P2P SIP working group March 14, 2008 9:00-11:30 Chairs: David Bryan, Brian Rosen Scribe: Ted Hardie Jabber Scribe: Dan York Note on the Agenda: The agenda as published contained slots for discussion of ideas from alternate protocols to be included in the consensus document. Because of the outcome of the initial consensus call, this discussion was deemed premature and the agenda changed to remove that discussion. Consensus Calls Taken: (Note that decisions taken in a meeting are subject to confirmation on the mailing list.) The chairs called for consensus on the adoption of draft-bryan-p2psip-reload-03. There was no consensus to adopt the draft at this time. The chairs called for consensus on the question "Should we ask that protocols developed allow HIP to be a customer of the P2Psip service, within the constraints of the charter?" There was no opposition. The chairs called for consensus on the question "Should we structure the P2PSIP service such that hip is a) a mandatory part of the technical infrastructure b) an optional part of the technical infrastrcuture c) potentially present only when it replaces IP, with no other linkage. Rough consensus for b as the current answer, with further discussion as the technical documents describing p2psip protocols progress. Action items assigned: The chairs requested that the authors of draft-bryan-p2psip-reload-03 develop a successor document or documents which clarifies the protocol syntax and semnatics. The chairs requested that Philip Mathews work with the Transport Area Area Directors for discussion of where the general aspects of his work ID-LOC should be discussed. The chairs requested Ted Hardie to submit a follow-on version of his draft using the draft-authorname-p2psip-topic convention. The chairs requested that discussion take place on the mailing list on the documents which did not get timeslots or for which there was no time for discussion at the meeting. Songs Sung: "Wild Colonial Boy" (Martin Dolly) Review of meeting discussion: Bruce Lowekamp presented http://www3.ietf.org/proceedings/08mar/slides/p2psip-0.ppt, The document represents a merger of ASP, reload-01, and p2pp. While the authors agree on many fundamental aspects of the peer protocol and have incorporated lessons learned from 4 implmentations, many issues remain open. The author team's intent was to produce a base peer protocol with support for different apps, with a very small number of drafts needed to specify a working system; they also intended to reuse base components where available. During discussion of the document, there were a number of speakers who raised issues with the readability of the draft and the resulting difficulty in analyzing whether it met the needs of the working group. The author team agreed that the three different sources of terminology had created some conflicts. Another issue raised focused on the layering: Usage layer Overlay routing and storage layer Forwarding layer Transport layer Concerns were expressed about the arrangement of functional components in both the usage layer and overlay routing/storage layer. The relationship to HIP was difficult to understand in the document as written (since it can be in multiple places), as was the use of a forwarding layer as a "link layer" on top of tls/dtls. Related concerns were raised with the bootstrapping mechanism as currently described. Another issue raised was the difficulty in understanding the encoding. The binary protocol evolved from TLV and the authors still consider it a work in progress; their intent, as expressed by Eric Rescorla, was to avoid a "bag of attributes" style encoding where you have to make sense of the bag, hence the aim to have a fixed order. Phil replied that routing area experience has been that TLV encodings are easier to extend and understand. Henny Sinnreich raised the issue of appointing an editor with time to spend on cleaning up the document. It was pointed out by the chairs that they could only appoint an editor for working group documents, and that as long as it remains an individual submission it is up to the author team to resolve editorial issues. There was, however, general agreement that the current draft could use editing, with support for this position from the author team. The author team expressed happiness that the work was done, but wished they had had more time to polish the result. Philip and Alan raised points about the destination lists and the possibility of routing loops. At the moment, the peer can inspect for routing loops, but there is nothing that prevents them. Philip was also concerned about the individual peers' keeping state. The group then discussed the adoption of the draft, resulting in the consensus call and results noted above. The working group chairs noted that they will attempt to set a deadline for this decision before the Dublin meeting, but that there are always challenges to deciding consensus with only mailing list input. Gonzallo Camarillo then presented: http://www3.ietf.org/proceedings/08mar/slides/p2psip-1.ppt The group discussed the various ways HIP could be used in relation to p2p SIP, concluding in the consensus calls and results noted above. Philip Matthews then presented: http://www3.ietf.org/proceedings/08mar/slides/p2psip-1.ppt Group discussion focused on how this differed from HIP and whether this extraction of certain elements from HIP was especially appropriate to the p2p context. The arguments against HIP's use of ORCHIDs appears to apply here, as do many other benefits or detriments from the use of HIP. The working group came to the conclusion that this work was better understood and pursued in a broader context. The time pressures on the working group meeting were fairly extreme at this point, and the following presentations were made over a fairly short period: Marcin Matuszewski presented http://www3.ietf.org/proceedings/08mar/slides/p2psip-4.ppt. The group discussed why specific devices might choose to be a client, and discussed the reasons they might choose to pay the costs of being a peer instead. Discussion of battery life related to keepalives vs. peer overhead noted that they have different potential optimizations (see draft-ietf-sip-outbound) Henning Schulzrinne presented http://www3.ietf.org/proceedings/08mar/slides/p2psip-5.ppt. Jiang XingFeng presented http://www3.ietf.org/proceedings/08mar/slides/p2psip-6.ppt. Ted Hardie presented: http://www3.ietf.org/proceedings/08mar/slides/p2psip-7.ppt No discussion time was available for the last three presentations, and further discussion will take place on the list.