From: Alan Johnston <alan(_dot_)b(_dot_)johnston(_at_)gmail(_dot_)com>
Date: October 27, 2011 10:53:27 AM CDT
To: Ben Campbell <ben(_at_)nostrum(_dot_)com>
Cc: draft-ietf-cuss-sip-uui-reqs(_dot_)all(_at_)tools(_dot_)ietf(_dot_)org,
"gen-art(_at_)ietf(_dot_)org Review Team" <gen-art(_at_)ietf(_dot_)org>,
"Cuss(_at_)ietf(_dot_)org" <cuss(_at_)ietf(_dot_)org>
Subject: Re: Gen-ART Last Call review of draft-ietf-cuss-sip-uui-reqs-06
Ben,
Thanks for your review of the draft. See my comments below. I have also
revised the draft as draft-ietf-cuss-sip-uui-reqs-07 which I believe
addresses all the issues.
- Alan -
On Oct 12, 2011, at 4:27 PM, Ben Campbell wrote:
I am the assigned Gen-ART reviewer for this draft. For background on
Gen-ART, please see the FAQ at
<http://wiki.tools.ietf.org/area/gen/trac/wiki/GenArtfaq>.
Please resolve these comments along with any other Last Call comments you
may receive.
Document: draft-ietf-cuss-sip-uui-reqs-06
Reviewer: Ben Campbell
Review Date: 2011-10-12
IETF LC End Date: 2011-10-13
Summary: This draft is almost ready for publication as an informational RFC.
I have a few minor questions and comments that may be worth addressing first.
Major issues:
None
Minor issues:
-- section 1, 2nd paragraph, last sentence: "In particular, this mechanism
creates no requirements on intermediaries such as proxies."
What about SBCs, B2BUAs, etc?
There are no requirements on them either - we'll add text saying that.
-- REQ-4: "… any other form of redirection of the request."
"Any other form" seems a pretty strong statement. What about a b2bua doing
weird stuff?
Sure, no protocol mechanism can prevent a B2BUA from doing something. We
will clarify to just say redirection, since that operation is well defined in
RFC 3261.
-- REQ-8: "If the UAS does not understand the UUI mechanism, the request
will fail."
Based on the routing requirement, shouldn't that say that if the request
cannot be routed to a UAS that understands the UUI mechanism, the request
will fail?
Yes, this is clearer.
-- REQ-12:
What degree of certainty is required here? (i.e. strong identity?) If
implied by the SIP dialog, does that impact expectations on what sort of
authn must happen at the SIP layer?
This is not meant to imply strong identity. And since UUI data can appear in
a response, there aren't really any strong methods available with SIP. The
UUI mechanism does not introduce stronger authorization requirements for SIP,
but instead the mechanism needs to be able to utilize existing SIP approaches.
-- REQ 13:
I'm not sure I understand how this interacts with the ability for
intermediaries to remove UUI. Should this be detectable by the endpoints? Or
is that ability limited to the hop-by-hop case, or require no integrity
protection?
Yes, there are tradeoffs between this requirement and requirement REQ-9.
Hop-by-hop protection is one way to resolve this interaction.
Nits/editorial comments:
-- section 4, 2nd paragraph: "The UUI mechanisim should support both of
these approaches"
Should that be a numbered requirement? You've got requirements to support
e2e and hop-by-hop, but no requirement that mentions SIP layer vs
application layer.
Actually, this sentence is misplaced. There isn't really a requirement to
support both of these. I'll remove it to avoid confusion.