the recent comments debating downgrading =
[MUST] to [SHOULD] or [MAY].  The decision to utilize end-to-end=20
=09
	cryptography is use case topology dependent.  If these use cases are =
deemed to be in scope, then doesn't it make sense for this to be a=20
=09
	[MUST] requirement and amend the verbiage to read "... MUST be =
optionally capable of providing ..."?
=09
=09
	Page 34 Section 7.2
=09
	I noticed the former PA-8 requirement in draft-ietf-nea-requirements-03 =
was dropped; apparently in favor of the more generalized new C-9=20
=09
	requirement in draft-ietf-nea-requirements-04.  I believe there still =
exists a need for a PA specific I18N string encoding requirement for =
non-
=09
	displayable request and response attributes. Request attributes for end =
point artifacts such as File, Directory, Process Name, etc are used in=20
=09
	common practice today in NEA equivalent deployments I am aware of.  =
Again, as mentioned in the comment regarding the C-9 requirement,=20
=09
	UTF-8 makes things unambiguous and at a minimum must be a supported =
encoding.
=09
	Page 34, Section 7.3 Requirement PB-3
=09
	Is there any reason in PB-3 not to specify identifiers be based upon =
SMI enterprise numbers?  This requirement looks much like that=20
=09
	already faced by SNMP and LDAP; and we already have the IANA machinery =
to leverage here.
=09
	Page 35, Section 7.4 Requirement PT-2
=09
	In addition to mutual authentication specified in PT-2, the PT protocol =
must also be capable of supporting server side only authentication=20
=09
	as an alternative to mutual authentication.  Server side only =
authentication is common practice today in SSL VPNs. With this in mind, =
please=20
=09
	consider changing "mutual authentication" to "mutual and server side =
only authentication schemes".
=09
	Page 44, Section 9.1 Paragraph 2
=09
	In order to prevent privacy abuse by network owners, PA request =
attributes must be restricted to information  known apriori  rather=20
=09
	than be open ended in order to prevent dredging for artifact =
information. For example a PV looking for a domain named=20
=09
	"mydomain.com" should be allowed, whereas requesting a domain named "*" =
should be disallowed by the PC.
=09
	Page 46, Section 9.2  Paragraph 2
=09
	I don't think Model 1 is practical or even possible in some cases.  =
This model works only when both the PC and PV have an immutable=20
=09
	set of attributes known apriori by each.  However, it is common =
practice in existing SSL VPN s to request information on the existence =
of files,=20
=09
	directories, processes and other artifacts that are configurable =
assessment policy on the NEA Server. In order for model 1 to potentially =
work,=20
=09
	the NEA client would have to send enormous amounts of information; =
potentially an enumeration of its entire file name space, process name =
space,=20
=09
	registry name space, etc.=20
=09
	Page 46, Section 9.2 Paragraph 4
=09
	Model 3 is used in common practice today by many of the NEA like =
deployments.  At a minimum, it must be supported .  I can also see model =
2=20
=09
	being used effectively by a subset of PC/PV providers.  It occurs to =
me, the decision to use either model 2 or model 3 is one the PC/PV =
provider=20
=09
	makes and that the NEA protocols could consider supporting both models =
if necessary; but Model 3 appears to be a superset of Model 2 and as=20
=09
	such so it might make sense to simply the protocol and only support =
model 3. My preference is to always make things simpler if possible.
=09
=09
	Gary Tomlinson
=09
=09
=09
________________________________

	_______________________________________________
	Nea mailing list
	Nea@ietf.org
	https://www1.ietf.org/mailman/listinfo/nea
=09





_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Wed Sep 12 00:46:56 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVK7q-0005gt-6z; Wed, 12 Sep 2007 00:46:54 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVK7p-0005go-MZ
	for nea-confirm+ok@megatron.ietf.org; Wed, 12 Sep 2007 00:46:53 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVK7p-0005gg-CV
	for nea@ietf.org; Wed, 12 Sep 2007 00:46:53 -0400
Received: from sj-iport-6.cisco.com ([171.71.176.117])
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IVK7n-00069r-VY
	for nea@ietf.org; Wed, 12 Sep 2007 00:46:53 -0400
X-IronPort-AV: E=Sophos;i="4.20,242,1186383600"; d="scan'208";a="216473331"
Received: from sj-dkim-6.cisco.com ([171.68.10.81])
	by sj-iport-6.cisco.com with ESMTP; 11 Sep 2007 21:46:51 -0700
Received: from sj-core-2.cisco.com (sj-core-2.cisco.com [171.71.177.254])
	by sj-dkim-6.cisco.com (8.12.11/8.12.11) with ESMTP id l8C4kpwj014996; 
	Tue, 11 Sep 2007 21:46:51 -0700
Received: from xbh-sjc-211.amer.cisco.com (xbh-sjc-211.cisco.com
	[171.70.151.144])
	by sj-core-2.cisco.com (8.12.10/8.12.6) with ESMTP id l8C4kpat000671;
	Wed, 12 Sep 2007 04:46:51 GMT
Received: from xmb-sjc-22d.amer.cisco.com ([128.107.191.68]) by
	xbh-sjc-211.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.1830); 
	Tue, 11 Sep 2007 21:46:51 -0700
Received: from 10.21.92.241 ([10.21.92.241]) by xmb-sjc-22d.amer.cisco.com
	([128.107.191.68]) with Microsoft Exchange Server HTTP-DAV ; 
	Wed, 12 Sep 2007 04:46:50 +0000
User-Agent: Microsoft-Entourage/11.3.6.070618
Date: Tue, 11 Sep 2007 21:46:16 -0700
Subject: Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
From: Kaushik Narayan <kaushik@cisco.com>
To: <gtomlinson@SonicWALL.com>, <nea@ietf.org>
Message-ID: <C30CBD28.AC65%kaushik@cisco.com>
Thread-Topic: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acfz5qfsPz1SyEhoRu2ib5KW8a/VZAAHAHudACY6VtAAFxHHrw==
In-Reply-To: <4DF8F0A275770B4387764386ABEF4D7A112236@us0exb10.us.sonicwall.com>
Mime-version: 1.0
Content-type: text/plain;
	charset="US-ASCII"
Content-transfer-encoding: 7bit
X-OriginalArrivalTime: 12 Sep 2007 04:46:51.0140 (UTC)
	FILETIME=[EF400C40:01C7F4F7]
DKIM-Signature: v=0.5; a=rsa-sha256; q=dns/txt; l=5681; t=1189572411;
	x=1190436411; c=relaxed/simple; s=sjdkim6002;
	h=Content-Type:From:Subject:Content-Transfer-Encoding:MIME-Version;
	d=cisco.com; i=kaushik@cisco.com;
	z=From:=20Kaushik=20Narayan=20<kaushik@cisco.com>
	|Subject:=20Re=3A=20[Nea]=20Newcomer=20comments=20on=20draft-ietf-nea-req
	uirements-04 |Sender:=20;
	bh=UqY2etvJb/kJUQvCJJ94CgMrgCyFpWoEzqMdbOR8udM=;
	b=YjbgCS7QrzlsWIZCBdBBSOCk+F26gD4WNXDxX7x8RTmueS1gp4xTAPgWSnAYJbiNn0pTVCM6
	dkEUfqACB15Et/hw2ZcXJhiSokHzR2rdFdrG3ftJbUq2PN8kxTJMXbIm;
Authentication-Results: sj-dkim-6; header.From=kaushik@cisco.com; dkim=pass (
	sig from cisco.com/sjdkim6002 verified; ); 
X-Spam-Score: -4.0 (----)
X-Scan-Signature: 311e798ce51dbeacf5cdfcc8e9fda21b
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hi Gary,

Please find my reply inline.

<snipped>

>> 
>> C-9 currently tries to capture i18n requirements for
>> attributes that are both "text" and "non-text"at the PA and
>> PB layer. We did not want to get into specific encoding standards
>> and that's the reason for a very generically worded requirement
>> 
>> </Kaushik> 
>> 
> 
> <Gary>
> I'm not sure how "non-text" fits into i18n?  Did you mean
> "display-strings" and
> "non-display-strings"?  If so, why not change the phrasing of C-9 to
> "For all string
> attributes, the...."? Since this is a common requirement for PA & PB,
> why did you make
> the requirement so generically?  Does the WG plan to provide this much
> leeway to the
> PA and PB protocol design teams? If so, I'll retract my UTF-8 comments
> here and look to
> apply them to the forthcoming PA & PB protocol drafts.
> </Gary>
> 
>> 


<Kaushik-Response>


I did mean display strings (unicode). I think your suggestion is valid and
we will augment text in requirement C-9.


</Kaushik-Response>


>> 
>> 
>> Page 34, Section 7.3 Requirement PB-3
>> Is there any reason in PB-3 not to specify identifiers
>> be based upon SMI enterprise numbers?  This requirement looks
>> much like that 
>> 
>> already faced by SNMP and LDAP; and we already have the
>> IANA machinery to leverage here.
>> 
>> 
>> <Kaushik> 
>> 
>> Are you referring to the use of SMI for  static
>> identifiers for collectors & validators ? I think it might be
>> a useful suggestion
>> for a potential solution, not sure if need to talk
>> about this in the requirements.
>> 
>> </Kaushik>
>> 
>> 
> 
> <Gary>
> Yes, I am referring to SMI for PV & PC identifiers.  I tend to want
> tight
> requirements and made the leap of incorporating an already existing IETF
> standard
> in here; but that may be beyond what the WG wants at this stage.  At a
> minimum, I 
> do think PB-3 needs to specify identifiers must be globally unique
> within the NEA 
> PB name space.
> </Gary>
> 

<Kaushik-Response>

I agree about the suggestion of unique identifier, we will add text to PB-3
to specify that identifiers are required to be unique.

</Kaushik-Response>





>> 
>> 
>> 
>> Page 35, Section 7.4 Requirement PT-2
>> In addition to mutual authentication specified in PT-2,
>> the PT protocol must also be capable of supporting server
>> side only authentication
>> 
>> as an alternative to mutual authentication.  Server
>> side only authentication is common practice today in SSL
>> VPNs. With this in mind, please
>> 
>> consider changing "mutual authentication" to "mutual
>> and server side only authentication schemes".
>> 
>> 
>> <Kaushik>
>> 
>> 
>> The requirement is a MUST for mutual authentication
>> between NEA clients & NEA servers. The mutual authentication
>> could be achieved as a result
>> of two independent authentication transactions (with
>> binding for certain PTs), for e.g. For password based
>> credentials, the NEA server will be authenticated
>> while a secure channel is being established (such as
>> TLS) and NEA client will be subsequently authenticated within
>> that secure channel.
>> 
>> 
>> </Kaushik> 
>> 
> 
> <Gary>
> Understood on the multi-layer approach. However, this requirement is
> beyond the scope of the PT protocol and I think a new common C-XX
> requirement is necessary that states the NEA Client and NEA Server
> must provide  a secure channel via mutual authentication,
> confidentiality 
> and integrity through one or more of the PA, PB and PT protocols
> collectively.  
> As to my comment on PT-2, I believe that still stands as it is required
> for
> many if not most SSL VPN TLS connections.
> </Gary>
> 

<Kaushik-Response>

The common requirement would be redundant because we already have individual
requirements in PA, PB and PT for authentication, integrity & privacy
protection. The reason for individual requirements is because priority for
the PA, PB and PT layer requirements are different.


Re-wording PT-2 to change "mutual authentication" to "mutual and server side
only authentication schemes" would mean that it would acceptable have
server-only authentication for PT which does not seem sufficient.


</Kaushik-Response>


>> 
>> <Kaushik> 
>> 
>> Just want to point out that all the models described in
>> section 9 are implementation considerations and not
>> requirements for NEA protocol specification. The models
>> explain how NEA implementations can address privacy concerns
>> and it is upto to implementations to support one or more of
>> these models.
>> 
>> regards,
>>  kaushik!
>> 
>> </Kaushik>
>> 
>> 
> 
> <Gary>
> Understood, although I don't think section 9.2 paragraph 1 clearly
> states this.  I suggest rephrasing this paragraph to explicitly state
> the NEA reference architecture and its protocols can be used with any
> and
> all of the 3 NEA Client to NEA Server interaction models; and that the
> selection of an interaction model by a PC/PV implementer is use case
> dependent. Once this context is established, I think it is fine to
> switch 
> the context to protocol consideration guidance for the forthcoming PB
> and 
> PA design teams.
> 


<Kaushik-Response>

The second sentence in Section 9 states

" This section will discuss a few of the potential privacy concerns raised
by the deployment of this technology and offer some guidance to
implementers."

Does that seem adequate or do you think we need additional text in section
9.2 


regards,
 kaushik

</Kaushik-Response>



_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Wed Sep 12 11:55:38 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVUYz-0008Ln-45; Wed, 12 Sep 2007 11:55:37 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVUYy-0008HB-4V
	for nea-confirm+ok@megatron.ietf.org; Wed, 12 Sep 2007 11:55:36 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVUYx-0008F3-Nn
	for nea@ietf.org; Wed, 12 Sep 2007 11:55:35 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IVUYw-000855-Mi
	for nea@ietf.org; Wed, 12 Sep 2007 11:55:35 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id ht05h00171s3 for <nea@ietf.org>;
	Wed, 12 Sep 2007 08:51:44 -0700 (envelope-from
	<gtomlinson@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9143)
	with ESMTP; Wed, 12 Sep 2007 08:51:44 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Wed, 12 Sep 2007 08:55:33 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Date: Wed, 12 Sep 2007 08:55:21 -0700
Message-ID: <4DF8F0A275770B4387764386ABEF4D7A100ACF@us0exb10.us.sonicwall.com>
In-Reply-To: <01a101c7f4af$8d5bb440$174f7d40@home.glassey.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acf0r4xdUk2TlaW1TaKOYNDP09/nsgAIz3JQ
From: <gtomlinson@SonicWALL.com>
To: <tglassey@earthlink.net>,
	<nea@ietf.org>
X-OriginalArrivalTime: 12 Sep 2007 15:55:33.0621 (UTC)
	FILETIME=[5A1CCA50:01C7F555]
X-Mlf-Version: 6.0.1.9143
X-Mlf-UniqueId: o200709121551430072641
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 4b66a1e94d7d92973ece9e5da449ff80
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org


 Todd,

Please find my response inline.

[snippet]

>=20
> >>
> > Page 33, Section 7.2, Requirement PA-2
> > The NEA model implies a plurality of extensible vendor
> > attribute name spaces and concurrent instances of each
> > schema, but PA-2 does not  incorporate this as a requirement.
> >  Nor does PA-2 specify a requirement for vendor attribute
> > name space schemas to be uniquely identifiable.
> >
> > Again - a complete set of use models would eliminate this.
>=20
> I'm not sure I can describe a complete set of use cases;=20
> although again I=20
> assumed the same use case as in PA-1 and thus the need to=20
> distinguish the=20
> vendor specific  attributes from each other and from the=20
> standard attributes=20
> as well.
>=20
> TSG: Gary - You nor anyone else here as individuals needs to=20
> docuemtn all of=20
> the use models or cases - the group collectively does need to=20
> document say=20
> about 5 or more solid use cases and the controls for each one.
>=20

<Gary>
It is always tough to decide when a sufficient number of use cases are
present to establish a baseline context.  I am familiar with assessment
used in conjunction with real-time access control enforcement and thus
have been able to mostly understand the requirements so far; however
one could argue a use case for remote access VPN usage might be useful.
I am only marginally familiar with assessment for compliance auditing
which seems to be your area of expertise and thus not in a position to
judge the utility of the current set of use cases for this purpose.
I'll
have defer to the WG at large on this one.
</Gary>

> TSG: It would also would be a good idea to map COBIT and=20
> ISO17799 use cases=20
> on top of the protocol as well so it is easier to absorb into Audit=20
> Regulatred environments.
>=20

<Gary>
Again, I'm not sure if this is in the current WG charter scope.  If it
is, I would imagine you could provide assistance in this mapping.
</Gary>

[snippet]

>=20
> TSG: Gary - How will that fit into the control framework of=20
> SNMP then? Can=20
> we use SNMP to talk to the NEA Agent PC/PV systems?

<Gary>
Sorry I wasn't clearer in my comments.  I'm not suggesting NEA have any
SNMP MIBs, but rather consider using the SMI numbering scheme already=20
employed in SNMP for PC & PV identifiers and possibly consider using it
to
enumerate attributes as well. But as Kaushik correctly pointed out, this

is beyond the scope of the requirements specification and is something
to=20
consider for the PB protocol work. For those interested, I'd see this=20
being done where the standard attribute name space would be orthogonal=20
to mib2 as an OID and vendor specific PC/PV would be rooted off of their

own enterprise OIDs.
</Gary>

[snippet]

> >>
> >> Page 33, Section 7.2, Requirement PA-4
> >> The network owner security policy may restrict giving
> >> the user complete visibility into why an assessment failed in
> >> order to prevent exposing unnecessary information to
> >> potential attackers, especially prior to user authentication.
> >>
> > Here then is a specific use in which the system provides
> > hierarchical error or failure messages with varying detail?
>=20
> Yes, although I think the reasons are outside the scope of=20
> the  protocol=20
> itself.  I'm not sure how to phrase the guidance here =20
> regarding how to=20
> provide useful information to the user as it is  at the=20
> discretion of the=20
> network operator's policy.
>=20
> TSG: Then perhaps... the scope of the protocol definition may=20
> need to be=20
> adjusted then since these control systems are key to a=20
> successful deployment=20
> of NEA IMHO.

<Gary>
Clearly the protocol must provide a complete state graph for error and
remediation transitions. We'll have to cross that bridge once the WG
gets
to doing the PA protocol work.
</Gary>

[snippet]

>=20
> TSG: This is one of the things we need to complete. The=20
> Inferring of Use=20
> Models with NEA is a no-no IMHO. The uses of NEA need to be=20
> tight and very=20
> very secure, meaning that they are well defined and=20
> controlled so there is=20
> little if any wiggle room in them to allow developers to improperly=20
> implement or operator's to improperly configure NEA systems=20
> (again IMHO).
>=20

<Gary>
As I previously stated, at some point we'll have sufficient use cases to
describe the model. The question is, do we now?  I'm not sure we can
ever=20
completely make this a closed system though as vendor PA name spaces are

under the control of 3rd parties. We can only provide guidance and let
the
market sort things out here for poor implementations.
</Gary>

[snippet]

> Good point, although I think my comment still stands in this case.
>=20
> Asking are you a member of "myuniversity.edu" isn't open ended
> in my mind since this is explicitly scoped as a specific group query.
>=20
> TSG: OK but in this case the system is one that already is=20
> qualified or has=20
> had something to point it at MYUniversity.COM. What happens=20
> when you pull a=20
> fresh machine from the box and plug it into the network. How=20
> is it going to=20
> get its basic info - like its IP address for instance, and=20
> what will setup=20
> the server. Seems to me that in that use-case, the hosting=20
> network would=20
> want the NEA assurance process to work at the DHCP level or=20
> have I missed=20
> something?
>=20

<Gary>
A classic way, and one that is used prevalently today starts the user
on a quarantined network (i.e. VLAN segmented) to interact with their
DHCP server.  Once their network layer config is provisioned, the NEA
assessment begins and if they are compliant, their end point is mapped
onto the trusted network.  This is of course a LAN example.  Hope that=20
helps?  Use case section 6.1.1 glossed over these important details
between steps 1 and 2.
</Gary>

>=20
> My comment was to provide guidance to PC & PV developers to exercise
> good judgment in maintaining user privacy in their disclosures via the
> PA protocol exchanges.
>=20
> TSG: arent those specific to the policy controls of the=20
> hosting org? - I=20
> understand the need to build a baseline but what features=20
> make that up?

<Gary>
Please see my comments dated 9/11 to Kaushik for more information.
</Gary>

[snippet]

>=20
> >
> > yes it would - but what is the value of the assurance it was
> > to provide if the Client side didnt present these
> > qualification datapoints?
> >
>=20
> I'm not following you on this one.  Can you provide more clarity on
> this comment?
>=20
> TSG: Sure - if NEA is to provide a stateful assurance of=20
> complaince with=20
> specific policies, then the scope of what triggers or forms=20
> those policies=20
> on the Device Under Test (aka a "DUT" in the Chip Test=20
> Systems world), then=20
> we need to be able to force both flattened and hierarchical=20
> compliance.=20
> Flattened compliance is a single set of tests wherein the=20
> client sends the=20
> querying system a notice saying it is in compliance with=20
> POLICY XYZ... and=20
> in the hierarchical mode the Client responds to remote test=20
> questions of=20
> compliance and the Statement of Compliance with the sum of=20
> those policies is=20
> generated by the evaluating system or machine.
>=20
> TSG: Does that give you enough detail?

<Gary>
Yes.  And now that Kaushik clarified my understanding that NEA
isn't standardizing an interaction model, I don't have issues
with model 1.

Regards
Gary

</Gary>


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Wed Sep 12 12:21:43 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVUyE-00035T-TJ; Wed, 12 Sep 2007 12:21:42 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVUyD-00034S-Oi
	for nea-confirm+ok@megatron.ietf.org; Wed, 12 Sep 2007 12:21:41 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVUyD-00034K-D6
	for nea@ietf.org; Wed, 12 Sep 2007 12:21:41 -0400
Received: from elasmtp-banded.atl.sa.earthlink.net ([209.86.89.70])
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IVUyC-0000ml-FE
	for nea@ietf.org; Wed, 12 Sep 2007 12:21:41 -0400
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=earthlink.net;
	b=j1YllJfrrNtq5qb2QJGehXS4Jxih8Ab8YlVA5M8QOLo5ZzzlO/Qjh+5p0UkYSQ2S;
	h=Received:Message-ID:From:To:References:Subject:Date:MIME-Version:Content-Type:Content-Transfer-Encoding:X-Priority:X-MSMail-Priority:X-Mailer:X-MimeOLE:X-ELNK-Trace:X-Originating-IP;
Received: from [64.125.79.23] (helo=gw)
	by elasmtp-banded.atl.sa.earthlink.net with asmtp (Exim 4.34)
	id 1IVUyB-0001Ef-PF; Wed, 12 Sep 2007 12:21:39 -0400
Message-ID: <002a01c7f559$06f6f1d0$174f7d40@home.glassey.com>
From: "todd glassey" <tglassey@earthlink.net>
To: <gtomlinson@SonicWALL.com>,
	<nea@ietf.org>
References: <4DF8F0A275770B4387764386ABEF4D7A100ACF@us0exb10.us.sonicwall.com>
Subject: Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Date: Wed, 12 Sep 2007 09:21:49 -0700
MIME-Version: 1.0
Content-Type: text/plain; format=flowed; charset="iso-8859-1";
	reply-type=original
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2900.3138
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.3138
X-ELNK-Trace: 01b7a7e171bdf5911aa676d7e74259b7b3291a7d08dfec7912f1178a290560aa18817002704d3da2350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c
X-Originating-IP: 64.125.79.23
X-Spam-Score: 0.1 (/)
X-Scan-Signature: dd7e0c3fd18d19cffdd4de99a114001d
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org


----- Original Message ----- 
From: <gtomlinson@SonicWALL.com>
To: <tglassey@earthlink.net>; <nea@ietf.org>
Sent: Wednesday, September 12, 2007 8:55 AM
Subject: RE: [Nea] Newcomer comments on draft-ietf-nea-requirements-04



 Todd,

Please find my response inline.

[snippet]

>
> >>
> > Page 33, Section 7.2, Requirement PA-2
> > The NEA model implies a plurality of extensible vendor
> > attribute name spaces and concurrent instances of each
> > schema, but PA-2 does not  incorporate this as a requirement.
> >  Nor does PA-2 specify a requirement for vendor attribute
> > name space schemas to be uniquely identifiable.
> >
> > Again - a complete set of use models would eliminate this.
>
> I'm not sure I can describe a complete set of use cases;
> although again I
> assumed the same use case as in PA-1 and thus the need to
> distinguish the
> vendor specific  attributes from each other and from the
> standard attributes
> as well.
>
> TSG: Gary - You nor anyone else here as individuals needs to
> docuemtn all of
> the use models or cases - the group collectively does need to
> document say
> about 5 or more solid use cases and the controls for each one.
>

<Gary>
It is always tough to decide when a sufficient number of use cases are
present to establish a baseline context.  I am familiar with assessment
used in conjunction with real-time access control enforcement and thus
have been able to mostly understand the requirements so far; however
one could argue a use case for remote access VPN usage might be useful.


TSG:  I agree. VPN for remote access will be a critical value-add for the 
Audit Community.

I am only marginally familiar with assessment for compliance auditing
which seems to be your area of expertise and thus not in a position to
judge the utility of the current set of use cases for this purpose.
I'll
have defer to the WG at large on this one.

TSG:  My concern is that the Auditor's ***will*** control whether NEA is 
rolled out in arena's they control and so its important to either get their 
buy-in conceptually early on, or to answer all of the major questions they 
would have in a manner that doesnt make it critically expensive to 
implement.

</Gary>

> TSG: It would also would be a good idea to map COBIT and
> ISO17799 use cases
> on top of the protocol as well so it is easier to absorb into Audit
> Regulatred environments.
>

<Gary>
Again, I'm not sure if this is in the current WG charter scope.  If it
is, I would imagine you could provide assistance in this mapping.
</Gary>

TSG: I dont think its in the defined charter - but as a matter of what IT 
operations contain these days, there is a formal requirement for regulatory 
compliance and technologies that make that easier are winners here and the 
rest - well... you get the picture I am sure.

[snippet]

>
> TSG: Gary - How will that fit into the control framework of
> SNMP then? Can
> we use SNMP to talk to the NEA Agent PC/PV systems?

<Gary>
Sorry I wasn't clearer in my comments.  I'm not suggesting NEA have any
SNMP MIBs, but rather consider using the SMI numbering scheme already
employed in SNMP for PC & PV identifiers and possibly consider using it
to
enumerate attributes as well. But as Kaushik correctly pointed out, this

is beyond the scope of the requirements specification and is something
to
consider for the PB protocol work. For those interested, I'd see this
being done where the standard attribute name space would be orthogonal
to mib2 as an OID and vendor specific PC/PV would be rooted off of their

own enterprise OIDs.
</Gary>

[snippet]

> >>
> >> Page 33, Section 7.2, Requirement PA-4
> >> The network owner security policy may restrict giving
> >> the user complete visibility into why an assessment failed in
> >> order to prevent exposing unnecessary information to
> >> potential attackers, especially prior to user authentication.
> >>
> > Here then is a specific use in which the system provides
> > hierarchical error or failure messages with varying detail?
>
> Yes, although I think the reasons are outside the scope of
> the  protocol
> itself.  I'm not sure how to phrase the guidance here
> regarding how to
> provide useful information to the user as it is  at the
> discretion of the
> network operator's policy.
>
> TSG: Then perhaps... the scope of the protocol definition may
> need to be
> adjusted then since these control systems are key to a
> successful deployment
> of NEA IMHO.

<Gary>
Clearly the protocol must provide a complete state graph for error and
remediation transitions. We'll have to cross that bridge once the WG
gets
to doing the PA protocol work.
</Gary>

[snippet]

>
> TSG: This is one of the things we need to complete. The
> Inferring of Use
> Models with NEA is a no-no IMHO. The uses of NEA need to be
> tight and very
> very secure, meaning that they are well defined and
> controlled so there is
> little if any wiggle room in them to allow developers to improperly
> implement or operator's to improperly configure NEA systems
> (again IMHO).
>

<Gary>
As I previously stated, at some point we'll have sufficient use cases to
describe the model. The question is, do we now?  I'm not sure we can
ever
completely make this a closed system though as vendor PA name spaces are

under the control of 3rd parties. We can only provide guidance and let
the
market sort things out here for poor implementations.
</Gary>

[snippet]

> Good point, although I think my comment still stands in this case.
>
> Asking are you a member of "myuniversity.edu" isn't open ended
> in my mind since this is explicitly scoped as a specific group query.
>
> TSG: OK but in this case the system is one that already is
> qualified or has
> had something to point it at MYUniversity.COM. What happens
> when you pull a
> fresh machine from the box and plug it into the network. How
> is it going to
> get its basic info - like its IP address for instance, and
> what will setup
> the server. Seems to me that in that use-case, the hosting
> network would
> want the NEA assurance process to work at the DHCP level or
> have I missed
> something?
>

<Gary>
A classic way, and one that is used prevalently today starts the user
on a quarantined network (i.e. VLAN segmented) to interact with their
DHCP server.  Once their network layer config is provisioned, the NEA
assessment begins and if they are compliant, their end point is mapped
onto the trusted network.  This is of course a LAN example.  Hope that
helps?  Use case section 6.1.1 glossed over these important details
between steps 1 and 2.
</Gary>

>
> My comment was to provide guidance to PC & PV developers to exercise
> good judgment in maintaining user privacy in their disclosures via the
> PA protocol exchanges.
>
> TSG: arent those specific to the policy controls of the
> hosting org? - I
> understand the need to build a baseline but what features
> make that up?

<Gary>
Please see my comments dated 9/11 to Kaushik for more information.
</Gary>

[snippet]

>
> >
> > yes it would - but what is the value of the assurance it was
> > to provide if the Client side didnt present these
> > qualification datapoints?
> >
>
> I'm not following you on this one.  Can you provide more clarity on
> this comment?
>
> TSG: Sure - if NEA is to provide a stateful assurance of
> complaince with
> specific policies, then the scope of what triggers or forms
> those policies
> on the Device Under Test (aka a "DUT" in the Chip Test
> Systems world), then
> we need to be able to force both flattened and hierarchical
> compliance.
> Flattened compliance is a single set of tests wherein the
> client sends the
> querying system a notice saying it is in compliance with
> POLICY XYZ... and
> in the hierarchical mode the Client responds to remote test
> questions of
> compliance and the Statement of Compliance with the sum of
> those policies is
> generated by the evaluating system or machine.
>
> TSG: Does that give you enough detail?

<Gary>
Yes.  And now that Kaushik clarified my understanding that NEA
isn't standardizing an interaction model, I don't have issues
with model 1.

Regards
Gary

</Gary> 



_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Wed Sep 12 12:48:59 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVVOc-0006kK-0V; Wed, 12 Sep 2007 12:48:58 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVVOZ-0006kD-Hh
	for nea-confirm+ok@megatron.ietf.org; Wed, 12 Sep 2007 12:48:55 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVVOZ-0006k5-81
	for nea@ietf.org; Wed, 12 Sep 2007 12:48:55 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IVVOX-0004Hm-V4
	for nea@ietf.org; Wed, 12 Sep 2007 12:48:55 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id ht0bou0171sq for <nea@ietf.org>;
	Wed, 12 Sep 2007 09:45:03 -0700 (envelope-from
	<panandam@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9143)
	with ESMTP; Wed, 12 Sep 2007 09:45:02 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Wed, 12 Sep 2007 09:48:52 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Date: Wed, 12 Sep 2007 09:48:50 -0700
Message-ID: <D37A04F05A87044A8DCEEC0BF9A3AB1D07D0BE@us0exb10.us.sonicwall.com>
In-Reply-To: <C30CBD28.AC65%kaushik@cisco.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acfz5qfsPz1SyEhoRu2ib5KW8a/VZAAHAHudACY6VtAAFxHHrwAYXsJQ
References: <4DF8F0A275770B4387764386ABEF4D7A112236@us0exb10.us.sonicwall.com>
	<C30CBD28.AC65%kaushik@cisco.com>
From: <panandam@SonicWALL.com>
To: <nea@ietf.org>
X-OriginalArrivalTime: 12 Sep 2007 16:48:52.0504 (UTC)
	FILETIME=[CCCB9980:01C7F55C]
X-Mlf-Version: 6.0.1.9143
X-Mlf-UniqueId: o200709121645020085406
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 34d35111647d654d033d58d318c0d21a
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hi Kausik,

[PA] One comment inline...

Regards,
Parvez

-----Original Message-----
From: Kaushik Narayan [mailto:kaushik@cisco.com]=20
Sent: Tuesday, September 11, 2007 9:46 PM
To: Gary Tomlinson; nea@ietf.org
Subject: Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04

Hi Gary,

Please find my reply inline.

<snipped>

>>=20
>>=20
>>=20
>> Page 35, Section 7.4 Requirement PT-2
>> In addition to mutual authentication specified in PT-2,
>> the PT protocol must also be capable of supporting server
>> side only authentication
>>=20
>> as an alternative to mutual authentication.  Server
>> side only authentication is common practice today in SSL
>> VPNs. With this in mind, please
>>=20
>> consider changing "mutual authentication" to "mutual
>> and server side only authentication schemes".
>>=20
>>=20
>> <Kaushik>
>>=20
>>=20
>> The requirement is a MUST for mutual authentication
>> between NEA clients & NEA servers. The mutual authentication
>> could be achieved as a result
>> of two independent authentication transactions (with
>> binding for certain PTs), for e.g. For password based
>> credentials, the NEA server will be authenticated
>> while a secure channel is being established (such as
>> TLS) and NEA client will be subsequently authenticated within
>> that secure channel.
>>=20
>>=20
>> </Kaushik>=20
>>=20
>=20
> <Gary>
> Understood on the multi-layer approach. However, this requirement is
> beyond the scope of the PT protocol and I think a new common C-XX
> requirement is necessary that states the NEA Client and NEA Server
> must provide  a secure channel via mutual authentication,
> confidentiality=20
> and integrity through one or more of the PA, PB and PT protocols
> collectively. =20
> As to my comment on PT-2, I believe that still stands as it is
required
> for
> many if not most SSL VPN TLS connections.
> </Gary>
>=20

<Kaushik-Response>

The common requirement would be redundant because we already have
individual
requirements in PA, PB and PT for authentication, integrity & privacy
protection. The reason for individual requirements is because priority
for
the PA, PB and PT layer requirements are different.


Re-wording PT-2 to change "mutual authentication" to "mutual and server
side
only authentication schemes" would mean that it would acceptable have
server-only authentication for PT which does not seem sufficient.


</Kaushik-Response>


[PA] The example you give above is a good one: 1) the NEA server is
first authenticated while a secure channel is being established and 2)
The NEA client subsequently authenticates within that secure channel. In
this example, I would have thought that 1) is part of the PT protocol
whereas 2) would be part of a higher-level protocol such as PB, in which
case the mutual authentication does not happen exclusively within PT. Am
I incorrect in this assumption? Thanks for clarifying...
 =20



_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Wed Sep 12 12:52:39 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVVSB-0003Gd-Bk; Wed, 12 Sep 2007 12:52:39 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVVSA-0003G0-6F
	for nea-confirm+ok@megatron.ietf.org; Wed, 12 Sep 2007 12:52:38 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVVS9-0003Fq-Ra
	for nea@ietf.org; Wed, 12 Sep 2007 12:52:37 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IVVS9-0001fP-C7
	for nea@ietf.org; Wed, 12 Sep 2007 12:52:37 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id ht0c6u0171s3 for <nea@ietf.org>;
	Wed, 12 Sep 2007 09:48:46 -0700 (envelope-from
	<panandam@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9143)
	with ESMTP; Wed, 12 Sep 2007 09:48:46 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Wed, 12 Sep 2007 09:52:36 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Date: Wed, 12 Sep 2007 09:52:35 -0700
Message-ID: <D37A04F05A87044A8DCEEC0BF9A3AB1D07D0BF@us0exb10.us.sonicwall.com>
In-Reply-To: <002a01c7f559$06f6f1d0$174f7d40@home.glassey.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acf1WQTEFu+xyScISb2o7/E7sQEEMAAA8png
References: <4DF8F0A275770B4387764386ABEF4D7A100ACF@us0exb10.us.sonicwall.com>
	<002a01c7f559$06f6f1d0$174f7d40@home.glassey.com>
From: <panandam@SonicWALL.com>
To: <nea@ietf.org>
X-OriginalArrivalTime: 12 Sep 2007 16:52:36.0933 (UTC)
	FILETIME=[5290BF50:01C7F55D]
X-Mlf-Version: 6.0.1.9143
X-Mlf-UniqueId: o200709121648460086367
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 41c17b4b16d1eedaa8395c26e9a251c4
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hi Todd,

[PA] One comment inline...

Regards,
Parvez

-----Original Message-----
From: todd glassey [mailto:tglassey@earthlink.net]=20
Sent: Wednesday, September 12, 2007 9:22 AM
To: Gary Tomlinson; nea@ietf.org
Subject: Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04


----- Original Message -----=20
From: <gtomlinson@SonicWALL.com>
To: <tglassey@earthlink.net>; <nea@ietf.org>
Sent: Wednesday, September 12, 2007 8:55 AM
Subject: RE: [Nea] Newcomer comments on draft-ietf-nea-requirements-04



 Todd,

Please find my response inline.

[snippet]

>
> >>
> > Page 33, Section 7.2, Requirement PA-2
> > The NEA model implies a plurality of extensible vendor
> > attribute name spaces and concurrent instances of each
> > schema, but PA-2 does not  incorporate this as a requirement.
> >  Nor does PA-2 specify a requirement for vendor attribute
> > name space schemas to be uniquely identifiable.
> >
> > Again - a complete set of use models would eliminate this.
>
> I'm not sure I can describe a complete set of use cases;
> although again I
> assumed the same use case as in PA-1 and thus the need to
> distinguish the
> vendor specific  attributes from each other and from the
> standard attributes
> as well.
>
> TSG: Gary - You nor anyone else here as individuals needs to
> docuemtn all of
> the use models or cases - the group collectively does need to
> document say
> about 5 or more solid use cases and the controls for each one.
>

<Gary>
It is always tough to decide when a sufficient number of use cases are
present to establish a baseline context.  I am familiar with assessment
used in conjunction with real-time access control enforcement and thus
have been able to mostly understand the requirements so far; however
one could argue a use case for remote access VPN usage might be useful.


TSG:  I agree. VPN for remote access will be a critical value-add for
the=20
Audit Community.


[PA] Gary and I would be happy to write a TLS VPN remote access use
case, if the WG wishes.


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Wed Sep 12 13:48:20 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVWK2-00059C-VH; Wed, 12 Sep 2007 13:48:18 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVWK1-00058z-GC
	for nea-confirm+ok@megatron.ietf.org; Wed, 12 Sep 2007 13:48:17 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVWK1-00058q-6I
	for nea@ietf.org; Wed, 12 Sep 2007 13:48:17 -0400
Received: from elasmtp-dupuy.atl.sa.earthlink.net ([209.86.89.62])
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IVWJz-0005LW-Ra
	for nea@ietf.org; Wed, 12 Sep 2007 13:48:17 -0400
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=earthlink.net;
	b=tZ0TFJRGKrjc/1fIhzMUzETPBeZ2Zu2K53JUppEgsa3Q9aKBgczJSC4knx0/5DHJ;
	h=Received:Message-ID:From:To:References:Subject:Date:MIME-Version:Content-Type:Content-Transfer-Encoding:X-Priority:X-MSMail-Priority:X-Mailer:X-MimeOLE:X-ELNK-Trace:X-Originating-IP;
Received: from [64.125.79.23] (helo=gw)
	by elasmtp-dupuy.atl.sa.earthlink.net with asmtp (Exim 4.34)
	id 1IVWJl-0000Kk-85; Wed, 12 Sep 2007 13:48:01 -0400
Message-ID: <006201c7f565$176a8f70$174f7d40@home.glassey.com>
From: "todd glassey" <tglassey@earthlink.net>
To: <panandam@SonicWALL.com>,
	<nea@ietf.org>
References: <4DF8F0A275770B4387764386ABEF4D7A100ACF@us0exb10.us.sonicwall.com><002a01c7f559$06f6f1d0$174f7d40@home.glassey.com>
	<D37A04F05A87044A8DCEEC0BF9A3AB1D07D0BF@us0exb10.us.sonicwall.com>
Subject: Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Date: Wed, 12 Sep 2007 10:47:20 -0700
MIME-Version: 1.0
Content-Type: text/plain; format=flowed; charset="iso-8859-1";
	reply-type=original
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2900.3138
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.3138
X-ELNK-Trace: 01b7a7e171bdf5911aa676d7e74259b7b3291a7d08dfec7982d5c1daeb934f562c95b832198bfe29350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c
X-Originating-IP: 64.125.79.23
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 14582b0692e7f70ce7111d04db3781c8
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org


[PA] Gary and I would be happy to write a TLS VPN remote access use
case, if the WG wishes.

TSG: PLEASE DO! - I will work on the COBITization of the control 
descriptions and we can produce a COBIT compliance document for the use of 
NEA. We can then literally just put ISACA on notice that we are designing a 
protocol specifically with COBIT compliance built in from the gate, and they 
will probably be pretty happy about that I would think.

Todd


----- Original Message ----- 
From: <panandam@SonicWALL.com>
To: <nea@ietf.org>
Sent: Wednesday, September 12, 2007 9:52 AM
Subject: RE: [Nea] Newcomer comments on draft-ietf-nea-requirements-04


Hi Todd,

[PA] One comment inline...

Regards,
Parvez

-----Original Message-----
From: todd glassey [mailto:tglassey@earthlink.net]
Sent: Wednesday, September 12, 2007 9:22 AM
To: Gary Tomlinson; nea@ietf.org
Subject: Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04


----- Original Message ----- 
From: <gtomlinson@SonicWALL.com>
To: <tglassey@earthlink.net>; <nea@ietf.org>
Sent: Wednesday, September 12, 2007 8:55 AM
Subject: RE: [Nea] Newcomer comments on draft-ietf-nea-requirements-04



 Todd,

Please find my response inline.

[snippet]

>
> >>
> > Page 33, Section 7.2, Requirement PA-2
> > The NEA model implies a plurality of extensible vendor
> > attribute name spaces and concurrent instances of each
> > schema, but PA-2 does not  incorporate this as a requirement.
> >  Nor does PA-2 specify a requirement for vendor attribute
> > name space schemas to be uniquely identifiable.
> >
> > Again - a complete set of use models would eliminate this.
>
> I'm not sure I can describe a complete set of use cases;
> although again I
> assumed the same use case as in PA-1 and thus the need to
> distinguish the
> vendor specific  attributes from each other and from the
> standard attributes
> as well.
>
> TSG: Gary - You nor anyone else here as individuals needs to
> docuemtn all of
> the use models or cases - the group collectively does need to
> document say
> about 5 or more solid use cases and the controls for each one.
>

<Gary>
It is always tough to decide when a sufficient number of use cases are
present to establish a baseline context.  I am familiar with assessment
used in conjunction with real-time access control enforcement and thus
have been able to mostly understand the requirements so far; however
one could argue a use case for remote access VPN usage might be useful.


TSG:  I agree. VPN for remote access will be a critical value-add for
the
Audit Community.


[PA] Gary and I would be happy to write a TLS VPN remote access use
case, if the WG wishes.


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea 



_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Wed Sep 12 19:47:33 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVbvf-0004SW-6v; Wed, 12 Sep 2007 19:47:31 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVbve-0004SQ-Hw
	for nea-confirm+ok@megatron.ietf.org; Wed, 12 Sep 2007 19:47:30 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVbve-0004SI-6T
	for nea@ietf.org; Wed, 12 Sep 2007 19:47:30 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IVbvd-0002YQ-DR
	for nea@ietf.org; Wed, 12 Sep 2007 19:47:30 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id ht1sqc0171ss for <nea@ietf.org>;
	Wed, 12 Sep 2007 16:43:33 -0700 (envelope-from
	<gtomlinson@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9143)
	with ESMTP; Wed, 12 Sep 2007 16:43:30 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Wed, 12 Sep 2007 16:47:25 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Date: Wed, 12 Sep 2007 16:47:12 -0700
Message-ID: <4DF8F0A275770B4387764386ABEF4D7A100B4C@us0exb10.us.sonicwall.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: RE: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acf1lz1lbPJI7lC5RgKSgOOSpelueQ==
From: <gtomlinson@SonicWALL.com>
To: <kaushik@cisco.com>,
	<nea@ietf.org>
X-OriginalArrivalTime: 12 Sep 2007 23:47:25.0239 (UTC)
	FILETIME=[4526DC70:01C7F597]
X-Mlf-Version: 6.0.1.9143
X-Mlf-UniqueId: o200709122343300057311
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 25eb6223a37c19d53ede858176b14339
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hello Kaushik,

My please find my response inline.

[snipped]

>=20
>=20
>=20
> >>=20
> >>=20
> >>=20
> >> Page 35, Section 7.4 Requirement PT-2 In addition to mutual=20
> >> authentication specified in PT-2, the PT protocol must also be=20
> >> capable of supporting server side only authentication
> >>=20
> >> as an alternative to mutual authentication.  Server side only=20
> >> authentication is common practice today in SSL VPNs. With this in=20
> >> mind, please
> >>=20
> >> consider changing "mutual authentication" to "mutual and
> server side
> >> only authentication schemes".
> >>=20
> >>=20
> >> <Kaushik>
> >>=20
> >>=20
> >> The requirement is a MUST for mutual authentication between NEA=20
> >> clients & NEA servers. The mutual authentication could be
> achieved as
> >> a result of two independent authentication transactions
> (with binding
> >> for certain PTs), for e.g. For password based credentials, the NEA=20
> >> server will be authenticated while a secure channel is being=20
> >> established (such as
> >> TLS) and NEA client will be subsequently authenticated within that=20
> >> secure channel.
> >>=20
> >>=20
> >> </Kaushik>
> >>=20
> >=20
> > <Gary>
> > Understood on the multi-layer approach. However, this
> requirement is
> > beyond the scope of the PT protocol and I think a new common C-XX=20
> > requirement is necessary that states the NEA Client and NEA Server=20
> > must provide  a secure channel via mutual authentication,=20
> > confidentiality and integrity through one or more of the
> PA, PB and PT
> > protocols collectively.
> > As to my comment on PT-2, I believe that still stands as it is=20
> > required for many if not most SSL VPN TLS connections.
> > </Gary>
> >=20
>=20
> <Kaushik-Response>
>=20
> The common requirement would be redundant because we already have=20
> individual requirements in PA, PB and PT for authentication, integrity

> & privacy protection. The reason for individual requirements is=20
> because priority for the PA, PB and PT layer requirements are=20
> different.
>=20
>=20
> Re-wording PT-2 to change "mutual authentication" to "mutual and=20
> server side only authentication schemes" would mean that it would=20
> acceptable have server-only authentication for PT which does not seem=20
> sufficient.
>=20
>=20
> </Kaushik-Response>
>=20
>=20

<Gary>
My only concern is not explicitly stating the common requirement in the
normative section 7, but rather in the teachings of the non normative
section 8.2. If the WG is comfortable with that, then I'll concur. =20

As to mutual authentication being necessary at the PT layer, I don't
concur at all as it isn't necessary if there is mutual authentication
being provided by an upper layer. For PT's such as 802.1X or IKE, PT
mutual authentication support is a given and thus PB and PA can rely on
it. However, for other
PT protocols, such as TLS it isn't a given at all. In fact, with TLS it
is unlikely mutual authentication will be employed in most remote access
use cases due to the lack of client cert adoption.  This is why I
recommended the changes in order to utilize PT simply for
confidentiality and rely on an upper protocol for authentication and
integrity.  An example might help here: In our SSL VPN solution as is
the case with other SSL VPN solutions I'm aware of, server side only
authentication is employed and the PB equivalent protocol provides
cryptographic mutual authentication between the PB handlers for trust
and integrity.  In this case, the PT protocol (i.e. TLS) is relied upon
for confidentiality.
</Gary>

[snippet]

> >>=20
> >> <Kaushik>
> >>=20
> >> Just want to point out that all the models described in section 9=20
> >> are implementation considerations and not requirements for NEA=20
> >> protocol specification. The models explain how NEA implementations=20
> >> can address privacy concerns and it is upto to implementations to=20
> >> support one or more of these models.
> >>=20
> >> regards,
> >>  kaushik!
> >>=20
> >> </Kaushik>
> >>=20
> >>=20
> >=20
> > <Gary>
> > Understood, although I don't think section 9.2 paragraph 1 clearly=20
> > states this.  I suggest rephrasing this paragraph to
> explicitly state
> > the NEA reference architecture and its protocols can be
> used with any
> > and
> > all of the 3 NEA Client to NEA Server interaction models;
> and that the
> > selection of an interaction model by a PC/PV implementer is use case

> > dependent. Once this context is established, I think it is fine to=20
> > switch the context to protocol consideration guidance for the
> forthcoming PB
> > and
> > PA design teams.
> >=20
>=20
>=20
> <Kaushik-Response>
>=20
> The second sentence in Section 9 states
>=20
> " This section will discuss a few of the potential privacy concerns=20
> raised by the deployment of this technology and offer some guidance to

> implementers."
>=20
> Does that seem adequate or do you think we need additional text in=20
> section
> 9.2
>=20
>=20
> regards,
>  kaushik
>=20
> </Kaushik-Response>
>=20
>=20

<Gary>
This part of the text is fine.  The part that through me for a loop is
"... whether a preferred approach will emerge potentially impacting the
protocols ...". I think we should plan to satisfy the three models in
the protocol work.  I'm comfortable with the verbiage now that I
understand the context better.

Regards,
Gary

</Gary>


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Thu Sep 13 12:09:32 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVrFz-0008Lj-S6; Thu, 13 Sep 2007 12:09:31 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVrFy-0008Le-M5
	for nea-confirm+ok@megatron.ietf.org; Thu, 13 Sep 2007 12:09:30 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVrFy-0008LW-CF
	for nea@ietf.org; Thu, 13 Sep 2007 12:09:30 -0400
Received: from sj-iport-6.cisco.com ([171.71.176.117])
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IVrFx-0000JR-5D
	for nea@ietf.org; Thu, 13 Sep 2007 12:09:30 -0400
X-IronPort-AV: E=Sophos;i="4.20,250,1186383600"; d="scan'208";a="217542166"
Received: from sj-dkim-2.cisco.com ([171.71.179.186])
	by sj-iport-6.cisco.com with ESMTP; 13 Sep 2007 09:09:28 -0700
Received: from sj-core-2.cisco.com (sj-core-2.cisco.com [171.71.177.254])
	by sj-dkim-2.cisco.com (8.12.11/8.12.11) with ESMTP id l8DG9S1n024959; 
	Thu, 13 Sep 2007 09:09:28 -0700
Received: from xbh-sjc-211.amer.cisco.com (xbh-sjc-211.cisco.com
	[171.70.151.144])
	by sj-core-2.cisco.com (8.12.10/8.12.6) with ESMTP id l8DG9Sat006156;
	Thu, 13 Sep 2007 16:09:28 GMT
Received: from xmb-sjc-22d.amer.cisco.com ([128.107.191.68]) by
	xbh-sjc-211.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.1830); 
	Thu, 13 Sep 2007 09:09:28 -0700
Received: from 10.21.117.47 ([10.21.117.47]) by xmb-sjc-22d.amer.cisco.com
	([128.107.191.68]) with Microsoft Exchange Server HTTP-DAV ; 
	Thu, 13 Sep 2007 16:09:28 +0000
User-Agent: Microsoft-Entourage/11.3.6.070618
Date: Thu, 13 Sep 2007 09:09:26 -0700
Subject: Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
From: Kaushik Narayan <kaushik@cisco.com>
To: <panandam@SonicWALL.com>, <nea@ietf.org>
Message-ID: <C30EAEC6.AECA%kaushik@cisco.com>
Thread-Topic: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acfz5qfsPz1SyEhoRu2ib5KW8a/VZAAHAHudACY6VtAAFxHHrwAYXsJQADHH1ag=
In-Reply-To: <D37A04F05A87044A8DCEEC0BF9A3AB1D07D0BE@us0exb10.us.sonicwall.com>
Mime-version: 1.0
Content-type: text/plain;
	charset="US-ASCII"
Content-transfer-encoding: 7bit
X-OriginalArrivalTime: 13 Sep 2007 16:09:28.0204 (UTC)
	FILETIME=[75F9CCC0:01C7F620]
DKIM-Signature: v=0.5; a=rsa-sha256; q=dns/txt; l=877; t=1189699768;
	x=1190563768; c=relaxed/simple; s=sjdkim2002;
	h=Content-Type:From:Subject:Content-Transfer-Encoding:MIME-Version;
	d=cisco.com; i=kaushik@cisco.com;
	z=From:=20Kaushik=20Narayan=20<kaushik@cisco.com>
	|Subject:=20Re=3A=20[Nea]=20Newcomer=20comments=20on=20draft-ietf-nea-req
	uirements-04 |Sender:=20;
	bh=oVHSK4nmZ/Ktxxw/nhtCW/RqCyME7PjQtbUJNUljdrY=;
	b=bFJdy47K2ujtRSbgw6PUCKmNUkjpinjwBLlhcUZY7/5sUZBgpwR0IX+WIM7o3xHhcyCptI87
	LP45QyaLdnBa5kxzzXjeG5o5773Ss67ZNXnCWalGDLQeqR6KrgIEy1sN;
Authentication-Results: sj-dkim-2; header.From=kaushik@cisco.com; dkim=pass (
	sig from cisco.com/sjdkim2002 verified; ); 
X-Spam-Score: -4.0 (----)
X-Scan-Signature: ffa9dfbbe7cc58b3fa6b8ae3e57b0aa3
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hi Parvez,

Please find my reply inline.

<snipped>

> 
> 
> [PA] The example you give above is a good one: 1) the NEA server is
> first authenticated while a secure channel is being established and 2)
> The NEA client subsequently authenticates within that secure channel. In
> this example, I would have thought that 1) is part of the PT protocol
> whereas 2) would be part of a higher-level protocol such as PB, in which
> case the mutual authentication does not happen exclusively within PT. Am
> I incorrect in this assumption? Thanks for clarifying...
>   
> 

The PT protocol MUST authenticate the NEA client, it's fairly
straight-forward for protocols such as TLS or EAP to support this. In
general the NEA reference model requires the NEA protocols, i.e. PA, PB & PT
to satisfy the security requirements independently.

regards,
  kaushik 
 


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Thu Sep 13 18:00:21 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVwjT-0003eW-Sl; Thu, 13 Sep 2007 18:00:19 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVwjS-0003eQ-Nw
	for nea-confirm+ok@megatron.ietf.org; Thu, 13 Sep 2007 18:00:18 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVwjS-0003eH-E2
	for nea@ietf.org; Thu, 13 Sep 2007 18:00:18 -0400
Received: from sj-iport-5.cisco.com ([171.68.10.87])
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IVwjR-0000Hq-2q
	for nea@ietf.org; Thu, 13 Sep 2007 18:00:18 -0400
X-IronPort-AV: E=Sophos;i="4.20,251,1186383600"; d="scan'208";a="175910597"
Received: from sj-dkim-3.cisco.com ([171.71.179.195])
	by sj-iport-5.cisco.com with ESMTP; 13 Sep 2007 15:00:16 -0700
Received: from sj-core-5.cisco.com (sj-core-5.cisco.com [171.71.177.238])
	by sj-dkim-3.cisco.com (8.12.11/8.12.11) with ESMTP id l8DM0Gcn022943; 
	Thu, 13 Sep 2007 15:00:16 -0700
Received: from xbh-sjc-231.amer.cisco.com (xbh-sjc-231.cisco.com
	[128.107.191.100])
	by sj-core-5.cisco.com (8.12.10/8.12.6) with ESMTP id l8DM0GoJ028596;
	Thu, 13 Sep 2007 22:00:16 GMT
Received: from xmb-sjc-22d.amer.cisco.com ([128.107.191.68]) by
	xbh-sjc-231.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.1830); 
	Thu, 13 Sep 2007 15:00:16 -0700
Received: from 171.69.75.24 ([171.69.75.24]) by xmb-sjc-22d.amer.cisco.com
	([128.107.191.68]) with Microsoft Exchange Server HTTP-DAV ; 
	Thu, 13 Sep 2007 22:00:16 +0000
User-Agent: Microsoft-Entourage/11.3.6.070618
Date: Thu, 13 Sep 2007 15:00:15 -0700
Subject: Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
From: Kaushik Narayan <kaushik@cisco.com>
To: <gtomlinson@SonicWALL.com>, <nea@ietf.org>
Message-ID: <C30F00FF.AF6C%kaushik@cisco.com>
Thread-Topic: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acf1lz1lbPJI7lC5RgKSgOOSpelueQAujlzT
In-Reply-To: <4DF8F0A275770B4387764386ABEF4D7A100B4C@us0exb10.us.sonicwall.com>
Mime-version: 1.0
Content-type: text/plain;
	charset="US-ASCII"
Content-transfer-encoding: 7bit
X-OriginalArrivalTime: 13 Sep 2007 22:00:16.0393 (UTC)
	FILETIME=[77AC8F90:01C7F651]
DKIM-Signature: v=0.5; a=rsa-sha256; q=dns/txt; l=2844; t=1189720816;
	x=1190584816; c=relaxed/simple; s=sjdkim3002;
	h=Content-Type:From:Subject:Content-Transfer-Encoding:MIME-Version;
	d=cisco.com; i=kaushik@cisco.com;
	z=From:=20Kaushik=20Narayan=20<kaushik@cisco.com>
	|Subject:=20Re=3A=20[Nea]=20Newcomer=20comments=20on=20draft-ietf-nea-req
	uirements-04 |Sender:=20;
	bh=gYtzS9C9R+wLBONfYLl/L+1Uhd9MPInRA0lpJEOK3dQ=;
	b=sO2iVM2sVXAUF6YpzLlw3W72ZiuiwqPSrX/f043ucKbFO4vCAVdCMmBKZMaNAe7eCz9ou56h
	0Rq8O9n2EnZG0K8gnJjebpu2DCjasQYUv5Cer2ILqXFoPL0AYdQEcuCz;
Authentication-Results: sj-dkim-3; header.From=kaushik@cisco.com; dkim=pass (
	sig from cisco.com/sjdkim3002 verified; ); 
X-Spam-Score: -1.4 (-)
X-Scan-Signature: bdc523f9a54890b8a30dd6fd53d5d024
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hi Gary,

Please find my reply inline.

<snipped>

>>> 
>> 
>> <Kaushik-Response>
>> 
>> The common requirement would be redundant because we already have
>> individual requirements in PA, PB and PT for authentication, integrity
> 
>> & privacy protection. The reason for individual requirements is
>> because priority for the PA, PB and PT layer requirements are
>> different.
>> 
>> 
>> Re-wording PT-2 to change "mutual authentication" to "mutual and
>> server side only authentication schemes" would mean that it would
>> acceptable have server-only authentication for PT which does not seem
>> sufficient.
>> 
>> 
>> </Kaushik-Response>
>> 
>> 
> 
> <Gary>
> My only concern is not explicitly stating the common requirement in the
> normative section 7, but rather in the teachings of the non normative
> section 8.2. If the WG is comfortable with that, then I'll concur.

<Kaushik-Response>

I am still not sure how the common requirement won't be redundant if we have
individual requirements for PA, PB & PT (all of these are in Section 7). The
common requirements essentially apply to these three protocols.

</Kaushik-Response>


> 
> As to mutual authentication being necessary at the PT layer, I don't
> concur at all as it isn't necessary if there is mutual authentication
> being provided by an upper layer. For PT's such as 802.1X or IKE, PT
> mutual authentication support is a given and thus PB and PA can rely on
> it. However, for other
> PT protocols, such as TLS it isn't a given at all. In fact, with TLS it
> is unlikely mutual authentication will be employed in most remote access
> use cases due to the lack of client cert adoption.  This is why I
> recommended the changes in order to utilize PT simply for
> confidentiality and rely on an upper protocol for authentication and
> integrity.  An example might help here: In our SSL VPN solution as is
> the case with other SSL VPN solutions I'm aware of, server side only
> authentication is employed and the PB equivalent protocol provides
> cryptographic mutual authentication between the PB handlers for trust
> and integrity.  In this case, the PT protocol (i.e. TLS) is relied upon
> for confidentiality.
> </Gary>


<Kaushik-Response>

I am not sure that client-side authentication would be limited to
certificates with TLS as transport. There are options for TLS to
authenticate using Kerberos (RFC2717), pre-shared keys (RFC4279)
& username/passwords (draft-ietf-tls-srp-14.txt). TLS should be able to meet
the mutual authentication requirements.


I think this discussion points out that we probably missing a table as part
of the PT requirements that indicates the type of client/server
authentication methods the PT is required to support.

regards,
 kaushik

</Kaushik-Response>



_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Thu Sep 13 19:58:32 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IVyZr-00028o-Dw; Thu, 13 Sep 2007 19:58:31 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IVyZq-0001ve-6Q
	for nea-confirm+ok@megatron.ietf.org; Thu, 13 Sep 2007 19:58:30 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IVyZp-0001u6-Rc
	for nea@ietf.org; Thu, 13 Sep 2007 19:58:29 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IVyZo-0002XP-Ik
	for nea@ietf.org; Thu, 13 Sep 2007 19:58:29 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id ht76qm0171sq for <nea@ietf.org>;
	Thu, 13 Sep 2007 16:54:18 -0700 (envelope-from
	<panandam@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9143)
	with ESMTP; Thu, 13 Sep 2007 16:54:18 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Thu, 13 Sep 2007 16:58:26 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Date: Thu, 13 Sep 2007 16:58:23 -0700
Message-ID: <D37A04F05A87044A8DCEEC0BF9A3AB1D07D117@us0exb10.us.sonicwall.com>
In-Reply-To: <C30EAEC6.AECA%kaushik@cisco.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acfz5qfsPz1SyEhoRu2ib5KW8a/VZAAHAHudACY6VtAAFxHHrwAYXsJQADHH1agAED89UA==
References: <D37A04F05A87044A8DCEEC0BF9A3AB1D07D0BE@us0exb10.us.sonicwall.com>
	<C30EAEC6.AECA%kaushik@cisco.com>
From: <panandam@SonicWALL.com>
To: <nea@ietf.org>
X-OriginalArrivalTime: 13 Sep 2007 23:58:26.0517 (UTC)
	FILETIME=[F9B7A050:01C7F661]
X-Mlf-Version: 6.0.1.9143
X-Mlf-UniqueId: o200709132354150006302
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 21c69d3cfc2dd19218717dbe1d974352
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hi Kaushik,

[PA] Please find my reply inline.

Regards,
Parvez


<snipped>

>=20
>=20
> [PA] The example you give above is a good one: 1) the NEA server is
> first authenticated while a secure channel is being established and 2)
> The NEA client subsequently authenticates within that secure channel.
In
> this example, I would have thought that 1) is part of the PT protocol
> whereas 2) would be part of a higher-level protocol such as PB, in
which
> case the mutual authentication does not happen exclusively within PT.
Am
> I incorrect in this assumption? Thanks for clarifying...
>  =20
>=20

The PT protocol MUST authenticate the NEA client, it's fairly
straight-forward for protocols such as TLS or EAP to support this. In
general the NEA reference model requires the NEA protocols, i.e. PA, PB
& PT
to satisfy the security requirements independently.

regards,
  kaushik=20
=20

[PA] I do agree with you that TLS and EAP certainly support mutual
authentication and implementing that is not the problem. What I'm having
a hard time with is why it's so important for PT to have the MUST for
mutual authentication whereas the PB protocol has a MAY. After all,
isn't the main trust relationship between the Posture Broker Client and
the Posture Broker Server? If that trust relationship can be
transitively derived from the PT layer, well and good, but I'm not sure
I see why that transitivity needs to be mandated.=20



_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Fri Sep 14 01:58:28 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IW4CB-000392-AH; Fri, 14 Sep 2007 01:58:27 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IW4CA-00038w-Qu
	for nea-confirm+ok@megatron.ietf.org; Fri, 14 Sep 2007 01:58:26 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IW4CA-00038o-GX
	for nea@ietf.org; Fri, 14 Sep 2007 01:58:26 -0400
Received: from sj-iport-6.cisco.com ([171.71.176.117])
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IW4C9-0002h0-8c
	for nea@ietf.org; Fri, 14 Sep 2007 01:58:26 -0400
X-IronPort-AV: E=Sophos;i="4.20,253,1186383600"; d="scan'208";a="217954443"
Received: from sj-dkim-1.cisco.com ([171.71.179.21])
	by sj-iport-6.cisco.com with ESMTP; 13 Sep 2007 22:58:24 -0700
Received: from sj-core-2.cisco.com (sj-core-2.cisco.com [171.71.177.254])
	by sj-dkim-1.cisco.com (8.12.11/8.12.11) with ESMTP id l8E5wOni018965; 
	Thu, 13 Sep 2007 22:58:24 -0700
Received: from xbh-sjc-231.amer.cisco.com (xbh-sjc-231.cisco.com
	[128.107.191.100])
	by sj-core-2.cisco.com (8.12.10/8.12.6) with ESMTP id l8E5wIat010304;
	Fri, 14 Sep 2007 05:58:20 GMT
Received: from xmb-sjc-22d.amer.cisco.com ([128.107.191.68]) by
	xbh-sjc-231.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.1830); 
	Thu, 13 Sep 2007 22:58:18 -0700
Received: from 10.21.92.152 ([10.21.92.152]) by xmb-sjc-22d.amer.cisco.com
	([128.107.191.68]) with Microsoft Exchange Server HTTP-DAV ; 
	Fri, 14 Sep 2007 05:58:17 +0000
User-Agent: Microsoft-Entourage/11.3.6.070618
Date: Thu, 13 Sep 2007 22:58:17 -0700
Subject: Re: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
From: Kaushik Narayan <kaushik@cisco.com>
To: <panandam@SonicWALL.com>, <nea@ietf.org>
Message-ID: <C30F7109.AFF8%kaushik@cisco.com>
Thread-Topic: [Nea] Newcomer comments on draft-ietf-nea-requirements-04
Thread-Index: Acfz5qfsPz1SyEhoRu2ib5KW8a/VZAAHAHudACY6VtAAFxHHrwAYXsJQADHH1agAED89UAAMs0LC
In-Reply-To: <D37A04F05A87044A8DCEEC0BF9A3AB1D07D117@us0exb10.us.sonicwall.com>
Mime-version: 1.0
Content-type: text/plain;
	charset="US-ASCII"
Content-transfer-encoding: 7bit
X-OriginalArrivalTime: 14 Sep 2007 05:58:18.0214 (UTC)
	FILETIME=[3F5F2060:01C7F694]
DKIM-Signature: v=0.5; a=rsa-sha256; q=dns/txt; l=1176; t=1189749504;
	x=1190613504; c=relaxed/simple; s=sjdkim1004;
	h=Content-Type:From:Subject:Content-Transfer-Encoding:MIME-Version;
	d=cisco.com; i=kaushik@cisco.com;
	z=From:=20Kaushik=20Narayan=20<kaushik@cisco.com>
	|Subject:=20Re=3A=20[Nea]=20Newcomer=20comments=20on=20draft-ietf-nea-req
	uirements-04 |Sender:=20;
	bh=cGhdItDXeIwokZtRIySy9vVhTADRE+VcSO7dOfvHT1Y=;
	b=MYgm7+/KhYHXJRzpNavEMnk38Ys4ajdegdB2jVJwY1FS46RbgVbBxTvPDkmPsnWvTp3fXuzs
	ylm91Fryx7YId0b70RSYFkLaVPT5ELwD6j19VH1T8N5VgC+v3023maeWNWhHJOey8i6I7a5gXd
	k+h0aFfbjmyF5WKenC1+b/nVw=;
Authentication-Results: sj-dkim-1; header.From=kaushik@cisco.com; dkim=pass (
	sig from cisco.com/sjdkim1004 verified; ); 
X-Spam-Score: -4.0 (----)
X-Scan-Signature: 97adf591118a232206bdb5a27b217034
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hi Parvez,

Please find my reply inline.

<snipped>
>> 
> 
> The PT protocol MUST authenticate the NEA client, it's fairly
> straight-forward for protocols such as TLS or EAP to support this. In
> general the NEA reference model requires the NEA protocols, i.e. PA, PB
> & PT
> to satisfy the security requirements independently.
> 
>  
> 
> [PA] I do agree with you that TLS and EAP certainly support mutual
> authentication and implementing that is not the problem. What I'm having
> a hard time with is why it's so important for PT to have the MUST for
> mutual authentication whereas the PB protocol has a MAY. After all,
> isn't the main trust relationship between the Posture Broker Client and
> the Posture Broker Server? If that trust relationship can be
> transitively derived from the PT layer, well and good, but I'm not sure
> I see why that transitivity needs to be mandated.
> 
> 


Protocols such as TLS, EAP etc. that NEA will rely on to provide posture
transport are already capable of providing mutual authentication,
confidentiality and privacy protection. This allows simplification of the PB
protocol.

Regards,
 kaushik


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Sat Sep 15 16:29:46 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IWeGq-0005OE-5G; Sat, 15 Sep 2007 16:29:40 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IWeGo-0005O9-W2
	for nea-confirm+ok@megatron.ietf.org; Sat, 15 Sep 2007 16:29:38 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IWeGo-0005O1-Bl
	for nea@ietf.org; Sat, 15 Sep 2007 16:29:38 -0400
Received: from extu-mxob-1.symantec.com ([216.10.194.28])
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IWeGm-0002tm-SM
	for nea@ietf.org; Sat, 15 Sep 2007 16:29:38 -0400
Received: from tus1opsmtapin01.ges.symantec.com
	(tus1opsmtapin01.ges.symantec.com [192.168.214.43])
	by extu-mxob-1.symantec.com (8.14.1/8.14.1) with ESMTP id
	l8FKTVIE025381
	(version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO)
	for <nea@ietf.org>; Sat, 15 Sep 2007 13:29:36 -0700
Received: from reserved-155-64-230-18.ges.symantec.com ([155.64.230.18]
	helo=TUS1XCHECNPIN01.enterprise.veritas.com)
	by tus1opsmtapin01.ges.symantec.com with esmtp (Exim 4.67)
	(envelope-from <Paul_Sangster@symantec.com>) id 1IWeGX-0007nF-8L
	for nea@ietf.org; Sat, 15 Sep 2007 13:29:21 -0700
Received: from TUS1XCHEVSPIN05.enterprise.veritas.com ([155.64.231.27]) by
	TUS1XCHECNPIN01.enterprise.veritas.com with Microsoft
	SMTPSVC(6.0.3790.1830); Sat, 15 Sep 2007 13:29:20 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Date: Sat, 15 Sep 2007 13:29:20 -0700
Message-ID: <AB96CED633A7C246BDC661DBEE1CF01FD48F76@TUS1XCHCLUPIN11.enterprise.veritas.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: Should we downgrade PA security to a SHOULD, MAY or remove it?
Thread-Index: AcfwzTq2XBTSVc2lRlGEdm6kXscoXwG9SksA
From: "Paul Sangster" <Paul_Sangster@symantec.com>
To: <nea@ietf.org>
X-OriginalArrivalTime: 15 Sep 2007 20:29:20.0925 (UTC)
	FILETIME=[18C9B4D0:01C7F7D7]
X-Spam-Score: -4.0 (----)
X-Scan-Signature: a492040269d440726bfd84680622cee7
Subject: [Nea] RE: Should we downgrade PA security to a SHOULD,
	MAY or remove it?
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Content-Type: multipart/mixed; boundary="===============0859409855=="
Errors-To: nea-bounces@ietf.org

This is a multi-part message in MIME format.

--===============0859409855==
Content-class: urn:content-classes:message
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C7F7D7.18B2BFEC"

This is a multi-part message in MIME format.

------_=_NextPart_001_01C7F7D7.18B2BFEC
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

Unfortunately, this thread has failed to have traction.  Since we
currently have a MUST for PA security would someone like to assert that
this MUST should be lowered and provide some rationale to start this
discussion?


________________________________

	From: Paul Sangster=20
	Sent: Thursday, September 06, 2007 2:31 PM
	To: nea@ietf.org
	Subject: Should we downgrade PA security to a SHOULD, MAY or
remove it?
=09
=09
	Hopefully everyone noticed that our -04 revision was published
and is=20
	awaiting comments from the WG.  As mentioned in my prior e-mail
on 8/24/07,
	we've made a significant number of changes based on comments
received=20
	during WGLC.  We tried to address those comments but one
significant
	question that was not addressed relates to our requirement PA-5
below:
	=20
	 "PA-5 The PA protocol MUST provide authentication, integrity,
and=20
	       confidentiality of attributes communicated between a
Posture=20
	       Collector and Posture Validator. This enables end-to-end
security=20
	       across a NEA deployment that might involve traversal of
several=20
	       systems."
	=20
	The question is should this requirement be a SHOULD, MAY, or
removed=20
	entirely?  This requirement guides our selection process for the
PA
	protocol and currently would prevent candidate PA protocols from
being
	selected if they lack proper security protections.  In the past
	we've discussed the overlapping requirements for security upon
PA (MUST),
	PB (MAY) and PT (MUST) and in Prague this factored into our
decision to
	reduce PB to a MAY (instead of a MUST).  This topic is also
discussed in=20
	section 8.2 and other parts of the spec.
	=20
	The current PA security protections were included to provide for
end to end=20
	(e.g. Posture Collector to Posture Validator) authentication,
integrity and
	confidentiality protection of the PA attributes.  These
attributes might travel to remote=20
	systems hosting the posture validator that are separate from the
posture=20
	transport server thus PT protections are no longer present.
Several folks=20
	have mentioned that the protection of the link between the
posture transport=20
	server and remote posture validator could be provided in other
less complex=20
	ways and so requiring a standards based PA security protocol
isn't critical=20
	and will just make our deliverable more difficult.  We also
would need to
	specify a solution consistent with the principals of RFC 4107
which could
	be challenging.
	=20
	We would like to request that the working group discuss whether
we should=20
	downgrade PA-5 to SHOULD, MAY or remove it all together.  In
order
	to bound the discussion we would like to conclude the discussion
by 10PM PDT
	on thursday, Sept 20th (2 weeks).  This will allows the editors
time to make=20
	any revisions to the specification and re-submit in a timely
manner so we can
	product a new I-D ready for a re-start the WGLC process.
	=20
	Paul


------_=_NextPart_001_01C7F7D7.18B2BFEC
Content-Type: text/html;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=3DContent-Type content=3D"text/html; =
charset=3Dus-ascii">
<META content=3D"MSHTML 6.00.2900.3157" name=3DGENERATOR></HEAD>
<BODY>
<DIV dir=3Dltr align=3Dleft><FONT face=3DArial color=3D#0000ff =
size=3D2><SPAN=20
class=3D546070118-15092007>Unfortunately, this thread has failed to have =

traction.&nbsp; Since we currently have a MUST for PA security would =
someone=20
like to assert that this MUST should be lowered and provide some =
rationale to=20
start this discussion?</SPAN></FONT></DIV><BR>
<BLOCKQUOTE dir=3Dltr=20
style=3D"PADDING-LEFT: 5px; MARGIN-LEFT: 5px; BORDER-LEFT: #0000ff 2px =
solid; MARGIN-RIGHT: 0px">
  <DIV class=3DOutlookMessageHeader lang=3Den-us dir=3Dltr align=3Dleft>
  <HR tabIndex=3D-1>
  <FONT face=3DTahoma size=3D2><B>From:</B> Paul Sangster =
<BR><B>Sent:</B> Thursday,=20
  September 06, 2007 2:31 PM<BR><B>To:</B> =
nea@ietf.org<BR><B>Subject:</B>=20
  Should we downgrade PA security to a SHOULD, MAY or remove=20
  it?<BR></FONT><BR></DIV>
  <DIV></DIV>
  <DIV><FONT face=3DArial size=3D2>Hopefully everyone noticed that our =
-04 revision=20
  was published and is <BR>awaiting comments from the WG.&nbsp; As =
mentioned in=20
  my prior e-mail on 8/24/07,<BR>we've made a significant number of =
changes=20
  based on comments received <BR>during WGLC.&nbsp; We tried to address =
those=20
  comments but one significant<BR>question that was not addressed =
relates to our=20
  requirement PA-5 below:</FONT></DIV>
  <DIV>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>&nbsp;"PA-5 The PA protocol MUST =
provide=20
  authentication, integrity, and =
<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;=20
  confidentiality of attributes communicated between a Posture=20
  <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Collector and Posture =
Validator. This=20
  enables end-to-end security <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
across a=20
  NEA deployment that might involve traversal of several=20
  <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; systems."</FONT></DIV>
  <DIV>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>The question is should this =
requirement be a=20
  SHOULD, MAY, or removed <BR>entirely?&nbsp; This requirement guides =
our=20
  selection process for the PA<BR>protocol and currently would prevent =
candidate=20
  PA protocols from being<BR>selected if they lack proper security=20
  protections.&nbsp; In the past<BR>we've discussed the overlapping =
requirements=20
  for security upon PA (MUST),<BR>PB (MAY) and PT (MUST) and in Prague =
this=20
  factored into our decision to<BR>reduce PB to a MAY (instead of a =
MUST).&nbsp;=20
  This topic is also discussed in <BR>section 8.2 and other parts of the =

  spec.</FONT></DIV>
  <DIV>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>The current PA security protections =
were included=20
  to provide for end to end <BR>(e.g. Posture Collector to Posture =
Validator)=20
  authentication, integrity and</FONT></DIV>
  <DIV><FONT face=3DArial size=3D2>confidentiality protection of the PA=20
  attributes<SPAN class=3D734362521-06092007>.&nbsp; These=20
  attributes&nbsp;m</SPAN>ight travel to remote <BR>systems hosting the =
posture=20
  validator that are separate from the posture <BR>transport server thus =
PT=20
  protections are no longer present.&nbsp; Several folks <BR>have =
mentioned that=20
  the protection of the link between the posture transport <BR>server =
and remote=20
  posture validator could be provided in other less complex <BR>ways and =
so=20
  requiring a standards based PA security protocol isn't critical =
<BR>and will=20
  just make our deliverable more difficult.&nbsp; We also would need=20
  to<BR>specify a solution consistent with the principals of RFC 4107 =
which=20
  could<BR>be challenging.</FONT></DIV>
  <DIV><FONT face=3DArial size=3D2></FONT>&nbsp;</DIV>
  <DIV><FONT face=3DArial size=3D2>We would like to request that the =
working group=20
  discuss whether we should <BR>downgrade PA-5 to SHOULD, MAY or remove =
it all=20
  together.&nbsp; In order<BR>to bound the discussion we would like to =
conclude=20
  the discussion by 10PM PDT<BR>on thursday, Sept 20th (2 weeks).&nbsp; =
This=20
  will allows the editors time to make <BR>any revisions to the =
specification=20
  and re-submit in a timely manner so we can<BR><SPAN=20
  class=3D734362521-06092007>product a new I-D ready for a =
</SPAN>re-start the=20
  WGLC process.</FONT></DIV>
  <DIV>&nbsp;</DIV>
  <DIV><FONT face=3DArial =
size=3D2>Paul</FONT></DIV></BLOCKQUOTE></BODY></HTML>

------_=_NextPart_001_01C7F7D7.18B2BFEC--



--===============0859409855==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea

--===============0859409855==--





From nea-bounces@ietf.org Mon Sep 17 19:13:23 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IXPkf-0000fy-PG; Mon, 17 Sep 2007 19:11:37 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IXPkf-0000fp-3u
	for nea-confirm+ok@megatron.ietf.org; Mon, 17 Sep 2007 19:11:37 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IXPke-0000fh-Pe
	for nea@ietf.org; Mon, 17 Sep 2007 19:11:36 -0400
Received: from sj-iport-3-in.cisco.com ([171.71.176.72]
	helo=sj-iport-3.cisco.com) by ietf-mx.ietf.org with esmtp (Exim 4.43)
	id 1IXPkd-0004sv-G4 for nea@ietf.org; Mon, 17 Sep 2007 19:11:36 -0400
X-IronPort-AV: E=Sophos;i="4.20,266,1186383600"; d="scan'208";a="524864955"
Received: from sj-dkim-2.cisco.com ([171.71.179.186])
	by sj-iport-3.cisco.com with ESMTP; 17 Sep 2007 16:11:33 -0700
Received: from sj-core-4.cisco.com (sj-core-4.cisco.com [171.68.223.138])
	by sj-dkim-2.cisco.com (8.12.11/8.12.11) with ESMTP id l8HNBX8w019082; 
	Mon, 17 Sep 2007 16:11:33 -0700
Received: from xbh-sjc-211.amer.cisco.com (xbh-sjc-211.cisco.com
	[171.70.151.144])
	by sj-core-4.cisco.com (8.12.10/8.12.6) with ESMTP id l8HNBFug004472;
	Mon, 17 Sep 2007 23:11:33 GMT
Received: from xmb-sjc-225.amer.cisco.com ([128.107.191.38]) by
	xbh-sjc-211.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.1830); 
	Mon, 17 Sep 2007 16:11:26 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="US-ASCII"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] RE: Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Date: Mon, 17 Sep 2007 16:11:31 -0700
Message-ID: <AC1CFD94F59A264488DC2BEC3E890DE504855466@xmb-sjc-225.amer.cisco.com>
In-Reply-To: <AB96CED633A7C246BDC661DBEE1CF01FD48F76@TUS1XCHCLUPIN11.enterprise.veritas.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] RE: Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Thread-Index: AcfwzTq2XBTSVc2lRlGEdm6kXscoXwG9SksAAG8JNZA=
From: "Joseph Salowey \(jsalowey\)" <jsalowey@cisco.com>
To: "Paul Sangster" <Paul_Sangster@symantec.com>, <nea@ietf.org>
X-OriginalArrivalTime: 17 Sep 2007 23:11:26.0841 (UTC)
	FILETIME=[12B63690:01C7F980]
DKIM-Signature: v=0.5; a=rsa-sha256; q=dns/txt; l=4356; t=1190070693;
	x=1190934693; c=relaxed/simple; s=sjdkim2002;
	h=Content-Type:From:Subject:Content-Transfer-Encoding:MIME-Version;
	d=cisco.com; i=jsalowey@cisco.com;
	z=From:=20=22Joseph=20Salowey=20\(jsalowey\)=22=20<jsalowey@cisco.com>
	|Subject:=20RE=3A=20[Nea]=20RE=3A=20Should=20we=20downgrade=20PA=20securi
	ty=20to=20a=20SHOULD,MAY=20or=20remove=20it? |Sender:=20;
	bh=UIwI5tKdLKpWigaVt3K1uFIHWs/ZOl3NyjVlutnaA3c=;
	b=2Jmyr5PCYEqNPMdUYw8b3d/zH/y4dXqysjkCZkdboI63sRcChmq0c/Cmw8iJJvbtzFJOLxcq
	INq5h8IIRsEwiKXJsTDkYQ/Wwcudoq7JObQYHqeue1eWQPmkTs/dKX8v;
Authentication-Results: sj-dkim-2; header.From=jsalowey@cisco.com; dkim=pass (
	sig from cisco.com/sjdkim2002 verified; ); 
X-Spam-Score: -4.0 (----)
X-Scan-Signature: c83ccb5cc10e751496398f1233ca9c3a
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

My concern is that applying security at the PA layer has the most
management complexity.  There is a fair amount of complexity in key
management at this layer.  This will require the working group to spend
a significant amount of time and energy to define. I believe that the
security relationships are more manageable at the broker or transport
layer.  It is also not clear to me what are the practical advantages
requiring standard security mechanisms at the PA layer.

Note that if you are going to go down the path of stand alone security
for the PA layer then you need to work on the description of trust in
the requirements document, since the current document discusses trust
between a NEA client and an NEA server instead of the PA components. =20

Cheers,

Joe

> -----Original Message-----
> From: Paul Sangster [mailto:Paul_Sangster@symantec.com]=20
> Sent: Saturday, September 15, 2007 1:29 PM
> To: nea@ietf.org
> Subject: [Nea] RE: Should we downgrade PA security to a=20
> SHOULD,MAY or remove it?
>=20
> Unfortunately, this thread has failed to have traction. =20
> Since we currently have a MUST for PA security would someone=20
> like to assert that this MUST should be lowered and provide=20
> some rationale to start this discussion?
>=20
>=20
> ________________________________
>=20
> 	From: Paul Sangster=20
> 	Sent: Thursday, September 06, 2007 2:31 PM
> 	To: nea@ietf.org
> 	Subject: Should we downgrade PA security to a SHOULD,=20
> MAY or remove it?
> =09
> =09
> 	Hopefully everyone noticed that our -04 revision was=20
> published and is=20
> 	awaiting comments from the WG.  As mentioned in my=20
> prior e-mail on 8/24/07,
> 	we've made a significant number of changes based on=20
> comments received=20
> 	during WGLC.  We tried to address those comments but=20
> one significant
> 	question that was not addressed relates to our=20
> requirement PA-5 below:
> 	=20
> 	 "PA-5 The PA protocol MUST provide authentication,=20
> integrity, and=20
> 	       confidentiality of attributes communicated=20
> between a Posture=20
> 	       Collector and Posture Validator. This enables=20
> end-to-end security=20
> 	       across a NEA deployment that might involve=20
> traversal of several=20
> 	       systems."
> 	=20
> 	The question is should this requirement be a SHOULD,=20
> MAY, or removed=20
> 	entirely?  This requirement guides our selection=20
> process for the PA
> 	protocol and currently would prevent candidate PA=20
> protocols from being
> 	selected if they lack proper security protections.  In the past
> 	we've discussed the overlapping requirements for=20
> security upon PA (MUST),
> 	PB (MAY) and PT (MUST) and in Prague this factored into=20
> our decision to
> 	reduce PB to a MAY (instead of a MUST).  This topic is=20
> also discussed in=20
> 	section 8.2 and other parts of the spec.
> 	=20
> 	The current PA security protections were included to=20
> provide for end to end=20
> 	(e.g. Posture Collector to Posture Validator)=20
> authentication, integrity and
> 	confidentiality protection of the PA attributes.  These=20
> attributes might travel to remote=20
> 	systems hosting the posture validator that are separate=20
> from the posture=20
> 	transport server thus PT protections are no longer=20
> present.  Several folks=20
> 	have mentioned that the protection of the link between=20
> the posture transport=20
> 	server and remote posture validator could be provided=20
> in other less complex=20
> 	ways and so requiring a standards based PA security=20
> protocol isn't critical=20
> 	and will just make our deliverable more difficult.  We=20
> also would need to
> 	specify a solution consistent with the principals of=20
> RFC 4107 which could
> 	be challenging.
> 	=20
> 	We would like to request that the working group discuss=20
> whether we should=20
> 	downgrade PA-5 to SHOULD, MAY or remove it all=20
> together.  In order
> 	to bound the discussion we would like to conclude the=20
> discussion by 10PM PDT
> 	on thursday, Sept 20th (2 weeks).  This will allows the=20
> editors time to make=20
> 	any revisions to the specification and re-submit in a=20
> timely manner so we can
> 	product a new I-D ready for a re-start the WGLC process.
> 	=20
> 	Paul
>=20
>=20


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Mon Sep 17 19:55:19 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IXQQC-00043w-Hf; Mon, 17 Sep 2007 19:54:32 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IXQQB-00042b-0L
	for nea-confirm+ok@megatron.ietf.org; Mon, 17 Sep 2007 19:54:31 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IXQQA-0003wr-77
	for nea@ietf.org; Mon, 17 Sep 2007 19:54:30 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IXQQ4-0000x0-JC
	for nea@ietf.org; Mon, 17 Sep 2007 19:54:25 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id hts9ki0171s4 for <nea@ietf.org>;
	Mon, 17 Sep 2007 16:52:40 -0700 (envelope-from
	<gtomlinson@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9159)
	with ESMTP; Mon, 17 Sep 2007 16:52:40 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Mon, 17 Sep 2007 16:54:28 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] RE: Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Date: Mon, 17 Sep 2007 16:54:19 -0700
Message-ID: <4DF8F0A275770B4387764386ABEF4D7A100E88@us0exb10.us.sonicwall.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: RE: [Nea] RE: Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Thread-Index: Acf5hhAxbDGGJzWyTzKSIksmrrqTig==
From: <gtomlinson@SonicWALL.com>
To: <nea@ietf.org>,
	<Paul_Sangster@symantec.com>
X-OriginalArrivalTime: 17 Sep 2007 23:54:28.0935 (UTC)
	FILETIME=[15C27D70:01C7F986]
X-Mlf-Version: 6.0.1.9159
X-Mlf-UniqueId: o200709172352400052270
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 0fa76816851382eb71b0a882ccdc29ac
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hello Paul,

Parvez and I have been thinking about this quite a bit, especially since
discussing the requirements of PT mutual authentication with Kaushik,
which has lead us to quite a bit of trust model analysis.

Our PA-5 response is inline.  We will be commenting separately on the
PB/PT trust relationship which is germane to the PT mutual
authentication discussion thread.

> Date: Sat, 15 Sep 2007 13:29:20 -0700
> From: "Paul Sangster" <Paul_Sangster@symantec.com>
> Subject: [Nea] RE: Should we downgrade PA security to a=20
> SHOULD,	MAY or
> 	remove it?
> To: <nea@ietf.org>
> Message-ID:
> =09
> <AB96CED633A7C246BDC661DBEE1CF01FD48F76@TUS1XCHCLUPIN11.enterp
> rise.veritas.com>
> =09
> Content-Type: text/plain; charset=3D"us-ascii"
>=20
> Unfortunately, this thread has failed to have traction. =20
> Since we currently have a MUST for PA security would someone like to=20
> assert that this MUST should be lowered and provide some rationale to=20
> start this discussion?
>=20


We'll take a stab at asserting PA-5 should be a [MUST].  Here is our
argument.=20

Unfortunately, we've come to the group late in the requirements game and
thus didn't participate in the Prague discussions and as such don't have
the context that lead to the downgrading of PB authentication and
shifting all of this responsibility to PT and PA respectively. However,
we think the most important question is how does the trust model work. =20

If a PV does not utilize PA authentication then it implicitly is relying
upon the PB protocol and/or the PT protocol for authentication of its
associated PC. Such transitive trust is possible in use cases where a
NEA Client can provide assurances on the authenticity of the PCs it
hosts, such as in systems with immutable software images. However, for
use cases where such assurances can't be made, such as when PCs can be
installed on systems for which the PB Client can't ascertain
authenticity in behalf of the PV, the transitive trust relationship
breaks down. Given such use cases exist, we don't see how PA-5
authentication can be dropped. In fact, thinking about this more, it
occurs to us that the minimum trust model required for NEA is PA-5 with
mutual authentication and trust between the PB Server and its associated
PVs via a (out of WG scope) secure interface.  Our rationale for this
assertion is, if each PV/PC combination has a cryptographically sound
communication channel, then it isn't possible for a PB Client, or PT
stack to spoof said messages and if PB Server has a secure trusted
relationship with each PV, then PB Server can rely entirely on
assertions from the PV's in composing the complete end point posture
state.
We're not advocating this minimalist model as the only trust
requirements, just stating we think it is the minimum set of trust
requirements.

Now comes the semantic fun.  What is the meaning of [MUST], [SHOULD] and
[MAY] in the context of PA-5?  Our assertion is PA-5 is a protocol
requirement and as such, given our assertion there is a need for PA-5
mutual authentication, then the PA protocol [MUST] be capable of
providing mutual authentication.  Since, the need for mutual
authentication is use case dependent, we assert implementation of it is
an optional requirement and therefore think a [MAY] implementation
requirement is appropriate to consider in the forthcoming PA protocol
RFC work.

[omitted]

Regards,

Gary


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Mon Sep 17 23:46:23 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IXU0J-00075c-Nq; Mon, 17 Sep 2007 23:44:03 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IXU0I-0006zi-2g
	for nea-confirm+ok@megatron.ietf.org; Mon, 17 Sep 2007 23:44:02 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IXU0H-0006xR-O6
	for nea@ietf.org; Mon, 17 Sep 2007 23:44:01 -0400
Received: from sj-iport-5.cisco.com ([171.68.10.87])
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IXU0B-0002rp-6m
	for nea@ietf.org; Mon, 17 Sep 2007 23:44:01 -0400
X-IronPort-AV: E=Sophos;i="4.20,266,1186383600"; d="scan'208";a="176496717"
Received: from sj-dkim-2.cisco.com ([171.71.179.186])
	by sj-iport-5.cisco.com with ESMTP; 17 Sep 2007 20:43:49 -0700
Received: from sj-core-5.cisco.com (sj-core-5.cisco.com [171.71.177.238])
	by sj-dkim-2.cisco.com (8.12.11/8.12.11) with ESMTP id l8I3hnYF018541; 
	Mon, 17 Sep 2007 20:43:49 -0700
Received: from xbh-sjc-231.amer.cisco.com (xbh-sjc-231.cisco.com
	[128.107.191.100])
	by sj-core-5.cisco.com (8.12.10/8.12.6) with ESMTP id l8I3hjfD024251;
	Tue, 18 Sep 2007 03:43:47 GMT
Received: from xmb-sjc-225.amer.cisco.com ([128.107.191.38]) by
	xbh-sjc-231.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.1830); 
	Mon, 17 Sep 2007 20:43:45 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="US-ASCII"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] RE: Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Date: Mon, 17 Sep 2007 20:43:51 -0700
Message-ID: <AC1CFD94F59A264488DC2BEC3E890DE504855533@xmb-sjc-225.amer.cisco.com>
In-Reply-To: <4DF8F0A275770B4387764386ABEF4D7A100E88@us0exb10.us.sonicwall.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] RE: Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Thread-Index: Acf5hhAxbDGGJzWyTzKSIksmrrqTigAHjzVA
From: "Joseph Salowey \(jsalowey\)" <jsalowey@cisco.com>
To: <gtomlinson@SonicWALL.com>, <nea@ietf.org>, <Paul_Sangster@symantec.com>
X-OriginalArrivalTime: 18 Sep 2007 03:43:45.0946 (UTC)
	FILETIME=[1D93C7A0:01C7F9A6]
DKIM-Signature: v=0.5; a=rsa-sha256; q=dns/txt; l=4858; t=1190087029;
	x=1190951029; c=relaxed/simple; s=sjdkim2002;
	h=Content-Type:From:Subject:Content-Transfer-Encoding:MIME-Version;
	d=cisco.com; i=jsalowey@cisco.com;
	z=From:=20=22Joseph=20Salowey=20\(jsalowey\)=22=20<jsalowey@cisco.com>
	|Subject:=20RE=3A=20[Nea]=20RE=3A=20Should=20we=20downgrade=20PA=20securi
	ty=20to=20a=20SHOULD,MAY=20or=20remove=20it? |Sender:=20;
	bh=Qo0DXh1PFfo9XP+VAjK7+3+AHVMJL6kx5hcJFKQAzQk=;
	b=JJ3QecatOBLXk08OrgS5KgsV0X87tmgAoZ1D37e5bEZ6jzDnfsfw/FvC2v6o+4BRNR3Ycmh0
	KQiI392YU7OwSoMN3BVUVufUziMsVagJF0hIuoVzsw7spPTwvIgg/Q0i;
Authentication-Results: sj-dkim-2; header.From=jsalowey@cisco.com; dkim=pass (
	sig from cisco.com/sjdkim2002 verified; ); 
X-Spam-Score: -4.0 (----)
X-Scan-Signature: 2857c5c041d6c02d7181d602c22822c8
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hi Gary,

Comments inline.=20

> -----Original Message-----
> From: gtomlinson@SonicWALL.com [mailto:gtomlinson@SonicWALL.com]=20
> Sent: Monday, September 17, 2007 4:54 PM
> To: nea@ietf.org; Paul_Sangster@symantec.com
> Subject: RE: [Nea] RE: Should we downgrade PA security to a=20
> SHOULD,MAY or remove it?
>=20
> Hello Paul,
>=20
> Parvez and I have been thinking about this quite a bit,=20
> especially since discussing the requirements of PT mutual=20
> authentication with Kaushik, which has lead us to quite a bit=20
> of trust model analysis.
>=20
> Our PA-5 response is inline.  We will be commenting=20
> separately on the PB/PT trust relationship which is germane=20
> to the PT mutual authentication discussion thread.
>=20
> > Date: Sat, 15 Sep 2007 13:29:20 -0700
> > From: "Paul Sangster" <Paul_Sangster@symantec.com>
> > Subject: [Nea] RE: Should we downgrade PA security to a=20
> > SHOULD,	MAY or
> > 	remove it?
> > To: <nea@ietf.org>
> > Message-ID:
> > =09
> > <AB96CED633A7C246BDC661DBEE1CF01FD48F76@TUS1XCHCLUPIN11.enterp
> > rise.veritas.com>
> > =09
> > Content-Type: text/plain; charset=3D"us-ascii"
> >=20
> > Unfortunately, this thread has failed to have traction. =20
> > Since we currently have a MUST for PA security would=20
> someone like to=20
> > assert that this MUST should be lowered and provide some=20
> rationale to=20
> > start this discussion?
> >=20
>=20
>=20
> We'll take a stab at asserting PA-5 should be a [MUST].  Here=20
> is our argument.=20
>=20
> Unfortunately, we've come to the group late in the=20
> requirements game and thus didn't participate in the Prague=20
> discussions and as such don't have the context that lead to=20
> the downgrading of PB authentication and shifting all of this=20
> responsibility to PT and PA respectively. However, we think=20
> the most important question is how does the trust model work. =20
>=20
> If a PV does not utilize PA authentication then it implicitly=20
> is relying upon the PB protocol and/or the PT protocol for=20
> authentication of its associated PC. Such transitive trust is=20
> possible in use cases where a NEA Client can provide=20
> assurances on the authenticity of the PCs it hosts, such as=20
> in systems with immutable software images. However, for use=20
> cases where such assurances can't be made, such as when PCs=20
> can be installed on systems for which the PB Client can't=20
> ascertain authenticity in behalf of the PV, the transitive=20
> trust relationship breaks down. Given such use cases exist,=20

[Joe] OK, but its not clear to that if the NEA client cannot provide
assurances of authenticity for its host that the PC will be a better
place to root trust than the PB or PT.  It would seem that in many
architectures the PT and/or PB will run with higher privilege than the
PC. =20

> we don't see how PA-5 authentication can be dropped. In fact,=20
> thinking about this more, it occurs to us that the minimum=20
> trust model required for NEA is PA-5 with mutual=20
> authentication and trust between the PB Server and its=20
> associated PVs via a (out of WG scope) secure interface.  Our=20
> rationale for this assertion is, if each PV/PC combination=20
> has a cryptographically sound communication channel, then it=20
> isn't possible for a PB Client, or PT stack to spoof said=20
> messages and if PB Server has a secure trusted relationship=20
> with each PV, then PB Server can rely entirely on assertions=20
> from the PV's in composing the complete end point posture state.
> We're not advocating this minimalist model as the only trust=20
> requirements, just stating we think it is the minimum set of=20
> trust requirements.
>=20
> Now comes the semantic fun.  What is the meaning of [MUST],=20
> [SHOULD] and [MAY] in the context of PA-5?  Our assertion is=20
> PA-5 is a protocol requirement and as such, given our=20
> assertion there is a need for PA-5 mutual authentication,=20
> then the PA protocol [MUST] be capable of providing mutual=20
> authentication.  Since, the need for mutual authentication is=20
> use case dependent, we assert implementation of it is an=20
> optional requirement and therefore think a [MAY]=20
> implementation requirement is appropriate to consider in the=20
> forthcoming PA protocol RFC work.
>=20
[Joe] In looking at the complexity of managing keys and cryptography at
the PA I think most deployments will opt for managing trust at the PB or
PT layer.  I would rather see the working group focus on things that
will be deployed as a higher priority requirement. =20

> [omitted]
>=20
> Regards,
>=20
> Gary
>=20
>=20
> _______________________________________________
> Nea mailing list
> Nea@ietf.org
> https://www1.ietf.org/mailman/listinfo/nea
>=20


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Wed Sep 19 08:27:31 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IXycr-0008UX-C2; Wed, 19 Sep 2007 08:25:53 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IXycq-0008US-9q
	for nea-confirm+ok@megatron.ietf.org; Wed, 19 Sep 2007 08:25:52 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IXycp-0008UK-M5
	for nea@ietf.org; Wed, 19 Sep 2007 08:25:51 -0400
Received: from michael.checkpoint.com ([194.29.32.68])
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IXyco-0006iP-5p
	for nea@ietf.org; Wed, 19 Sep 2007 08:25:51 -0400
Received: from [91.90.139.90] (localhost [127.0.0.1])
	by michael.checkpoint.com (8.12.10+Sun/8.12.10) with ESMTP id
	l8JCPmjs006766
	for <nea@ietf.org>; Wed, 19 Sep 2007 14:25:48 +0200 (IST)
Message-ID: <46F1154A.4080204@checkpoint.com>
Date: Wed, 19 Sep 2007 14:25:46 +0200
From: Yaron Sheffer <yaronf@checkpoint.com>
User-Agent: Thunderbird 2.0.0.6 (Windows/20070728)
MIME-Version: 1.0
To: nea@ietf.org
X-Spam-Score: 0.0 (/)
X-Scan-Signature: d9238570526f12788af3d33c67f37625
Subject: [Nea] Comments on draft -04
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Content-Type: multipart/mixed; boundary="===============1889439003=="
Errors-To: nea-bounces@ietf.org

This is a multi-part message in MIME format.
--===============1889439003==
Content-Type: multipart/alternative;
	boundary="------------070701090809050501030709"

This is a multi-part message in MIME format.
--------------070701090809050501030709
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Hi,


attached is a (rather long) list of comments. I appreciate the huge 
effort that went into this draft and apologize for being so late with my 
comments. I hope you can consider them for the next revision.


Thanks,

    Yaron

    * Sec. 2: Sessions should be in the scope of the requirements if
      they are relevant to the integrity (security) of the exchange. For
      example if the protocol is susceptible to message reordering. See
      last sentence of Sec. 8.1.3.
    * Sec. 2: "Owner" - the language is a bit convoluted. "The legal,
      rightful possessor", followed by "even if the asset is not within
      the owner's possession".
    * Sec. 2: the "user" is a person, not an entity.
    * Sec. 3.1: the first (positive) part of the Scope should be
      normative, since it sets the stage for the more detailed Sec. 7.
      Use stronger language, rather than "the WG intends".
    * Sec. 3.1: even if PT is/are defined in other workgroups, a
      NEA-specific PT profile may be specified within NEA for some
      particular PT protocol(s). It would be clearer to say that NEA
      prefers to use existing protocols for PT (this is in fact
      mentioned elsewhere).
    * General: the word "privacy" is used quite liberally throughout the
      document. In most cases when it is used, the problem is much more
      that of security exposure than of privacy, in the usual sense of
      the word. Most people would not consider their OS version and
      build number to be privacy sensitive, but this information is
      certainly security sensitive, since it allows an adversary to
      tailor an attack to the particular OS.
    * Sec. 4: what is "gateway access"?
    * Sec. 5: although the "vertical lines" are out of scope, it is
      important to note which ones are expected to be available to
      component developers.
    * Sec. 5: the paragraph about discovery is out of sync with the
      common 802.1X deployment. In that case, the client is not
      configured with the server address, it just talks to the
      enforcement point. On a different level, is there discussion on
      the discovery of protocol peers at the PA/PB layers?
    * Sec. 5.1.1.2: there may be scenarios where the Posture Broker
      Client should receive remediation attributes and handle them
      itself, e.g. by restarting (or even installling) individual
      Posture Collectors.
    * General: the model is unclear on whether there is a 1:1
      correspondence between PC and PV, or whether there may be one PV
      communicating with multiple PC etc.
    * General: it is assumed here that there is one global yes/no policy
      decision. But consider the following scenario: a computer is
      evaluated against two policies, a Contrator policy and an Employee
      policy. It is compliant to the first, and non-compliant to the
      second (and receives remediation instructions). The Posture Broker
      and/or the user may decide whether it wants to remediate its
      posture to be compliant with the stricter policy.
    * Sec. 6.1.1.2, step (15): typo, "the each".
    * Sec. 6.1.2.1: I wonder about this use case, since in a "shared
      terminal" environment, the most straightforward attack would be to
      simply replace the GUI front end, causing it to always report
      success to the user. I believe the general NEA model breaks in
      such an extremely insecure environment.
    * Sec. 6.2.1.2: I would expect such a policy-driven reassessment to
      be initiated by the server, rather than a client-based agent. This
      would also be more secure. Moreover, user interaction may not be
      appropriate in this scenario (or any other "reassessment"
      scenario), as there's nothing to guarantee that there's a person
      there.
    * Sec. 7: the WG certainly needs to define the attribute format, as
      well as a good number of attributes. Why are there no requirements
      on the format and the particular attributes?
    * PA-5 (security of PA protocol): security at this level is probably
      infeasible, especially if PB entities should be able to intercept
      PA messages (not just to mux and demux them). This is also related
      to the comment above regarding 1:1 correspondence between PC and PV.
    * PA-6: I propose to extend this requirement: The PA protocol MUST
      be capable of carrying *fully typed* attributes, including
      non-binary and binary data, large binary objects and encrypted
      content.
    * PA-*: there is also addressing at this level, why are there no
      related requirements? PA requirements should not be relegated to
      the PB section.
    * PB-*: I propose to add: PB MUST provide reliable flow control,
      ensuring that the failure of a single PC or PV does not impact
      other such components.
    * Sec. 7.4: this is implying we will end up with a single PT
      protocol, which quite magically can be embedded in IKE and 802.1X
      and also carried in raw form over TCP. This is very optimistic.
      There will probably be several PT *profiles* (for IKE and 802.1X,
      for example), and a similar but different protocol for running on
      UDP or TCP.
    * Sec. 8.1.3: this section should state whether there is trust
      between different PC components on a single client, and between
      different PV components on a server. This would influence the
      security requirements at the PA level.


--------------070701090809050501030709
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html style="direction: ltr;">
<head>
</head>
<body style="direction: ltr;" bgcolor="#ffffff" text="#000000">
<p style="margin-bottom: 0cm; margin-top: 0pt;">Hi,</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;"><br>
</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;">attached is a (rather
long) list of comments. I appreciate the huge effort that went into
this draft and apologize for being so late with my comments. I hope you
can consider them for the next revision.</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;"><br>
</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;">Thanks,</p>
<p style="margin-bottom: 0cm; margin-top: 0pt;">&nbsp;&nbsp;&nbsp; Yaron<br>
</p>
<ul>
  <li>Sec. 2: Sessions should be in the scope of the requirements if
they are relevant to the integrity (security) of the exchange. For
example if the protocol is susceptible to message reordering. See last
sentence of Sec. 8.1.3.</li>
  <li>Sec. 2: "Owner" - the language is a bit convoluted. "The legal,
rightful possessor", followed by "even if the asset is not within the
owner's possession".</li>
  <li>Sec. 2: the "user" is a person, not an entity.</li>
  <li>Sec. 3.1: the first (positive) part of the Scope should be
normative, since it sets the stage for the more detailed Sec. 7. Use
stronger language, rather than "the WG intends".</li>
  <li>Sec. 3.1: even if PT is/are defined in other workgroups, a
NEA-specific PT profile may be specified within NEA for some particular
PT protocol(s). It would be clearer to say that NEA prefers to use
existing protocols for PT (this is in fact mentioned elsewhere). </li>
  <li>General: the word "privacy" is used quite liberally throughout
the document. In most cases when it is used, the problem is much more
that of security exposure than of privacy, in the usual sense of the
word. Most people would not consider their OS version and build number
to be privacy sensitive, but this information is certainly security
sensitive, since it allows an adversary to tailor an attack to the
particular OS.</li>
  <li>Sec. 4: what is "gateway access"?</li>
  <li>Sec. 5: although the "vertical lines" are out of scope, it is
important to note which ones are expected to be available to component
developers.</li>
  <li>Sec. 5: the paragraph about discovery is out of sync with the
common 802.1X deployment. In that case, the client is not configured
with the server address, it just talks to the enforcement point. On a
different level, is there discussion on the discovery of protocol peers
at the PA/PB layers?</li>
  <li>Sec. 5.1.1.2: there may be scenarios where the Posture Broker
Client should receive remediation attributes and handle them itself,
e.g. by restarting (or even installling) individual Posture Collectors.</li>
  <li>General: the model is unclear on whether there is a 1:1
correspondence between PC and PV, or whether there may be one PV
communicating with multiple PC etc.</li>
  <li>General: it is assumed here that there is one global yes/no
policy decision. But consider the following scenario: a computer is
evaluated against two policies, a Contrator policy and an Employee
policy. It is compliant to the first, and non-compliant to the second
(and receives remediation instructions). The Posture Broker and/or the
user may decide whether it wants to remediate its posture to be
compliant with the stricter policy.</li>
  <li>Sec. 6.1.1.2, step (15): typo, "the each".</li>
  <li>Sec. 6.1.2.1: I wonder about this use case, since in a "shared
terminal" environment, the most straightforward attack would be to
simply replace the GUI front end, causing it to always report success
to the user. I believe the general NEA model breaks in such an
extremely insecure environment.</li>
  <li>Sec. 6.2.1.2: I would expect such a policy-driven reassessment to
be initiated by the server, rather than a client-based agent. This
would also be more secure. Moreover, user interaction may not be
appropriate in this scenario (or any other "reassessment" scenario), as
there's nothing to guarantee that there's a person there.</li>
  <li>Sec. 7: the WG certainly needs to define the attribute format, as
well as a good number of attributes. Why are there no requirements on
the format and the particular attributes?</li>
  <li>PA-5 (security of PA protocol): security at this level is
probably infeasible, especially if PB entities should be able to
intercept PA messages (not just to mux and demux them). This is also
related to the comment above regarding 1:1 correspondence between PC
and PV.</li>
  <li>PA-6: I propose to extend this requirement: The PA protocol MUST
be capable of carrying *fully typed* attributes, including non-binary
and binary data, large binary objects and encrypted content.</li>
  <li>PA-*: there is also addressing at this level, why are there no
related requirements? PA requirements should not be relegated to the PB
section.</li>
  <li>PB-*: I propose to add: PB MUST provide reliable flow control,
ensuring that the failure of a single PC or PV does not impact other
such components.</li>
  <li>Sec. 7.4: this is implying we will end up with a single PT
protocol, which quite magically can be embedded in IKE and 802.1X and
also carried in raw form over TCP. This is very optimistic. There will
probably be several PT *profiles* (for IKE and 802.1X, for example),
and a similar but different protocol for running on UDP or TCP.</li>
  <li>Sec. 8.1.3: this section should state whether there is trust
between different PC components on a single client, and between
different PV components on a server. This would influence the security
requirements at the PA level.</li>
</ul>
</body>
</html>

--------------070701090809050501030709--



--===============1889439003==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea

--===============1889439003==--





From nea-bounces@ietf.org Thu Sep 20 12:58:52 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IYPLc-0005Fn-67; Thu, 20 Sep 2007 12:57:52 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IYPLa-0005Ea-K2
	for nea-confirm+ok@megatron.ietf.org; Thu, 20 Sep 2007 12:57:50 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IYPLZ-0005EI-Vf
	for nea@ietf.org; Thu, 20 Sep 2007 12:57:50 -0400
Received: from sj-iport-6.cisco.com ([171.71.176.117])
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IYPLY-000554-CX
	for nea@ietf.org; Thu, 20 Sep 2007 12:57:49 -0400
X-IronPort-AV: E=Sophos;i="4.20,279,1186383600"; 
	d="scan'208,217";a="221922962"
Received: from sj-dkim-3.cisco.com ([171.71.179.195])
	by sj-iport-6.cisco.com with ESMTP; 20 Sep 2007 09:57:22 -0700
Received: from sj-core-5.cisco.com (sj-core-5.cisco.com [171.71.177.238])
	by sj-dkim-3.cisco.com (8.12.11/8.12.11) with ESMTP id l8KGvMR8018637; 
	Thu, 20 Sep 2007 09:57:22 -0700
Received: from xbh-sjc-211.amer.cisco.com (xbh-sjc-211.cisco.com
	[171.70.151.144])
	by sj-core-5.cisco.com (8.12.10/8.12.6) with ESMTP id l8KGv2E5025344;
	Thu, 20 Sep 2007 16:57:22 GMT
Received: from xmb-sjc-22d.amer.cisco.com ([128.107.191.68]) by
	xbh-sjc-211.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.1830); 
	Thu, 20 Sep 2007 09:57:20 -0700
Received: from 10.21.116.156 ([10.21.116.156]) by xmb-sjc-22d.amer.cisco.com
	([128.107.191.68]) with Microsoft Exchange Server HTTP-DAV ; 
	Thu, 20 Sep 2007 16:57:19 +0000
User-Agent: Microsoft-Entourage/11.3.6.070618
Date: Thu, 20 Sep 2007 09:57:18 -0700
Subject: Re: [Nea] Comments on draft -04
From: Kaushik Narayan <kaushik@cisco.com>
To: Yaron Sheffer <yaronf@checkpoint.com>, <nea@ietf.org>
Message-ID: <C317F47E.B71B%kaushik@cisco.com>
Thread-Topic: [Nea] Comments on draft -04
Thread-Index: Acf7p01mi80wZGeaEdyOMQAX8tamGQ==
In-Reply-To: <46F1154A.4080204@checkpoint.com>
Mime-version: 1.0
X-OriginalArrivalTime: 20 Sep 2007 16:57:20.0951 (UTC)
	FILETIME=[4F287470:01C7FBA7]
DKIM-Signature: v=0.5; a=rsa-sha256; q=dns/txt; l=33109; t=1190307442;
	x=1191171442; c=relaxed/simple; s=sjdkim3002;
	h=Content-Type:From:Subject:Content-Transfer-Encoding:MIME-Version;
	d=cisco.com; i=kaushik@cisco.com;
	z=From:=20Kaushik=20Narayan=20<kaushik@cisco.com>
	|Subject:=20Re=3A=20[Nea]=20Comments=20on=20draft=20-04
	|Sender:=20; bh=BMA9SrarZAfK2UPt3f9GWUp5TeuGNTfyX0uRSuMpWu8=;
	b=JbGwfXwynZwSBiIU+YhdhHpsEXH63ByYmxrRVCnKmITqOCOAuWrnkNMYRiJxHzlbpq586i/n
	4M02CwnzK6tpxNfQlEXkdeEoqb3tTKyLrJdbGUgiO/edN6lVWjQFksqc;
Authentication-Results: sj-dkim-3; header.From=kaushik@cisco.com; dkim=pass (
	sig from cisco.com/sjdkim3002 verified; ); 
X-Spam-Score: 0.0 (/)
X-Scan-Signature: e02509f7a15c0aaea12155beed3396f0
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Content-Type: multipart/mixed; boundary="===============0911364686=="
Errors-To: nea-bounces@ietf.org

> This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

--===============0911364686==
Content-type: multipart/alternative;
	boundary="B_3273127038_1853930"

> This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

--B_3273127038_1853930
Content-type: text/plain;
	charset="ISO-8859-1"
Content-transfer-encoding: quoted-printable

Hi Yaron,

Thanks for your comments. I haven=B9t commented on your editorial suggestions=
,
we will review the relevant text in light of your editorial suggestions and
update appropriately

Please find my reply inline.

<snipped>
>=20
>    =20
> Sec. 2: Sessions should be in the scope of the requirements if they are
> relevant to the integrity (security) of the exchange. For example if the
> protocol is susceptible to message reordering. See last sentence of Sec.
> 8.1.3.=20
>=20
> <Kaushik>=20
>=20
> The term session is used in this document to represent the context shared
> between the Posture Broker Client and Posture Broker Server.  The require=
ments
> leave it to the PB protocol design to define the specifics of the context
> shared at the PB layer and the security implications of a =B3session=B2 at th=
e PB
> layer won=B9t be clear unless the specifics of the session are defined. You=
 are
> right in pointing out that Section 8.1.3 talks about protection of sessio=
ns
> but the real intent of that text is to indicate the need for end-to-end
> security between the PA & PB layers in the event the NEA server side
> components are not co-located.  We will edit text in section 8.1.3 to cla=
rify
> the usage of the term session.
>=20
> </Kaushik>
>=20
>=20
> * Sec. 2: "Owner" - the language is a bit convoluted. "The legal, rightfu=
l
> possessor", followed by "even if the asset is not within the owner's
> possession".=20
> * Sec. 2: the "user" is a person, not an entity.
> * Sec. 3.1: the first (positive) part of the Scope should be normative, s=
ince
> it sets the stage for the more detailed Sec. 7. Use stronger language, ra=
ther
> than "the WG intends".
     =20
        =20


> * Sec. 3.1: even if PT is/are defined in other workgroups, a NEA-specific=
 PT
> profile may be specified within NEA for some particular PT protocol(s). I=
t
> would be clearer to say that NEA prefers to use existing protocols for PT
> (this is in fact mentioned elsewhere).

           <Kaushik>

            I am not sure about the NEA-specific PT profile since that migh=
t
depend on the specific transport protocol in question. You have a valid
point regarding the lack of text to
            indicate that we plan to use existing protocols for PT although
the examples in several sections (including PT requirements in section 7.4)
call out existing protocols such as TLS, EAP etc. .
            We will edit the text in section 3.1 to make this clear.

             </Kaushik>

> * General: the word "privacy" is used quite liberally throughout the docu=
ment.
> In most cases when it is used, the problem is much more that of security
> exposure than of privacy, in the usual sense of the word. Most people wou=
ld
> not consider their OS version and build number to be privacy sensitive, b=
ut
> this information is certainly security sensitive, since it allows an adve=
rsary
> to tailor an attack to the particular OS.
> * Sec. 4: what is "gateway access"?

              <Kaushik>

             I think the term =B3gateway access=B2 refers to points of network
interconnect within an enterprise, for e.g. A branch office being connected
to the central office via a gateway device.

            </Kaushik>

> * Sec. 5: although the "vertical lines" are out of scope, it is important=
 to
> note which ones are expected to be available to component developers.

            <Kaushik>

          Agreed, I am however not sure that this needs to be part of the
requirements document.
      =20
           </Kaushik>

> * Sec. 5: the paragraph about discovery is out of sync with the common 80=
2.1X
> deployment. In that case, the client is not configured with the server
> address, it just talks to the enforcement point. On a different level, is
> there discussion on the discovery of protocol peers at the PA/PB layers?

           <Kaushik>
=20
              Fair point. The reference model does not require the posture
transport to provide mechanisms for discovery, it is definitely simplifies
deployments if a particular PT provides such a
             capability, i.e. 802.1x as you point out. We will add text to
clarify this. The aren=B9t any requirements for PA & PB peers to discover eac=
h
other.
                  =20
          </Kaushik>
                  =20
> * Sec. 5.1.1.2: there may be scenarios where the Posture Broker Client sh=
ould
> receive remediation attributes and handle them itself, e.g. by restarting=
 (or
> even installling) individual Posture Collectors.

             <Kaushik>
            =20
             Agreed. Doesn=B9t the following text  from 5.1.1.2 capture that
=B3 The Posture Broker Client also handles the global assessment  decision
from the Posture Broker Server and may
             interact with  the user communicate the global assessment
decision and aid  in any necessary remediation steps.=B2
=20
             </Kaushik>



> * General: the model is unclear on whether there is a 1:1 correspondence
> between PC and PV, or whether there may be one PV communicating with mult=
iple
> PC etc.=20

            <Kaushik>


           The model is not limited to a 1:1 between a PC & PV. Here is tex=
t
from Section 5.1.2.1 =B3A Posture Validator is responsible for handling
Posture  Attributes from corresponding Posture
           Collector(s).  A Posture  Validator can handle Posture Attribute=
s
from one or more Posture  Collector(s) and vice-versa.=B2

         </Kaushik>


> * General: it is assumed here that there is one global yes/no policy deci=
sion.
> But consider the following scenario: a computer is evaluated against two
> policies, a Contrator policy and an Employee policy. It is compliant to t=
he
> first, and non-compliant to the second (and receives remediation
> instructions). The Posture Broker and/or the user may decide whether it w=
ants
> to remediate its posture to be compliant with the stricter policy.

             <Kaushik>

          There is an assumption that the posture broker produces one globa=
l
assessment decision per assessment. This decision need not be a yes/no, it
could be a grade/state (from healthy to
          infected). The scenario you point out can be implemented by a
cascading set of assessments.

       </Kaushik>



> * Sec. 6.1.1.2, step (15): typo, "the each".
> * Sec. 6.1.2.1: I wonder about this use case, since in a "shared terminal=
"
> environment, the most straightforward attack would be to simply replace t=
he
> GUI front end, causing it to always report success to the user. I believe=
 the
> general NEA model breaks in such an extremely insecure environment.

            <Kaushik>

  =20
              I agree that assessment in a shared environment is not
foolproof to attacks on the client. The example was to point out voluntary
assessment triggered by the user to protect the
            endpoint he is using. A slight variation to this example that
might be a student who is just back from using his laptop to browse in a
public network (kiosk)  & initiates an assessment
             before he connects to the university network to start his work=
.
 =20
        </Kaushik>
       =20

> * Sec. 6.2.1.2: I would expect such a policy-driven reassessment to be
> initiated by the server, rather than a client-based agent. This would als=
o be
> more secure. Moreover, user interaction may not be appropriate in this
> scenario (or any other "reassessment" scenario), as there's nothing to
> guarantee that there's a person there.

              <Kaushik>

            The policy is defined on the server in this example, the client
agent is useful in detecting a change to the client configuration and
initiating a re-assessment; the only way this change can be detected by a
server-only solution is via periodic re-assessment. The user interaction ma=
y
be necessary to reset the non-compliant configuration, i.e. Turn on the
desktop firewall in this example.

   </Kaushik>


> * Sec. 7: the WG certainly needs to define the attribute format, as well =
as a
> good number of attributes. Why are there no requirements on the format an=
d the
> particular attributes?

<Kaushik> This will be part of PA protocol specification. </Kaushik>

> * PA-5 (security of PA protocol): security at this level is probably
> infeasible, especially if PB entities should be able to intercept PA mess=
ages
> (not just to mux and demux them). This is also related to the comment abo=
ve
> regarding 1:1 correspondence between PC and PV.

<Kaushik> I am not sure that PA-5 implies that all of the PA data will
enveloped; PA security must ensure that there is enough data in the clear t=
o
allow for the PB layer to appropriately route the attribute messages. There
is also another thread on the WG mailing list that questions whether PA
security needs to be a MUST, do you have any input to that discussion ?
</Kaushik>

> * PA-6: I propose to extend this requirement: The PA protocol MUST be cap=
able
> of carrying *fully typed* attributes, including non-binary and binary dat=
a,
> large binary objects and encrypted content.
<Kaushik> I would consider large binary objects & encrypted content as
examples of binary data, We would like to avoid additional text in normativ=
e
requirements unless it is necessary </Kaushik>



> * PA-*: there is also addressing at this level, why are there no related
> requirements? PA requirements should not be relegated to the PB section.
> *=20
<Kaushik> We did not want to specify normative schemes for
addressing/identification at the PA layer in the requirements document sinc=
e
it might be better to address that in PA protocol design. We do have text i=
n
the reference model that provides guidance to protocol design. Here is the
text I am referring to =B3Each Posture Validator will be associated with one
or more  identifiers that enable it to be the specified as the  destination
in a PA message.  The Posture Broker Server uses  this identifier to route
messages to this Validator.  This  identifier might be dynamic (e.g.
generated by the Posture   Broker Server at run-time during registration) o=
r
more static   (e.g. pre-assigned to a Posture Validator at install-time and
passed to the Posture Broker Server during registration) or a   function of
the attribute messages the validator desires to  receive (e.g. message type
for subscription). </Kaushik>



> * PB-*: I propose to add: PB MUST provide reliable flow control, ensuring=
 that
> the failure of a single PC or PV does not impact other such components.

<Kaushik> I am not sure whether failure of  a PC/PV creates any requirement=
s
for the over-the-wire PB protocol between the NEA client and NEA server. Ca=
n
you elaborate on this ? </Kaushik>


> * Sec. 7.4: this is implying we will end up with a single PT protocol, wh=
ich
> quite magically can be embedded in IKE and 802.1X and also carried in raw=
 form
> over TCP. This is very optimistic. There will probably be several PT
> *profiles* (for IKE and 802.1X, for example), and a similar but different
> protocol for running on UDP or TCP.
<Kaushik> A mandatory to implement PT is mandated by the WG charter. I agre=
e
that NEA protocols realistically will need to utilize multiple PTs but I
guess the first step is to define the NEA with one specific PT protocol
</Kaushik>=20


=20
> * Sec. 8.1.3: this section should state whether there is trust between
> different PC components on a single client, and between different PV
> components on a server. This would influence the security requirements at=
 the
> PA level.=20

<Kaushik> The PV and PC components are expected to be independent, this
probably needs to be clarified in Section 8.1 </Kaushik>

Regards,
  kaushik
>=20
>=20
> _______________________________________________
> Nea mailing list
> Nea@ietf.org
> https://www1.ietf.org/mailman/listinfo/nea



--B_3273127038_1853930
Content-type: text/html;
	charset="ISO-8859-1"
Content-transfer-encoding: quoted-printable

<HTML>
<HEAD>
<TITLE>Re: [Nea] Comments on draft -04</TITLE>
</HEAD>
<BODY>
<FONT FACE=3D"Verdana, Helvetica, Arial"><SPAN STYLE=3D'font-size:12.0px'>Hi Ya=
ron,<BR>
<BR>
Thanks for your comments. I haven&#8217;t commented on your editorial sugge=
stions, we will review the relevant text in light of your editorial suggesti=
ons and update appropriately <BR>
<BR>
Please find my reply inline.<BR>
<BR>
&lt;snipped&gt;<BR>
</SPAN></FONT><BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPAN STYL=
E=3D'font-size:12.0px'><BR>
&nbsp;&nbsp;&nbsp;&nbsp;<BR>
Sec. 2: Sessions should be in the scope of the requirements if they are rel=
evant to the integrity (security) of the exchange. For example if the protoc=
ol is susceptible to message reordering. See last sentence of Sec. 8.1.3. <B=
R>
<BR>
&lt;Kaushik&gt; <BR>
<BR>
The term session is used in this document to represent the context shared b=
etween the Posture Broker Client and Posture Broker Server. &nbsp;The requir=
ements leave it to the PB protocol design to define the specifics of the con=
text shared at the PB layer and the security implications of a &#8220;sessio=
n&#8221; at the PB layer won&#8217;t be clear unless the specifics of the se=
ssion are defined. You are right in pointing out that Section 8.1.3 talks ab=
out protection of sessions but the real intent of that text is to indicate t=
he need for end-to-end security between the PA &amp; PB layers in the event =
the NEA server side components are not co-located. &nbsp;We will edit text i=
n section 8.1.3 to clarify the usage of the term session.<BR>
<BR>
&lt;/Kaushik&gt;<BR>
<BR>
<BR>
</SPAN></FONT><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><SPAN STYLE=3D'f=
ont-size:12.0px'>Sec. 2: &quot;Owner&quot; - the language is a bit convolute=
d. &quot;The legal, rightful possessor&quot;, followed by &quot;even if the =
asset is not within the owner's possession&quot;.
</SPAN></FONT><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><SPAN STYLE=3D'font-=
size:12.0px'>Sec. 2: the &quot;user&quot; is a person, not an entity.=20
</SPAN></FONT><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><SPAN STYLE=3D'font-=
size:12.0px'>Sec. 3.1: the first (positive) part of the Scope should be norm=
ative, since it sets the stage for the more detailed Sec. 7. Use stronger la=
nguage, rather than &quot;the WG intends&quot;. <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'> &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<BR>
<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>Sec. 3.1: even if PT is/are defined in other wo=
rkgroups, a NEA-specific PT profile may be specified within NEA for some par=
ticular PT protocol(s). It would be clearer to say that NEA prefers to use e=
xisting protocols for PT (this is in fact mentioned elsewhere). <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'><BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&lt;Kaush=
ik&gt; <BR>
<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;I a=
m not sure about the NEA-specific PT profile since that might depend on the =
specific transport protocol in question. You have a valid point regarding th=
e lack of text to <BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;ind=
icate that we plan to use existing protocols for PT although the examples in=
 several sections (including PT requirements in section 7.4) call out existi=
ng protocols such as TLS, EAP etc. . <BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;We =
will edit the text in section 3.1 to make this clear. <BR>
<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;&lt;/Kaushik&gt;<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>General: the word &quot;privacy&quot; is used q=
uite liberally throughout the document. In most cases when it is used, the p=
roblem is much more that of security exposure than of privacy, in the usual =
sense of the word. Most people would not consider their OS version and build=
 number to be privacy sensitive, but this information is certainly security =
sensitive, since it allows an adversary to tailor an attack to the particula=
r OS.=20
</SPAN></FONT><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><SPAN STYLE=3D'font-=
size:12.0px'>Sec. 4: what is &quot;gateway access&quot;? <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'><BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;&nbsp;&lt;Kaushik&gt;<BR>
<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;I think the term &#8220;gateway access&#8221; refers to points of network=
 interconnect within an enterprise, for e.g. A branch office being connected=
 to the central office via a gateway device. <BR>
<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&lt=
;/Kaushik&gt;<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>Sec. 5: although the &quot;vertical lines&quot;=
 are out of scope, it is important to note which ones are expected to be ava=
ilable to component developers. <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'><BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&lt=
;Kaushik&gt;<BR>
<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Agreed, I am ho=
wever not sure that this needs to be part of the requirements document.<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&lt;/Kaus=
hik&gt;<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>Sec. 5: the paragraph about discovery is out of=
 sync with the common 802.1X deployment. In that case, the client is not con=
figured with the server address, it just talks to the enforcement point. On =
a different level, is there discussion on the discovery of protocol peers at=
 the PA/PB layers? <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'><BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&lt;Kaush=
ik&gt;<BR>
&nbsp;<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;&nbsp;Fair point. The reference model does not require the posture transp=
ort to provide mechanisms for discovery, it is definitely simplifies deploym=
ents if a particular PT provides such a <BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;capability, i.e. 802.1x as you point out. We will add text to clarify thi=
s. The aren&#8217;t any requirements for PA &amp; PB peers to discover each =
other.<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&lt;/Kaushik&gt=
;<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>Sec. 5.1.1.2: there may be scenarios where the =
Posture Broker Client should receive remediation attributes and handle them =
itself, e.g. by restarting (or even installling) individual Posture Collecto=
rs. <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'><BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;&lt;Kaushik&gt; <BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;Agreed. Doesn&#8217;t the following text &nbsp;from 5.1.1.2 capture that =
&#8220; The Posture Broker Client also handles the global assessment &nbsp;d=
ecision from the Posture Broker Server and may <BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;interact with &nbsp;the user communicate the global assessment decision a=
nd aid &nbsp;in any necessary remediation steps.&#8221; &nbsp;<BR>
&nbsp;<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;&lt;/Kaushik&gt;<BR>
<BR>
<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>General: the model is unclear on whether there =
is a 1:1 correspondence between PC and PV, or whether there may be one PV co=
mmunicating with multiple PC etc. <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'><BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&lt=
;Kaushik&gt;<BR>
<BR>
<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;The model=
 is not limited to a 1:1 between a PC &amp; PV. Here is text from Section 5.=
1.2.1 &#8220;A Posture Validator is responsible for handling Posture &nbsp;A=
ttributes from corresponding Posture <BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Collector=
(s). &nbsp;A Posture &nbsp;Validator can handle Posture Attributes from one =
or more Posture &nbsp;Collector(s) and vice-versa.&#8221;<BR>
<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&lt;/Kaushik&gt;<BR>
<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>General: it is assumed here that there is one g=
lobal yes/no policy decision. But consider the following scenario: a compute=
r is evaluated against two policies, a Contrator policy and an Employee poli=
cy. It is compliant to the first, and non-compliant to the second (and recei=
ves remediation instructions). The Posture Broker and/or the user may decide=
 whether it wants to remediate its posture to be compliant with the stricter=
 policy. <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'><BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;&lt;Kaushik&gt;<BR>
<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;There is an ass=
umption that the posture broker produces one global assessment decision per =
assessment. This decision need not be a yes/no, it could be a grade/state (f=
rom healthy to <BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;infected). The =
scenario you point out can be implemented by a cascading set of assessments.=
<BR>
<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&lt;/Kaushik&gt;<BR>
<BR>
<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>Sec. 6.1.1.2, step (15): typo, &quot;the each&q=
uot;.=20
</SPAN></FONT><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><SPAN STYLE=3D'font-=
size:12.0px'>Sec. 6.1.2.1: I wonder about this use case, since in a &quot;sh=
ared terminal&quot; environment, the most straightforward attack would be to=
 simply replace the GUI front end, causing it to always report success to th=
e user. I believe the general NEA model breaks in such an extremely insecure=
 environment. <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'><BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&lt=
;Kaushik&gt;<BR>
<BR>
&nbsp;&nbsp;&nbsp;<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;&nbsp;I agree that assessment in a shared environment is not foolproof to=
 attacks on the client. The example was to point out voluntary assessment tr=
iggered by the user to protect the <BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;end=
point he is using. A slight variation to this example that might be a studen=
t who is just back from using his laptop to browse in a public network (kios=
k) &nbsp;&amp; initiates an assessment <BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;before he connects to the university network to start his work. <BR>
&nbsp;&nbsp;<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&lt;/Kaushik&gt;<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>Sec. 6.2.1.2: I would expect such a policy-driv=
en reassessment to be initiated by the server, rather than a client-based ag=
ent. This would also be more secure. Moreover, user interaction may not be a=
ppropriate in this scenario (or any other &quot;reassessment&quot; scenario)=
, as there's nothing to guarantee that there's a person there. <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'><BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nb=
sp;&nbsp;&lt;Kaushik&gt;<BR>
<BR>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;The=
 policy is defined on the server in this example, the client agent is useful=
 in detecting a change to the client configuration and initiating a re-asses=
sment; the only way this change can be detected by a server-only solution is=
 via periodic re-assessment. The user interaction may be necessary to reset =
the non-compliant configuration, i.e. Turn on the desktop firewall in this e=
xample.<BR>
<BR>
&nbsp;&nbsp;&nbsp;&lt;/Kaushik&gt;<BR>
<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>Sec. 7: the WG certainly needs to define the at=
tribute format, as well as a good number of attributes. Why are there no req=
uirements on the format and the particular attributes? <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'><BR>
&lt;Kaushik&gt; This will be part of PA protocol specification. &lt;/Kaushi=
k&gt;<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>PA-5 (security of PA protocol): security at thi=
s level is probably infeasible, especially if PB entities should be able to =
intercept PA messages (not just to mux and demux them). This is also related=
 to the comment above regarding 1:1 correspondence between PC and PV. <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'><BR>
&lt;Kaushik&gt; I am not sure that PA-5 implies that all of the PA data wil=
l enveloped; PA security must ensure that there is enough data in the clear =
to allow for the PB layer to appropriately route the attribute messages. The=
re is also another thread on the WG mailing list that questions whether PA s=
ecurity needs to be a MUST, do you have any input to that discussion ? &lt;/=
Kaushik&gt;<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>PA-6: I propose to extend this requirement: The=
 PA protocol MUST be capable of carrying *fully typed* attributes, including=
 non-binary and binary data, large binary objects and encrypted content. <BR=
>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'>&lt;Kaushik&gt; I would consider large binary obj=
ects &amp; encrypted content as examples of binary data, We would like to av=
oid additional text in normative requirements unless it is necessary &lt;/Ka=
ushik&gt;<BR>
<BR>
<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>PA-*: there is also addressing at this level, w=
hy are there no related requirements? PA requirements should not be relegate=
d to the PB section.=20
</SPAN></FONT><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><SPAN STYLE=3D'font-=
size:12.0px'><BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'>&lt;Kaushik&gt; We did not want to specify normat=
ive schemes for addressing/identification at the PA layer in the requirement=
s document since it might be better to address that in PA protocol design. W=
e do have text in the reference model that provides guidance to protocol des=
ign. Here is the text I am referring to &#8220;Each Posture Validator will b=
e associated with one or more &nbsp;identifiers that enable it to be the spe=
cified as the &nbsp;destination in a PA message. &nbsp;The Posture Broker Se=
rver uses &nbsp;this identifier to route messages to this Validator. &nbsp;T=
his &nbsp;identifier might be dynamic (e.g. generated by the Posture &nbsp;&=
nbsp;Broker Server at run-time during registration) or more static &nbsp;&nb=
sp;(e.g. pre-assigned to a Posture Validator at install-time and &nbsp;passe=
d to the Posture Broker Server during registration) or a &nbsp;&nbsp;functio=
n of the attribute messages the validator desires to &nbsp;receive (e.g. mes=
sage type for subscription). &lt;/Kaushik&gt;<BR>
<BR>
<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>PB-*: I propose to add: PB MUST provide reliabl=
e flow control, ensuring that the failure of a single PC or PV does not impa=
ct other such components. <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'><BR>
&lt;Kaushik&gt; I am not sure whether failure of &nbsp;a PC/PV creates any =
requirements for the over-the-wire PB protocol between the NEA client and NE=
A server. Can you elaborate on this ? &lt;/Kaushik&gt;<BR>
<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>Sec. 7.4: this is implying we will end up with =
a single PT protocol, which quite magically can be embedded in IKE and 802.1=
X and also carried in raw form over TCP. This is very optimistic. There will=
 probably be several PT *profiles* (for IKE and 802.1X, for example), and a =
similar but different protocol for running on UDP or TCP.<BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'>&lt;Kaushik&gt; A mandatory to implement PT is ma=
ndated by the WG charter. I agree that NEA protocols realistically will need=
 to utilize multiple PTs but I guess the first step is to define the NEA wit=
h one specific PT protocol &lt;/Kaushik&gt; <BR>
<BR>
<BR>
&nbsp;<BR>
</SPAN></FONT><BLOCKQUOTE><UL><LI><FONT FACE=3D"Verdana, Helvetica, Arial"><S=
PAN STYLE=3D'font-size:12.0px'>Sec. 8.1.3: this section should state whether t=
here is trust between different PC components on a single client, and betwee=
n different PV components on a server. This would influence the security req=
uirements at the PA level. <BR>
</SPAN></FONT></UL></BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPA=
N STYLE=3D'font-size:12.0px'><BR>
&lt;Kaushik&gt; The PV and PC components are expected to be independent, th=
is probably needs to be clarified in Section 8.1 &lt;/Kaushik&gt;<BR>
<BR>
Regards,<BR>
&nbsp;&nbsp;kaushik<BR>
</SPAN></FONT><BLOCKQUOTE><FONT FACE=3D"Verdana, Helvetica, Arial"><SPAN STYL=
E=3D'font-size:12.0px'><BR>
<HR ALIGN=3DCENTER SIZE=3D"3" WIDTH=3D"95%"></SPAN></FONT><SPAN STYLE=3D'font-size:=
12.0px'><FONT FACE=3D"Monaco, Courier New">___________________________________=
____________<BR>
Nea mailing list<BR>
Nea@ietf.org<BR>
https://www1.ietf.org/mailman/listinfo/nea<BR>
</FONT></SPAN></BLOCKQUOTE><SPAN STYLE=3D'font-size:12.0px'><FONT FACE=3D"Monac=
o, Courier New"><BR>
</FONT></SPAN>
</BODY>
</HTML>


--B_3273127038_1853930--



--===============0911364686==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea

--===============0911364686==--





From nea-bounces@ietf.org Fri Sep 21 18:47:39 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IYrGC-0005vn-Mh; Fri, 21 Sep 2007 18:46:08 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IYrGB-0005sr-Ni
	for nea-confirm+ok@megatron.ietf.org; Fri, 21 Sep 2007 18:46:07 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IYrGB-0005iS-5O
	for nea@ietf.org; Fri, 21 Sep 2007 18:46:07 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IYrG4-0005uh-LJ
	for nea@ietf.org; Fri, 21 Sep 2007 18:46:01 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id huh4h20171si for <nea@ietf.org>;
	Fri, 21 Sep 2007 15:43:28 -0700 (envelope-from
	<gtomlinson@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9159)
	with ESMTP; Fri, 21 Sep 2007 15:43:28 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Fri, 21 Sep 2007 15:46:08 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Date: Fri, 21 Sep 2007 15:44:12 -0700
Message-ID: <4DF8F0A275770B4387764386ABEF4D7A133CD2@us0exb10.us.sonicwall.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: PB Trust Model Authentication
Thread-Index: Acf8oO4JPN8Hi5BWTv2IUj/goaoigA==
From: <gtomlinson@SonicWALL.com>
To: <nea@ietf.org>
X-OriginalArrivalTime: 21 Sep 2007 22:46:08.0409 (UTC)
	FILETIME=[334EE090:01C7FCA1]
X-Mlf-Version: 6.0.1.9159
X-Mlf-UniqueId: o200709212243280015379
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 02ec665d00de228c50c93ed6b5e4fc1a
Subject: [Nea] PB Trust Model Authentication
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Sorry to not have posted this comment sooner, but Parvez and I have had
to think long and hard prior to posting. This comment is in part a
continuation from the trust model discussions regarding the requirement
for PA-5 and in part a discussion of mutual authentication of PT-2 in a
previous thread with Kaushik.


The discussion here is, which NEA component is central to the entire
trust model? Parvez and I assert the PB Server is the center of the
trust model as it is responsible for brokering all of the posture
assessments and by extension must provide assurances of posture state to
the network enforcement and network auditing machinery. Conversely, the
PB Client has a duality role, it must represent the interests of both
the NEA server in determining posture but also must protect the privacy
of the user.  As such, one could argue it is equally eligible as the
center of the trust model.  However, given the NEA mission is for
network security and the NEA Server acts entirely in the interests of
the network operator, we conclude it is the true center. As described in
the previous trust model comments regarding PA-5, the minimalist trust
model could involve just a direct trust between the PB Server and its
registered PVs and rely transitively on strong cryptography present in
the PA protocol between the PVs and their PCs.  In practice however,
this simple model isn't likely to be realized for a majority of use
cases due in part to key distribution complexities of PV/PC pairs, PB
Client trust of PB Server for authorized disclosures and run time
environments which preclude PB Client from ascertaining the authenticity
of PC code via an out of scope interface as illustrated in figure 1.
Given these impracticalities, it suggests a strong trust relationship
between PB Server and PB Client is necessary.  Given this need for such
PB mutual Trust, the question becomes should the PB protocol provide its
own authentication, confidentiality and integrity via requirement PB-5
or rely transitively on the PT protocol for mutual authentication of the
PB Server and PB Client credentials? We don't see a practical need for
both the PB (PB-5) and the PT (PT-2) to provide redundant services as
either can be employed to meet the trust relationship and thus think the
WG needs to make a choice on one or the other but not both in order to
simplify the complexities of PB/PT protocol stack requirements.

Heretofore our comments on PT-2 have been based on the assumption a
mixed model was needed in many TLS deployments where PB did its own
authentication and relied on PT (TLS) for confidentiality and integrity.
This seemed natural to us as common practice in SSL VPNs has been for
server side only authentication and upper level protocols doing the
authentication.  However, we've reconsidered our position after
examining available key distribution mechanisms and believe requiring
Mutual authentication for TLS is acceptable. With this in mind, we
retract our previous comments on
PT-2 and would prefer the WG simplify the PB protocol requirements and
rely entirely on PT-2 for PB cryptographic message dialogue; and
additionally request the elimination of requirement PB-5. =20

We still stand by our comments on PA-5 (separate thread) as there are
use cases where the PB Client can't establish sufficient authenticity of
PCs to enable the PB Server to establish transitive trust with the PVs.
However, we do think there are many use cases where it will be possible
to rely entirely on the PT for all cryptographic support including PA
without any the need for optionally employing PA-5 and that this will be
a common deployment.

Regards,
Gary


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Fri Sep 21 19:46:55 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IYsC6-0005ka-6A; Fri, 21 Sep 2007 19:45:58 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IYsC5-0005jY-3o
	for nea-confirm+ok@megatron.ietf.org; Fri, 21 Sep 2007 19:45:57 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IYsC4-0005Q1-Kx
	for nea@ietf.org; Fri, 21 Sep 2007 19:45:56 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IYsBz-0007S7-SR
	for nea@ietf.org; Fri, 21 Sep 2007 19:45:52 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id huhbhg0171sf for <nea@ietf.org>;
	Fri, 21 Sep 2007 16:43:19 -0700 (envelope-from
	<gtomlinson@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9159)
	with ESMTP; Fri, 21 Sep 2007 16:43:19 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Fri, 21 Sep 2007 16:46:00 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Subject: Re: [Nea] Comments on draft -04
Date: Fri, 21 Sep 2007 16:45:54 -0700
Message-ID: <4DF8F0A275770B4387764386ABEF4D7A133CEC@us0exb10.us.sonicwall.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: Re: [Nea] Comments on draft -04
Thread-Index: Acf8qYzvONfInb8WS+aSRBuasEvwIA==
From: <gtomlinson@SonicWALL.com>
To: <nea@ietf.org>
X-OriginalArrivalTime: 21 Sep 2007 23:46:00.0735 (UTC)
	FILETIME=[908052F0:01C7FCA9]
X-Mlf-Version: 6.0.1.9159
X-Mlf-UniqueId: o200709212343190020310
X-Spam-Score: 0.0 (/)
X-Scan-Signature: f66b12316365a3fe519e75911daf28a8
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hi Yaron and Kaushik,=20

I have a couple comments, inline regarding the PA protocol opacity with =
PB.

[snipped]=20

>=20
> > * PA-5 (security of PA protocol): security at this level is
> probably
> > infeasible, especially if PB entities should be able to
> intercept PA
> > messages (not just to mux and demux them). This is also
> related to the
> > comment above regarding 1:1 correspondence between PC and PV.
>=20
> <Kaushik> I am not sure that PA-5 implies that all of the PA data will =

> enveloped; PA security must ensure that there is enough data in the=20
> clear to allow for the PB layer to appropriately route the attribute=20
> messages. There is also another thread on the WG mailing list that=20
> questions whether PA security needs to be a MUST, do you have any=20
> input to that discussion ?
> </Kaushik>
>=20

<Gary>
PB-2 states PA data must be opaque to the PB protocol.  Is there a need =
for PB to inspect PA data?  I don't know of any and think addressing is =
a property of the interface between a PV/PB Server or PC/PB Client =
respectively.  Please refer to my subsequent comment regarding =
addressing for rationale of this assertion.
</Gary>

[snipped]

>=20
>=20
> > * PA-*: there is also addressing at this level, why are there no=20
> > related requirements? PA requirements should not be
> relegated to the PB section.
> > *
> <Kaushik> We did not want to specify normative schemes for=20
> addressing/identification at the PA layer in the requirements document =

> since it might be better to address that in PA protocol design. We do=20
> have text in the reference model that provides guidance to protocol=20
> design. Here is the text I am referring to =B3Each Posture Validator=20
> will be associated with one or more  identifiers that enable it to be=20
> the specified as the  destination in a PA message.  The Posture Broker =

> Server uses  this identifier to route messages to this Validator. =20
> This  identifier might be dynamic (e.g.
> generated by the Posture   Broker Server at run-time during=20
> registration) or
> more static   (e.g. pre-assigned to a Posture Validator at=20
> install-time and
> passed to the Posture Broker Server during registration) or a=20
>   function of
> the attribute messages the validator desires to  receive (e.g. message =

> type for subscription). </Kaushik>
>=20
>=20

<Gary>
There is much precedence in the IETF to place routing information not in
the payload but rather in the encapsulating protocol and convey routing =
information
via the interface to the encapsulating protocol.  A couple of examples =
are with TCP=20
and IP. An application utilizing TCP doesn't place the port numbers in =
its payload for
TCP to discover but rather coveys this in the interface.  The same holds =
true with the
TCP to IP interface in which protocol ID and IP addresses conveyed. With =
respect to
PB Identifiers, I assumed the PVs and PCs would be conveying them to PB =
via out of scope
interfaces.

As to local node scope dynamic assignment, I don't see how that would =
work for PB
Identifiers as I assumed these would be used for addressing purposes in =
both directions
(PV->PC) and (PC->PV) and would be assigned by some central authority, =
like say IANA. =20
</Gary>

[snipped]


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Fri Sep 21 19:47:10 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IYsD3-0000F8-SQ; Fri, 21 Sep 2007 19:46:57 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IYsD2-0000BF-Mr
	for nea-confirm+ok@megatron.ietf.org; Fri, 21 Sep 2007 19:46:56 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IYsD2-0000B7-D6
	for nea@ietf.org; Fri, 21 Sep 2007 19:46:56 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IYsD1-0003Nm-1e
	for nea@ietf.org; Fri, 21 Sep 2007 19:46:56 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id huhble0171sl for <nea@ietf.org>;
	Fri, 21 Sep 2007 16:44:12 -0700 (envelope-from
	<panandam@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9159)
	with ESMTP; Fri, 21 Sep 2007 16:44:12 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Fri, 21 Sep 2007 16:46:53 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] RE: Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Date: Fri, 21 Sep 2007 16:46:35 -0700
Message-ID: <D37A04F05A87044A8DCEEC0BF9A3AB1D0FFC01@us0exb10.us.sonicwall.com>
In-Reply-To: <AC1CFD94F59A264488DC2BEC3E890DE504855533@xmb-sjc-225.amer.cisco.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] RE: Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Thread-Index: Acf5hhAxbDGGJzWyTzKSIksmrrqTigAHjzVAAL60y1A=
References: <4DF8F0A275770B4387764386ABEF4D7A100E88@us0exb10.us.sonicwall.com>
	<AC1CFD94F59A264488DC2BEC3E890DE504855533@xmb-sjc-225.amer.cisco.com>
From: <panandam@SonicWALL.com>
To: <nea@ietf.org>
X-OriginalArrivalTime: 21 Sep 2007 23:46:53.0565 (UTC)
	FILETIME=[AFFD8AD0:01C7FCA9]
X-Mlf-Version: 6.0.1.9159
X-Mlf-UniqueId: o200709212344120020540
X-Spam-Score: 0.0 (/)
X-Scan-Signature: b1c41982e167b872076d0018e4e1dc3c
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hi Joseph,

[PA] Comment inline.

Thanks,
Parvez

-----Original Message-----
From: Joseph Salowey (jsalowey) [mailto:jsalowey@cisco.com]=20
Sent: Monday, September 17, 2007 8:44 PM
To: Gary Tomlinson; nea@ietf.org; Paul_Sangster@symantec.com
Subject: RE: [Nea] RE: Should we downgrade PA security to a SHOULD,MAY
or remove it?

Hi Gary,

Comments inline.=20

> -----Original Message-----
> From: gtomlinson@SonicWALL.com [mailto:gtomlinson@SonicWALL.com]=20
> Sent: Monday, September 17, 2007 4:54 PM
> To: nea@ietf.org; Paul_Sangster@symantec.com
> Subject: RE: [Nea] RE: Should we downgrade PA security to a=20
> SHOULD,MAY or remove it?
>=20
> Hello Paul,
>=20
> Parvez and I have been thinking about this quite a bit,=20
> especially since discussing the requirements of PT mutual=20
> authentication with Kaushik, which has lead us to quite a bit=20
> of trust model analysis.
>=20
> Our PA-5 response is inline.  We will be commenting=20
> separately on the PB/PT trust relationship which is germane=20
> to the PT mutual authentication discussion thread.
>=20
> > Date: Sat, 15 Sep 2007 13:29:20 -0700
> > From: "Paul Sangster" <Paul_Sangster@symantec.com>
> > Subject: [Nea] RE: Should we downgrade PA security to a=20
> > SHOULD,	MAY or
> > 	remove it?
> > To: <nea@ietf.org>
> > Message-ID:
> > =09
> > <AB96CED633A7C246BDC661DBEE1CF01FD48F76@TUS1XCHCLUPIN11.enterp
> > rise.veritas.com>
> > =09
> > Content-Type: text/plain; charset=3D"us-ascii"
> >=20
> > Unfortunately, this thread has failed to have traction. =20
> > Since we currently have a MUST for PA security would=20
> someone like to=20
> > assert that this MUST should be lowered and provide some=20
> rationale to=20
> > start this discussion?
> >=20
>=20
>=20
> We'll take a stab at asserting PA-5 should be a [MUST].  Here=20
> is our argument.=20
>=20
> Unfortunately, we've come to the group late in the=20
> requirements game and thus didn't participate in the Prague=20
> discussions and as such don't have the context that lead to=20
> the downgrading of PB authentication and shifting all of this=20
> responsibility to PT and PA respectively. However, we think=20
> the most important question is how does the trust model work. =20
>=20
> If a PV does not utilize PA authentication then it implicitly=20
> is relying upon the PB protocol and/or the PT protocol for=20
> authentication of its associated PC. Such transitive trust is=20
> possible in use cases where a NEA Client can provide=20
> assurances on the authenticity of the PCs it hosts, such as=20
> in systems with immutable software images. However, for use=20
> cases where such assurances can't be made, such as when PCs=20
> can be installed on systems for which the PB Client can't=20
> ascertain authenticity in behalf of the PV, the transitive=20
> trust relationship breaks down. Given such use cases exist,=20

[Joe] OK, but its not clear to that if the NEA client cannot provide
assurances of authenticity for its host that the PC will be a better
place to root trust than the PB or PT.  It would seem that in many
architectures the PT and/or PB will run with higher privilege than the
PC. =20

> we don't see how PA-5 authentication can be dropped. In fact,=20
> thinking about this more, it occurs to us that the minimum=20
> trust model required for NEA is PA-5 with mutual=20
> authentication and trust between the PB Server and its=20
> associated PVs via a (out of WG scope) secure interface.  Our=20
> rationale for this assertion is, if each PV/PC combination=20
> has a cryptographically sound communication channel, then it=20
> isn't possible for a PB Client, or PT stack to spoof said=20
> messages and if PB Server has a secure trusted relationship=20
> with each PV, then PB Server can rely entirely on assertions=20
> from the PV's in composing the complete end point posture state.
> We're not advocating this minimalist model as the only trust=20
> requirements, just stating we think it is the minimum set of=20
> trust requirements.
>=20
> Now comes the semantic fun.  What is the meaning of [MUST],=20
> [SHOULD] and [MAY] in the context of PA-5?  Our assertion is=20
> PA-5 is a protocol requirement and as such, given our=20
> assertion there is a need for PA-5 mutual authentication,=20
> then the PA protocol [MUST] be capable of providing mutual=20
> authentication.  Since, the need for mutual authentication is=20
> use case dependent, we assert implementation of it is an=20
> optional requirement and therefore think a [MAY]=20
> implementation requirement is appropriate to consider in the=20
> forthcoming PA protocol RFC work.
>=20
[Joe] In looking at the complexity of managing keys and cryptography at
the PA I think most deployments will opt for managing trust at the PB or
PT layer.  I would rather see the working group focus on things that
will be deployed as a higher priority requirement. =20


[PA] Gary and I agree with you that most deployments will opt for
managing trust at the PT layer. We still think that the PA protocol must
be capable of providing cryptographic security for at least the
following two reasons. First, a vendor may provide a combined solution
of a Posture Validator and a Posture Collector; they may want the
collected posture data to be opaque, to protect their intellectual
property. Second, without the possibility of using cryptography at the
PA layer, the "vertical lines" (the API lines, as opposed to the
horizontal protocol lines) need to provide strong security guarantees.
On some operating systems, this may be reasonable, but on others, these
guarantees might be non-existent or very difficult to implement. For
these reasons, we think that the PA layer should support the PA-5 [MUST]
for those who have no option but to rely on it.




_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Mon Sep 24 20:41:00 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IZySr-0005rd-JR; Mon, 24 Sep 2007 20:39:49 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IZySq-0005rY-Sx
	for nea-confirm+ok@megatron.ietf.org; Mon, 24 Sep 2007 20:39:48 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IZySq-0005rQ-JS
	for nea@ietf.org; Mon, 24 Sep 2007 20:39:48 -0400
Received: from sj-iport-1-in.cisco.com ([171.71.176.70]
	helo=sj-iport-1.cisco.com) by ietf-mx.ietf.org with esmtp (Exim 4.43)
	id 1IZySo-0003Zd-Du for nea@ietf.org; Mon, 24 Sep 2007 20:39:48 -0400
X-IronPort-AV: E=Sophos;i="4.20,293,1186383600"; d="scan'208";a="19757689"
Received: from sj-dkim-4.cisco.com ([171.71.179.196])
	by sj-iport-1.cisco.com with ESMTP; 24 Sep 2007 17:39:46 -0700
Received: from sj-core-5.cisco.com (sj-core-5.cisco.com [171.71.177.238])
	by sj-dkim-4.cisco.com (8.12.11/8.12.11) with ESMTP id l8P0djU8013367; 
	Mon, 24 Sep 2007 17:39:45 -0700
Received: from xbh-sjc-211.amer.cisco.com (xbh-sjc-211.cisco.com
	[171.70.151.144])
	by sj-core-5.cisco.com (8.12.10/8.12.6) with ESMTP id l8P0daEB022055;
	Tue, 25 Sep 2007 00:39:45 GMT
Received: from xmb-sjc-22d.amer.cisco.com ([128.107.191.68]) by
	xbh-sjc-211.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.1830); 
	Mon, 24 Sep 2007 17:39:36 -0700
Received: from 171.69.75.24 ([171.69.75.24]) by xmb-sjc-22d.amer.cisco.com
	([128.107.191.68]) with Microsoft Exchange Server HTTP-DAV ; 
	Tue, 25 Sep 2007 00:39:36 +0000
User-Agent: Microsoft-Entourage/11.3.6.070618
Date: Mon, 24 Sep 2007 17:39:36 -0700
Subject: Re: [Nea] PB Trust Model Authentication
From: Kaushik Narayan <kaushik@cisco.com>
To: <gtomlinson@SonicWALL.com>, <nea@ietf.org>
Message-ID: <C31DA6D8.BA65%kaushik@cisco.com>
Thread-Topic: [Nea] PB Trust Model Authentication
Thread-Index: Acf8oO4JPN8Hi5BWTv2IUj/goaoigACa54ka
In-Reply-To: <4DF8F0A275770B4387764386ABEF4D7A133CD2@us0exb10.us.sonicwall.com>
Mime-version: 1.0
Content-type: text/plain;
	charset="US-ASCII"
Content-transfer-encoding: 7bit
X-OriginalArrivalTime: 25 Sep 2007 00:39:36.0446 (UTC)
	FILETIME=[8C7421E0:01C7FF0C]
DKIM-Signature: v=0.5; a=rsa-sha256; q=dns/txt; l=4462; t=1190680785;
	x=1191544785; c=relaxed/simple; s=sjdkim4002;
	h=Content-Type:From:Subject:Content-Transfer-Encoding:MIME-Version;
	d=cisco.com; i=kaushik@cisco.com;
	z=From:=20Kaushik=20Narayan=20<kaushik@cisco.com>
	|Subject:=20Re=3A=20[Nea]=20PB=20Trust=20Model=20Authentication
	|Sender:=20; bh=DPqecdVx1EeTcYKtE/IXic+gRP3/iM7CA8ARvX8oVwc=;
	b=AwGulpL637vbNpaFt8KwFkGYDWfUb6i+2EG9Uq1iYaAVkbri4k7wgfG5GBt9ZLhXxlyX8vIs
	sMf4xJE8EKk38KnyGSCi4o5q+kXSPxIsXPVeJPo2MmWnF8MdZAphW5oa;
Authentication-Results: sj-dkim-4; header.From=kaushik@cisco.com; dkim=pass (
	sig from cisco.com/sjdkim4002 verified; ); 
X-Spam-Score: -1.4 (-)
X-Scan-Signature: 3002fc2e661cd7f114cb6bae92fe88f1
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hi Gary,

I just want to confirm my reading of your assessment of the NEA security
requirements

1. The mutual authentication requirements in PT-2 are fine and PT-2 should
be sufficient to provide security for most common deployments of NEA.

2. You are in favor of changing PA-5 from a MUST to a SHOULD to simplify key
management for the PA protocol.

regards,
 kaushik


On 9/21/07 3:44 PM, "gtomlinson@SonicWALL.com" <gtomlinson@SonicWALL.com>
wrote:

> Sorry to not have posted this comment sooner, but Parvez and I have had
> to think long and hard prior to posting. This comment is in part a
> continuation from the trust model discussions regarding the requirement
> for PA-5 and in part a discussion of mutual authentication of PT-2 in a
> previous thread with Kaushik.
> 
> 
> The discussion here is, which NEA component is central to the entire
> trust model? Parvez and I assert the PB Server is the center of theHi G
> trust model as it is responsible for brokering all of the posture
> assessments and by extension must provide assurances of posture state to
> the network enforcement and network auditing machinery. Conversely, the
> PB Client has a duality role, it must represent the interests of both
> the NEA server in determining posture but also must protect the privacy
> of the user.  As such, one could argue it is equally eligible as the
> center of the trust model.  However, given the NEA mission is for
> network security and the NEA Server acts entirely in the interests of
> the network operator, we conclude it is the true center. As described in
> the previous trust model comments regarding PA-5, the minimalist trust
> model could involve just a direct trust between the PB Server and its
> registered PVs and rely transitively on strong cryptography present in
> the PA protocol between the PVs and their PCs.  In practice however,
> this simple model isn't likely to be realized for a majority of use
> cases due in part to key distribution complexities of PV/PC pairs, PB
> Client trust of PB Server for authorized disclosures and run time
> environments which preclude PB Client from ascertaining the authenticity
> of PC code via an out of scope interface as illustrated in figure 1.
> Given these impracticalities, it suggests a strong trust relationship
> between PB Server and PB Client is necessary.  Given this need for such
> PB mutual Trust, the question becomes should the PB protocol provide its
> own authentication, confidentiality and integrity via requirement PB-5
> or rely transitively on the PT protocol for mutual authentication of the
> PB Server and PB Client credentials? We don't see a practical need for
> both the PB (PB-5) and the PT (PT-2) to provide redundant services as
> either can be employed to meet the trust relationship and thus think the
> WG needs to make a choice on one or the other but not both in order to
> simplify the complexities of PB/PT protocol stack requirements.
> 
> Heretofore our comments on PT-2 have been based on the assumption a
> mixed model was needed in many TLS deployments where PB did its own
> authentication and relied on PT (TLS) for confidentiality and integrity.
> This seemed natural to us as common practice in SSL VPNs has been for
> server side only authentication and upper level protocols doing the
> authentication.  However, we've reconsidered our position after
> examining available key distribution mechanisms and believe requiring
> Mutual authentication for TLS is acceptable. With this in mind, we
> retract our previous comments on
> PT-2 and would prefer the WG simplify the PB protocol requirements and
> rely entirely on PT-2 for PB cryptographic message dialogue; and
> additionally request the elimination of requirement PB-5.
> 
> We still stand by our comments on PA-5 (separate thread) as there are
> use cases where the PB Client can't establish sufficient authenticity of
> PCs to enable the PB Server to establish transitive trust with the PVs.
> However, we do think there are many use cases where it will be possible
> to rely entirely on the PT for all cryptographic support including PA
> without any the need for optionally employing PA-5 and that this will be
> a common deployment.
> 
> Regards,
> Gary
> 
> 
> _______________________________________________
> Nea mailing list
> Nea@ietf.org
> https://www1.ietf.org/mailman/listinfo/nea


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Tue Sep 25 09:53:30 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IaApt-0000mw-5B; Tue, 25 Sep 2007 09:52:25 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IaApr-0000mr-Uk
	for nea-confirm+ok@megatron.ietf.org; Tue, 25 Sep 2007 09:52:23 -0400
Received: from [10.91.34.44] (helo=ietf-mx.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IaApr-0000mj-L2
	for nea@ietf.org; Tue, 25 Sep 2007 09:52:23 -0400
Received: from rtp-iport-1.cisco.com ([64.102.122.148])
	by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1IaApl-0005xV-DA
	for nea@ietf.org; Tue, 25 Sep 2007 09:52:23 -0400
X-IronPort-AV: E=Sophos;i="4.20,295,1186372800"; d="scan'208";a="72061516"
Received: from rtp-dkim-2.cisco.com ([64.102.121.159])
	by rtp-iport-1.cisco.com with ESMTP; 25 Sep 2007 09:51:59 -0400
Received: from rtp-core-2.cisco.com (rtp-core-2.cisco.com [64.102.124.13])
	by rtp-dkim-2.cisco.com (8.12.11/8.12.11) with ESMTP id l8PDq3Kb009260; 
	Tue, 25 Sep 2007 09:52:03 -0400
Received: from xbh-rtp-211.amer.cisco.com (xbh-rtp-211.cisco.com
	[64.102.31.102])
	by rtp-core-2.cisco.com (8.12.10/8.12.6) with ESMTP id l8PDpNZ4000060; 
	Tue, 25 Sep 2007 13:51:54 GMT
Received: from xmb-rtp-212.amer.cisco.com ([64.102.31.111]) by
	xbh-rtp-211.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.1830); 
	Tue, 25 Sep 2007 09:51:50 -0400
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] RE: Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Date: Tue, 25 Sep 2007 09:52:34 -0400
Message-ID: <9958B444368E884DBB215F3FEF36F5B7054C1E0E@xmb-rtp-212.amer.cisco.com>
In-Reply-To: <D37A04F05A87044A8DCEEC0BF9A3AB1D0FFC01@us0exb10.us.sonicwall.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] RE: Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Thread-Index: Acf5hhAxbDGGJzWyTzKSIksmrrqTigAHjzVAAL60y1AAtsr8AA==
References: <4DF8F0A275770B4387764386ABEF4D7A100E88@us0exb10.us.sonicwall.com><AC1CFD94F59A264488DC2BEC3E890DE504855533@xmb-sjc-225.amer.cisco.com>
	<D37A04F05A87044A8DCEEC0BF9A3AB1D0FFC01@us0exb10.us.sonicwall.com>
From: "Hao Zhou \(hzhou\)" <hzhou@cisco.com>
To: <panandam@SonicWALL.com>, <nea@ietf.org>
X-OriginalArrivalTime: 25 Sep 2007 13:51:50.0005 (UTC)
	FILETIME=[38A9A250:01C7FF7B]
DKIM-Signature: v=0.5; a=rsa-sha256; q=dns/txt; l=8141; t=1190728323;
	x=1191592323; c=relaxed/simple; s=rtpdkim2001;
	h=Content-Type:From:Subject:Content-Transfer-Encoding:MIME-Version;
	d=cisco.com; i=hzhou@cisco.com;
	z=From:=20=22Hao=20Zhou=20\(hzhou\)=22=20<hzhou@cisco.com>
	|Subject:=20RE=3A=20[Nea]=20RE=3A=20Should=20we=20downgrade=20PA=20securi
	ty=20to=20a=20SHOULD,MAY=20or=20remove=20it? |Sender:=20
	|To:=20<panandam@SonicWALL.com>,=20<nea@ietf.org>;
	bh=tlnOR1Cq/r3L0PrN5svG/TOEKT1Xuc3FiVv5skpP97w=;
	b=mRVvs6V+Eu+Pk9JBquLaz3OSUgPTGBEIBT/RUILS5cTassn5tUfqqOT48j1AGdkypnkzKZfh
	InMgfMhJXVG+6t0SfQMnMmv0O3TTf3TCw+1FfB1P9OmwB/EMsugHCHxX;
Authentication-Results: rtp-dkim-2; header.From=hzhou@cisco.com; dkim=pass (
	sig from cisco.com/rtpdkim2001 verified; ); 
X-Spam-Score: -4.0 (----)
X-Scan-Signature: bf422c85703d3d847fb014987125ac48
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

> [PA] Gary and I agree with you that most deployments will opt=20
> for managing trust at the PT layer. We still think that the=20
> PA protocol must be capable of providing cryptographic=20
> security for at least the following two reasons. First, a=20
> vendor may provide a combined solution of a Posture Validator=20
> and a Posture Collector; they may want the collected posture=20
> data to be opaque, to protect their intellectual property.=20
> Second, without the possibility of using cryptography at the=20
> PA layer, the "vertical lines" (the API lines, as opposed to=20
> the horizontal protocol lines) need to provide strong=20
> security guarantees.
> On some operating systems, this may be reasonable, but on=20
> others, these guarantees might be non-existent or very=20
> difficult to implement. For these reasons, we think that the=20
> PA layer should support the PA-5 [MUST] for those who have no=20
> option but to rely on it.

A vendor can still provide cryptographic security using vendor specific
attribute if they want to. They can define their own key management
scheme and framework. Do we really want to mandate PA protocol to
support cryptographic security, only for a very small percentage of the
use cases, where PA is from one vendor and PA validator is from another
vendor?=20

My vote is [SHOULD] at most, to allow WG to focus on high priority
requirements.

> -----Original Message-----
> From: panandam@SonicWALL.com [mailto:panandam@SonicWALL.com]=20
> Sent: Friday, September 21, 2007 7:47 PM
> To: nea@ietf.org
> Subject: RE: [Nea] RE: Should we downgrade PA security to a=20
> SHOULD,MAY or remove it?
>=20
> Hi Joseph,
>=20
> [PA] Comment inline.
>=20
> Thanks,
> Parvez
>=20
> -----Original Message-----
> From: Joseph Salowey (jsalowey) [mailto:jsalowey@cisco.com]
> Sent: Monday, September 17, 2007 8:44 PM
> To: Gary Tomlinson; nea@ietf.org; Paul_Sangster@symantec.com
> Subject: RE: [Nea] RE: Should we downgrade PA security to a=20
> SHOULD,MAY or remove it?
>=20
> Hi Gary,
>=20
> Comments inline.=20
>=20
> > -----Original Message-----
> > From: gtomlinson@SonicWALL.com [mailto:gtomlinson@SonicWALL.com]
> > Sent: Monday, September 17, 2007 4:54 PM
> > To: nea@ietf.org; Paul_Sangster@symantec.com
> > Subject: RE: [Nea] RE: Should we downgrade PA security to a=20
> SHOULD,MAY=20
> > or remove it?
> >=20
> > Hello Paul,
> >=20
> > Parvez and I have been thinking about this quite a bit, especially=20
> > since discussing the requirements of PT mutual authentication with=20
> > Kaushik, which has lead us to quite a bit of trust model analysis.
> >=20
> > Our PA-5 response is inline.  We will be commenting=20
> separately on the=20
> > PB/PT trust relationship which is germane to the PT mutual=20
> > authentication discussion thread.
> >=20
> > > Date: Sat, 15 Sep 2007 13:29:20 -0700
> > > From: "Paul Sangster" <Paul_Sangster@symantec.com>
> > > Subject: [Nea] RE: Should we downgrade PA security to a=20
> > > SHOULD,	MAY or
> > > 	remove it?
> > > To: <nea@ietf.org>
> > > Message-ID:
> > > =09
> > > <AB96CED633A7C246BDC661DBEE1CF01FD48F76@TUS1XCHCLUPIN11.enterp
> > > rise.veritas.com>
> > > =09
> > > Content-Type: text/plain; charset=3D"us-ascii"
> > >=20
> > > Unfortunately, this thread has failed to have traction. =20
> > > Since we currently have a MUST for PA security would
> > someone like to
> > > assert that this MUST should be lowered and provide some
> > rationale to
> > > start this discussion?
> > >=20
> >=20
> >=20
> > We'll take a stab at asserting PA-5 should be a [MUST]. =20
> Here is our=20
> > argument.
> >=20
> > Unfortunately, we've come to the group late in the=20
> requirements game=20
> > and thus didn't participate in the Prague discussions and as such=20
> > don't have the context that lead to the downgrading of PB=20
> > authentication and shifting all of this responsibility to PT and PA=20
> > respectively. However, we think the most important question is how=20
> > does the trust model work.
> >=20
> > If a PV does not utilize PA authentication then it implicitly is=20
> > relying upon the PB protocol and/or the PT protocol for=20
> authentication=20
> > of its associated PC. Such transitive trust is possible in=20
> use cases=20
> > where a NEA Client can provide assurances on the=20
> authenticity of the=20
> > PCs it hosts, such as in systems with immutable software images.=20
> > However, for use cases where such assurances can't be made, such as=20
> > when PCs can be installed on systems for which the PB Client can't=20
> > ascertain authenticity in behalf of the PV, the transitive trust=20
> > relationship breaks down. Given such use cases exist,
>=20
> [Joe] OK, but its not clear to that if the NEA client cannot=20
> provide assurances of authenticity for its host that the PC=20
> will be a better place to root trust than the PB or PT.  It=20
> would seem that in many architectures the PT and/or PB will=20
> run with higher privilege than the PC. =20
>=20
> > we don't see how PA-5 authentication can be dropped. In=20
> fact, thinking=20
> > about this more, it occurs to us that the minimum trust=20
> model required=20
> > for NEA is PA-5 with mutual authentication and trust between the PB=20
> > Server and its associated PVs via a (out of WG scope) secure=20
> > interface.  Our rationale for this assertion is, if each PV/PC=20
> > combination has a cryptographically sound communication=20
> channel, then=20
> > it isn't possible for a PB Client, or PT stack to spoof=20
> said messages=20
> > and if PB Server has a secure trusted relationship with=20
> each PV, then=20
> > PB Server can rely entirely on assertions from the PV's in=20
> composing=20
> > the complete end point posture state.
> > We're not advocating this minimalist model as the only trust=20
> > requirements, just stating we think it is the minimum set of trust=20
> > requirements.
> >=20
> > Now comes the semantic fun.  What is the meaning of [MUST],=20
> [SHOULD]=20
> > and [MAY] in the context of PA-5?  Our assertion is
> > PA-5 is a protocol requirement and as such, given our=20
> assertion there=20
> > is a need for PA-5 mutual authentication, then the PA=20
> protocol [MUST]=20
> > be capable of providing mutual authentication.  Since, the need for=20
> > mutual authentication is use case dependent, we assert=20
> implementation=20
> > of it is an optional requirement and therefore think a [MAY]=20
> > implementation requirement is appropriate to consider in the=20
> > forthcoming PA protocol RFC work.
> >=20
> [Joe] In looking at the complexity of managing keys and=20
> cryptography at the PA I think most deployments will opt for=20
> managing trust at the PB or PT layer.  I would rather see the=20
> working group focus on things that will be deployed as a=20
> higher priority requirement. =20
>=20
>=20
> [PA] Gary and I agree with you that most deployments will opt=20
> for managing trust at the PT layer. We still think that the=20
> PA protocol must be capable of providing cryptographic=20
> security for at least the following two reasons. First, a=20
> vendor may provide a combined solution of a Posture Validator=20
> and a Posture Collector; they may want the collected posture=20
> data to be opaque, to protect their intellectual property.=20
> Second, without the possibility of using cryptography at the=20
> PA layer, the "vertical lines" (the API lines, as opposed to=20
> the horizontal protocol lines) need to provide strong=20
> security guarantees.
> On some operating systems, this may be reasonable, but on=20
> others, these guarantees might be non-existent or very=20
> difficult to implement. For these reasons, we think that the=20
> PA layer should support the PA-5 [MUST] for those who have no=20
> option but to rely on it.
>=20
>=20
>=20
>=20
> _______________________________________________
> Nea mailing list
> Nea@ietf.org
> https://www1.ietf.org/mailman/listinfo/nea
>=20


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Tue Sep 25 11:28:30 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IaCJr-00033e-Gx; Tue, 25 Sep 2007 11:27:27 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IaCJq-00033N-DH
	for nea-confirm+ok@megatron.ietf.org; Tue, 25 Sep 2007 11:27:26 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IaCJq-0002p9-1g
	for nea@ietf.org; Tue, 25 Sep 2007 11:27:26 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IaCJn-0002fL-Gr
	for nea@ietf.org; Tue, 25 Sep 2007 11:27:23 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id hv4k1g0171sk for <nea@ietf.org>;
	Tue, 25 Sep 2007 08:24:07 -0700 (envelope-from
	<gtomlinson@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9159)
	with ESMTP; Tue, 25 Sep 2007 08:24:07 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Tue, 25 Sep 2007 08:27:00 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] PB Trust Model Authentication
Date: Tue, 25 Sep 2007 08:26:59 -0700
Message-ID: <4DF8F0A275770B4387764386ABEF4D7A133DF2@us0exb10.us.sonicwall.com>
In-Reply-To: <C31DA6D8.BA65%kaushik@cisco.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] PB Trust Model Authentication
Thread-Index: Acf8oO4JPN8Hi5BWTv2IUj/goaoigACa54kaAB2Op9A=
From: <gtomlinson@SonicWALL.com>
To: <kaushik@cisco.com>,
	<nea@ietf.org>
X-OriginalArrivalTime: 25 Sep 2007 15:27:00.0858 (UTC)
	FILETIME=[8498A5A0:01C7FF88]
X-Mlf-Version: 6.0.1.9159
X-Mlf-UniqueId: o200709251524070052110
X-Spam-Score: 0.0 (/)
X-Scan-Signature: 7aafa0432175920a4b3e118e16c5cb64
Cc: 
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

Hi Kaushik,

> -----Original Message-----
> From: Kaushik Narayan [mailto:kaushik@cisco.com]=20
> Sent: Monday, September 24, 2007 5:40 PM
> To: Gary Tomlinson; nea@ietf.org
> Subject: Re: [Nea] PB Trust Model Authentication
>=20
> Hi Gary,
>=20
> I just want to confirm my reading of your assessment of the=20
> NEA security requirements
>=20
> 1. The mutual authentication requirements in PT-2 are fine=20
> and PT-2 should be sufficient to provide security for most=20
> common deployments of NEA.

<Gary>
Yes, PT-2 is sufficient as specified.
</Gary>
>=20
> 2. You are in favor of changing PA-5 from a MUST to a SHOULD=20
> to simplify key management for the PA protocol.

<Gary>
I think configuration of PA-5 at a site is optional and thus would be
a MAY in RFC parlance. However, as a requirement capability for a
standardized=20
protocol to provide I see it as a MUST as the optional behavior is not
for the=20
protocol developers but rather a policy of either a vendor or a network
operator.
This is why I stated PA-5 is MUST for the requirements RFC and a MAY
configuration=20
requirement for the PA protocol RFC. However, the more I think about it,
I think
PA-5 can be reduced to just authentication as I'm pretty sure
confidentiality and
integrity can always be met with PT-2. I find the semantics of MUST,
SHOULD, MAY tricky
between requirements of protocols and guidance to utilization of
protocol capabilities.
We could add in guidance to PA-5 stating its use should be employed when
authentication
of PCs can't be guaranteed by PB Client. Does this make sense?=20
</Gary>

>=20
> regards,
>  kaushik
>=20

Regards
Gary


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



From nea-bounces@ietf.org Tue Sep 25 16:28:23 2007
Return-path: <nea-bounces@ietf.org>
Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com)
	by megatron.ietf.org with esmtp (Exim 4.43)
	id 1IaGzU-0002WG-HF; Tue, 25 Sep 2007 16:26:44 -0400
Received: from nea by megatron.ietf.org with local (Exim 4.43)
	id 1IaGzR-0002Nt-VG
	for nea-confirm+ok@megatron.ietf.org; Tue, 25 Sep 2007 16:26:41 -0400
Received: from [10.90.34.44] (helo=chiedprmail1.ietf.org)
	by megatron.ietf.org with esmtp (Exim 4.43) id 1IaGzQ-0002DW-PK
	for nea@ietf.org; Tue, 25 Sep 2007 16:26:41 -0400
Received: from mail.sonicwall.com ([67.115.118.12] helo=es8000.sonicwall.com)
	by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1IaGzK-0004Ns-Cw
	for nea@ietf.org; Tue, 25 Sep 2007 16:26:35 -0400
Received: from es8000.sonicwall.com (127.0.0.1) by es8000.sonicwall.com
	(MlfMTA v3.2r9) id hv5n380171sc for <nea@ietf.org>;
	Tue, 25 Sep 2007 13:23:15 -0700 (envelope-from
	<panandam@SonicWALL.com>)
Received: from us0exf01.us.sonicwall.com ([10.50.128.221])
	by es8000.sonicwall.com (SonicWALL 6.0.1.9159)
	with ESMTP; Tue, 25 Sep 2007 13:23:15 -0700
Received: from us0exb10.us.sonicwall.com ([192.168.157.202]) by
	us0exf01.us.sonicwall.com with Microsoft SMTPSVC(6.0.3790.3959);
	Tue, 25 Sep 2007 13:26:46 -0700
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [Nea] RE: Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Date: Tue, 25 Sep 2007 13:26:32 -0700
Message-ID: <D37A04F05A87044A8DCEEC0BF9A3AB1D0FFCDC@us0exb10.us.sonicwall.com>
In-Reply-To: <9958B444368E884DBB215F3FEF36F5B7054C1E0E@xmb-rtp-212.amer.cisco.com>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [Nea] RE: Should we downgrade PA security to a SHOULD,
	MAY or remove it?
Thread-Index: Acf5hhAxbDGGJzWyTzKSIksmrrqTigAHjzVAAL60y1AAtsr8AAAJONwA
References: <4DF8F0A275770B4387764386ABEF4D7A100E88@us0exb10.us.sonicwall.com><AC1CFD94F59A264488DC2BEC3E890DE504855533@xmb-sjc-225.amer.cisco.com>
	<D37A04F05A87044A8DCEEC0BF9A3AB1D0FFC01@us0exb10.us.sonicwall.com>
	<9958B444368E884DBB215F3FEF36F5B7054C1E0E@xmb-rtp-212.amer.cisco.com>
From: <panandam@SonicWALL.com>
To: <nea@ietf.org>
X-OriginalArrivalTime: 25 Sep 2007 20:26:46.0276 (UTC)
	FILETIME=[64BDA440:01C7FFB2]
X-Mlf-Version: 6.0.1.9159
X-Mlf-UniqueId: o200709252023150078082
X-Spam-Score: 0.0 (/)
X-Scan-Signature: a4a24b484706be629f915bfb1a3e4771
X-BeenThere: nea@ietf.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Network Endpoint Assessment discussion list <nea.ietf.org>
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www1.ietf.org/pipermail/nea>
List-Post: <mailto:nea@ietf.org>
List-Help: <mailto:nea-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/nea>,
	<mailto:nea-request@ietf.org?subject=subscribe>
Errors-To: nea-bounces@ietf.org

At this point, I think we agree in spirit but it's the exact wording
that worries me.

First, let me state an assumption that may not be universally held. I am
assuming that there is going to be a SINGLE PA protocol, a SINGLE PB
protocol and MULTIPLE PT protocols. The reason I make this assumption is
because I don't see the necessity of having several PA and PB protocols
that accomplish the exact same purpose; since these protocols will be
designed from scratch, a single one of each should suffice.

In particular, I am assuming that the unique PA protocol will carry BOTH
the standard attributes and the vendor proprietary attributes. Thus, the
unique PA protocol will support multiple attribute name spaces.

Now, given that there is a unique PA protocol, if the "MUST provide
authentication, integrity, and confidentiality" requirement for this
protocol is made into a "SHOULD provide authentication, integrity, and
confidentiality" requirement, it is left up to the WG that designs this
unique PA protocol to include this capability or not. While I think that
it SHOULD be left up to the user of the protocol whether to employ this
capability, the unique PA protocol "MUST provide" this capability in
order for the user to have this choice.

I apologize for belaboring the point but this is a very important
requirement and I'd like to make sure I understand the implications.

Thanks,
Parvez Anandam


-----Original Message-----
From: Hao Zhou (hzhou) [mailto:hzhou@cisco.com]=20
Sent: Tuesday, September 25, 2007 6:53 AM
To: Parvez Anandam; nea@ietf.org
Subject: RE: [Nea] RE: Should we downgrade PA security to a SHOULD,MAY
or remove it?

> [PA] Gary and I agree with you that most deployments will opt=20
> for managing trust at the PT layer. We still think that the=20
> PA protocol must be capable of providing cryptographic=20
> security for at least the following two reasons. First, a=20
> vendor may provide a combined solution of a Posture Validator=20
> and a Posture Collector; they may want the collected posture=20
> data to be opaque, to protect their intellectual property.=20
> Second, without the possibility of using cryptography at the=20
> PA layer, the "vertical lines" (the API lines, as opposed to=20
> the horizontal protocol lines) need to provide strong=20
> security guarantees.
> On some operating systems, this may be reasonable, but on=20
> others, these guarantees might be non-existent or very=20
> difficult to implement. For these reasons, we think that the=20
> PA layer should support the PA-5 [MUST] for those who have no=20
> option but to rely on it.

A vendor can still provide cryptographic security using vendor specific
attribute if they want to. They can define their own key management
scheme and framework. Do we really want to mandate PA protocol to
support cryptographic security, only for a very small percentage of the
use cases, where PA is from one vendor and PA validator is from another
vendor?=20

My vote is [SHOULD] at most, to allow WG to focus on high priority
requirements.

> -----Original Message-----
> From: panandam@SonicWALL.com [mailto:panandam@SonicWALL.com]=20
> Sent: Friday, September 21, 2007 7:47 PM
> To: nea@ietf.org
> Subject: RE: [Nea] RE: Should we downgrade PA security to a=20
> SHOULD,MAY or remove it?
>=20
> Hi Joseph,
>=20
> [PA] Comment inline.
>=20
> Thanks,
> Parvez
>=20
> -----Original Message-----
> From: Joseph Salowey (jsalowey) [mailto:jsalowey@cisco.com]
> Sent: Monday, September 17, 2007 8:44 PM
> To: Gary Tomlinson; nea@ietf.org; Paul_Sangster@symantec.com
> Subject: RE: [Nea] RE: Should we downgrade PA security to a=20
> SHOULD,MAY or remove it?
>=20
> Hi Gary,
>=20
> Comments inline.=20
>=20
> > -----Original Message-----
> > From: gtomlinson@SonicWALL.com [mailto:gtomlinson@SonicWALL.com]
> > Sent: Monday, September 17, 2007 4:54 PM
> > To: nea@ietf.org; Paul_Sangster@symantec.com
> > Subject: RE: [Nea] RE: Should we downgrade PA security to a=20
> SHOULD,MAY=20
> > or remove it?
> >=20
> > Hello Paul,
> >=20
> > Parvez and I have been thinking about this quite a bit, especially=20
> > since discussing the requirements of PT mutual authentication with=20
> > Kaushik, which has lead us to quite a bit of trust model analysis.
> >=20
> > Our PA-5 response is inline.  We will be commenting=20
> separately on the=20
> > PB/PT trust relationship which is germane to the PT mutual=20
> > authentication discussion thread.
> >=20
> > > Date: Sat, 15 Sep 2007 13:29:20 -0700
> > > From: "Paul Sangster" <Paul_Sangster@symantec.com>
> > > Subject: [Nea] RE: Should we downgrade PA security to a=20
> > > SHOULD,	MAY or
> > > 	remove it?
> > > To: <nea@ietf.org>
> > > Message-ID:
> > > =09
> > > <AB96CED633A7C246BDC661DBEE1CF01FD48F76@TUS1XCHCLUPIN11.enterp
> > > rise.veritas.com>
> > > =09
> > > Content-Type: text/plain; charset=3D"us-ascii"
> > >=20
> > > Unfortunately, this thread has failed to have traction. =20
> > > Since we currently have a MUST for PA security would
> > someone like to
> > > assert that this MUST should be lowered and provide some
> > rationale to
> > > start this discussion?
> > >=20
> >=20
> >=20
> > We'll take a stab at asserting PA-5 should be a [MUST]. =20
> Here is our=20
> > argument.
> >=20
> > Unfortunately, we've come to the group late in the=20
> requirements game=20
> > and thus didn't participate in the Prague discussions and as such=20
> > don't have the context that lead to the downgrading of PB=20
> > authentication and shifting all of this responsibility to PT and PA=20
> > respectively. However, we think the most important question is how=20
> > does the trust model work.
> >=20
> > If a PV does not utilize PA authentication then it implicitly is=20
> > relying upon the PB protocol and/or the PT protocol for=20
> authentication=20
> > of its associated PC. Such transitive trust is possible in=20
> use cases=20
> > where a NEA Client can provide assurances on the=20
> authenticity of the=20
> > PCs it hosts, such as in systems with immutable software images.=20
> > However, for use cases where such assurances can't be made, such as=20
> > when PCs can be installed on systems for which the PB Client can't=20
> > ascertain authenticity in behalf of the PV, the transitive trust=20
> > relationship breaks down. Given such use cases exist,
>=20
> [Joe] OK, but its not clear to that if the NEA client cannot=20
> provide assurances of authenticity for its host that the PC=20
> will be a better place to root trust than the PB or PT.  It=20
> would seem that in many architectures the PT and/or PB will=20
> run with higher privilege than the PC. =20
>=20
> > we don't see how PA-5 authentication can be dropped. In=20
> fact, thinking=20
> > about this more, it occurs to us that the minimum trust=20
> model required=20
> > for NEA is PA-5 with mutual authentication and trust between the PB=20
> > Server and its associated PVs via a (out of WG scope) secure=20
> > interface.  Our rationale for this assertion is, if each PV/PC=20
> > combination has a cryptographically sound communication=20
> channel, then=20
> > it isn't possible for a PB Client, or PT stack to spoof=20
> said messages=20
> > and if PB Server has a secure trusted relationship with=20
> each PV, then=20
> > PB Server can rely entirely on assertions from the PV's in=20
> composing=20
> > the complete end point posture state.
> > We're not advocating this minimalist model as the only trust=20
> > requirements, just stating we think it is the minimum set of trust=20
> > requirements.
> >=20
> > Now comes the semantic fun.  What is the meaning of [MUST],=20
> [SHOULD]=20
> > and [MAY] in the context of PA-5?  Our assertion is
> > PA-5 is a protocol requirement and as such, given our=20
> assertion there=20
> > is a need for PA-5 mutual authentication, then the PA=20
> protocol [MUST]=20
> > be capable of providing mutual authentication.  Since, the need for=20
> > mutual authentication is use case dependent, we assert=20
> implementation=20
> > of it is an optional requirement and therefore think a [MAY]=20
> > implementation requirement is appropriate to consider in the=20
> > forthcoming PA protocol RFC work.
> >=20
> [Joe] In looking at the complexity of managing keys and=20
> cryptography at the PA I think most deployments will opt for=20
> managing trust at the PB or PT layer.  I would rather see the=20
> working group focus on things that will be deployed as a=20
> higher priority requirement. =20
>=20
>=20
> [PA] Gary and I agree with you that most deployments will opt=20
> for managing trust at the PT layer. We still think that the=20
> PA protocol must be capable of providing cryptographic=20
> security for at least the following two reasons. First, a=20
> vendor may provide a combined solution of a Posture Validator=20
> and a Posture Collector; they may want the collected posture=20
> data to be opaque, to protect their intellectual property.=20
> Second, without the possibility of using cryptography at the=20
> PA layer, the "vertical lines" (the API lines, as opposed to=20
> the horizontal protocol lines) need to provide strong=20
> security guarantees.
> On some operating systems, this may be reasonable, but on=20
> others, these guarantees might be non-existent or very=20
> difficult to implement. For these reasons, we think that the=20
> PA layer should support the PA-5 [MUST] for those who have no=20
> option but to rely on it.
>=20
>=20
>=20
>=20
> _______________________________________________
> Nea mailing list
> Nea@ietf.org
> https://www1.ietf.org/mailman/listinfo/nea
>=20


_______________________________________________
Nea mailing list
Nea@ietf.org
https://www1.ietf.org/mailman/listinfo/nea



