dccp@conference.ietf.jabber.com - 2002/11/21
[07:05] %% mrose has arrived.
[07:19] %% ietfwatch has arrived.
[07:21] %% falkad has arrived.
[07:24] %% falkad has left.
[07:25] %% Joseph has arrived.
[07:26] <Joseph> Ah, so now it's working
[07:26] * Joseph tries to catch up
[07:27] <Joseph> Slides for today's meeting at
http://www.isi.edu/~falk/dccp
[07:27] * Joseph has changed the subject to: agenda bashing
Aaron: agenda
[07:27] %% mallman has arrived.
[07:27] * Joseph has changed the subject to: agenda bashing
[07:28] %% da4089 has arrived.
[07:28] %% starsu has arrived.
[07:28] <Joseph> Current accoplishments
WG last called problem statement
Useful revisions
[07:28] <Joseph> Concerns:
1) Traffic is low (Low energy) - Is anyone paying attention to the work?
2) Lack of implementations
[07:31] %% gamze has arrived.
[07:38] <starsu> dccp is big fat guy?
[07:40] <starsu> The last file of the following URL is not downloaded
[07:40] <Joseph> Ugh, trying to catch up... typing in a buffer
[07:40] <Joseph> will transfer asap
[07:40] <starsu> http://www.isi.edu/~falk/dccp
[07:41] <Joseph> Mark Handley: 3 people have had implementations, although they are obsolete given the updates to the draft; knows of no new draft implementations.
[07:41] * Joseph has changed the subject to: DCCP Spec Update (Mark Handley)
[07:42] <Joseph> http://www.isi.edu/~falk/dccp/dccp-spec-handley.pdf
[07:49] <starsu> dccp-user-guide-lanphear.pdf is not downloaded
[07:50] %% starsu has left.
[07:50] %% Joseph has left.
[07:52] %% starsu has arrived.
[07:53] %% mallman has left.
[07:59] %% mallman has arrived.
[08:02] %% Joseph has arrived.
[08:03] %% Joseph has left.
[08:03] %% Joseph has arrived.
[08:03] %% Joseph has left.
[08:03] %% Joseph has arrived.
[08:03] <Joseph> Alison: Ask for feedback on the mailing list for changes... currently changes are not really put out for discussion.
Alison: It's the use case for DCCP, it is piece of the charter to deal with this problem (RTP as a major user of DCCP)
MIC: Assume DCCP is successfull, what is the probablity of header compression being incorporated.
Mark: Shouldn't be any more difficult than normal
Mark: Deployment issues, but possible and we should try
Steve: Real question for RTP/DCCP is motivation, service providers are conservative, why should they change, most would prefer not to care about congestion control (or are forced comercially). What is the added benefit.
Mark: UDP could be filtered, have a choice between the 2
Sally Floyd: Possible that with DCCP you could use ECN. So when routers are congeestion they could continue to use traffic that is ECN enabled. Draft of such a comment is in the works.
MIC: Streaming applications, unfair but aren't the "big bad guys"
Carl Perkins: Codecs to adapt with congestion control algorithms (possible limitation?)
Mark: Yes, good to consider.
Reinar: Is this being pushed forward as experimental
Aaron: No the groups goal is proposed standard
Andrie: Good to have some kind of API to work with
Aaron: Not goal to develop an api, but to discuss possible issues with the api
Alison: Good to have a example API...
Aaron: agrees
Alison/Sally/Other: [paraphrased highly - perhaps wrong?] There are other protocols that are similar to DCCP
Alison: SCTP has a type of unreliable mode, with _TCP_ congestion control. We don't want to have a protocol that has a multiple of functions.
Sally: Senses the room: Want to use SCTP to solve the problem. Why possibly not to use it...
1) Can use different congestion control algorithms (like TFRC) like Alison mentioned
2) Want to min. packet header size (SCTP uses chunks - larger headers and unneeded features (like multiple streams)
3) ECN
MIC: Header compression? Is there another IETF approved cong. control protocol.
Sally: Yes TFRC has been adopted for proposed standard... not protocol, but technique
Reiner: Moving to fast, not enough time to work on it, with it, think about it.
Alison: We're not trying to last call this today, just trying to charter.
Phil Conrad: We should consider DCCP and debate SCTP in paralell/later
Alison: No engineering is ever off the table, but there is a fealling that we have an architecture [functionality?] that we should write
[08:06] * Joseph notes that the consensous seems to be that people need to think about the problem more
[08:10] <Joseph> Spenser: Sense of the hum is if to adopt it as a working group doc. When?
Aaron: December. Milestones are intentionally agressive.
Mark H: Whatever the milestones say, doesn't matter, we should work on it and if at the end it gets accepted, ok, if not, then ok.
[08:14] * Joseph has changed the subject to: Sally F on TFRC
[08:14] %% Joseph has left.
[08:16] %% Joseph has arrived.
[08:16] %% Joseph has left.
[08:17] %% Joseph has arrived.
[08:17] <Joseph> (Reitteration of TFRC as a proposed standard an that its about when you can send a packet (ie: just an algorithm))
ccid3? and tfrc drafts, pretty much the same, modified to tack dccp onto them
[08:17] %% Joseph has left.
[08:17] %% Joseph has arrived.
[08:18] <Joseph> (troubles with jabber again)
[08:19] <starsu> what?
[08:24] %% starsu has left.
[08:25] %% starsu has arrived.
[08:27] <starsu> dccp is needed to jabber?
[08:27] <starsu> because of frequent disconnection?
[08:28] <Joseph> I'm not sure I follow... I was having trouble transcribing dccp wg session b/c of frequent disconnections :)
[08:34] * Joseph has changed the subject to: User Guide (API example) Damon Lanphear
[08:36] %% gamze has left.
[08:38] <Joseph> Real time media
Application Considerations
- Request for broader information needed on:
* IP Tel
* Unicast MultiM
* Games
* Other perspectives on streaming media
Open issues with API
* Kernel/User API??
* Loss signaling to application layer
should a loss be reported?
Request for ideas...
MIC: VoIP may need special considerations, for buffers etc?
Mark H.: Contribute
Sally: How to use loss information, sally's doing that and working on it.
Damon Lanphear on Implementations
- Hopefully have something ready by summer with updates on progress on the mailing list
Aaron: any other implementations? (no answer) Any other topics? (no answer)
[08:38] <Joseph> adjound
[08:39] <Joseph> heh, adjoun
[08:39] %% Joseph has left.
[08:45] %% da4089 has left.
[08:46] %% starsu has left.
[08:47] %% ietfwatch has left.
[08:54] %% mallman has left.
[09:33] %% mrose has left.
[11:20] %% Joseph has arrived.
[11:20] %% Joseph has left.