MINUTES - SIMPLE - IETF69 Wednesday July 25 1510-1610 Summary: * The group agreed to adopt -simple-simple as a working group item. The room felt that something closer to the hitchhiker's guide in SIP is what we want to produce rather than the earlier architecture draft attempts. * An error was identified and removed from the output numbers in the interdomain scaling analysis draft. The next revision will be ready for last call. * The room discussed pursuing nicknames in MSRP vs a non-chat specific mechanism. The conclusion of the discussion on the room was that we would continue to pursue the path currently in the draft. If a general mechanism appears before this draft is done, the group will consider shifting to use it. * All SIMPLE participants are encouraged to help finish the part of XCON needed to realize SIMPLE chatrooms, or help recruit someone who will * Brian Rosen will coordinate an effort to start work on a generalized nickname mechanism. Raw notes follow ---------------------------- Notes on SIMPLE at IETF 69 Recorded by Dean Willis Meeting chaired by Robert Sparks. Agenda accepted as presented. Status reviewed by chair. status issue: ; xcap-diff: According to JDR, a swamp. Caught up in requirements discussion and there are many open issues relating to OMA requirements, synchronization, granularity, and issues with xcap-config. Topic: SIMPLE Made Simple by Jonathan Rosenberg Slides presented. A hitch-hikers's guide to SIMPLE. Comment: This is a good start on the work item. But it needs a lot more use cases, and discourse on bringing the pieces together. It doesn't have enough context -- needs to be 35 pages or so that could be used to orient a new developer. Needs to have a network diagram showing where the components are and what protocols are used between them. Response: That would be the old architecture document that we failed to produce. JDR is not interested in attempting it. Avshalom Houri reports that he tried working on this several years ago, and it just kept getting bigger. We need a much clearer definition of the task. Poll on "who read this" revealed few had read it. Poll on "is this a good starting point" indicates a fair number of the people who read the draft thing it is a good starting point. Few people (15 people) responded, with about 2/3 thinking this draft is what we want, and about 1/3 think a lot more work is needed. Poll: Who will contribute text to a more detailed document? No takers . . . John Elwell suggested that though he thinks this is nearly done, he would like a picture or two describing how the parts fit together. JDR is willing to do a one pager. Several speakers noted we're not likely to get much better than this document. It would be nice, but it isn't likely to happen. Consensus: Adopt this document as WG, next rev to be submitted as simple-00. Topic: SIMPLE Problem Statement Led by Avshalom Houri Slides presented Changes since last version of document reviewed. A major calculation error (50%) has been corrected. This error was in the number of bytes only. Question: What is the source of the numbers? They appear to be rough assumptions. It would be very good to have real numbers here. Noted that while the numbers look large, they work out to a small number like 75 bytes per second per subscriber. Perhaps we simply have unreasonable assumptions. The traffic for 20M users will always be large. Response: The problem is in the fundamental assumptions of the architecture. For example, do we need currency of status for all presentities? Poll: Are we ready for WGLC? Consensus is yes. Chair will schedule. Topic: Chat using MSRP Led by Miguel Garcia Slides presented Major ongoing work is around nicknames. Major Issue: Nickname reservation mechanism A protocol is needed. SIP mechanisms including new methods and new headers have been proposed, as has the NICKNAME method in MSRP. Proposed by author that we us the MSRP method. Discussion: (Brian Rosen, Ben Campbell) Nicknames are useful in many scopes, and generality is desirable. However, there is a desire to expedite a solution. We could have TWO mechanisms. The general mechanism would appear to be needed in SIP. Nicknames inside MSRP are not a general solution -- won't work in mixed-media conferences, for example.. The parts of the network that know about identities (signaling) are also likely to be the parts that know about nicknames. Comments: (JDR) Draft starts on nicknames between domains. This is a big challenge and should be excised. If scope is restricted to similar to SSRC, then identifiers are scoped within the conference only and things are much easier to achieve. Comment: We really need to have SOMETHING that works well enough to use at IETF. This is very important. Comment: (Mary) There is clearly a more general problem that exceeds MSRP and SIP (it occurs also in XCON). It might be acceptable if tightly scoped to do it in MSRP, but we should probably have a second effort to do the general solution. Comment: (Paul K). Long-term reservation scopes this outside of MSRP. Restricting scope (to level of SSRC) makes the solution unsatisfactory. Comment: (Gonzalo) We should work toward a general solution, but do a minimal solution (ala MSRP) in the short term. AD Comment (Jon P): We know that the work here overlaps with other stuff in RAI, but we chartered a subset here in SIMPLE with the intent of getting something done sooner. Chair suggestion: WG to proceed down this path (MSRP) with people who think a general solution is needed to go work on it. If they get a general solution done in good order, we'll adopt it, else proceed with MSRP. No one stood up to say this is a show-stopper for them (though many people don't like it they aren't willing to stand up an alternative). Brian Rosen asked people to meet him today to work on a more general solution. Chair noted that XCON is working on abstract model that needs review and support from SIMPLE. Poll for volunteers shows general apathy . . . Further exhortation raised two or three hands. People who care are asked to come to XCON tomorrow. Meeting concluded.