[06:50:31] --- weddy has joined
[06:55:04] --- Melinda has joined
[06:59:20] --- admcd has joined
[07:02:15] --- jishac has joined
[07:05:00] --- csp has joined
[07:05:08] --- dotis has joined
[07:06:14] --- cgn has joined
[07:08:00] --- mallman has joined
[07:09:00] --- lars has joined
[07:10:30] --- mattz has joined
[07:13:56] --- csp has left
[07:14:01] --- falk has joined
[07:14:04] --- Ted Faber has joined
[07:14:16] <falk> I nominate Allman to co-chair tsvwg.
[07:14:32] --- touch has joined
[07:14:36] --- ray_atarashi has joined
[07:14:49] --- csp has joined
[07:15:17] --- arifumi@jabber.org has joined
[07:15:22] --- rbless has joined
[07:15:29] --- becarpenter has joined
[07:16:21] <mattz> francois is topic one
[07:16:25] <mattz> (dont' have agenda up)
[07:16:35] <mattz> I'm note taker, so may ignore jabber
[07:16:45] <mattz> rsvp-dste-00
[07:18:17] --- gorryf has joined
[07:20:07] --- csp has left
[07:21:08] --- csp has joined
[07:23:24] <mattz> bob braden at mike
[07:33:48] <mallman> falk just got out from under dccp ..... i think he probably he has more free cycles to co-chair tsvwg than i.
[07:35:19] --- ray_atarashi has left
[07:37:17] --- touch has left
[07:40:08] --- touch has joined
[07:44:33] --- gorryf has left: Disconnected
[07:44:57] --- csp has left
[07:46:33] --- cgn has left: Logged out
[07:49:06] --- gorryf has joined
[07:52:43] --- martin has joined
[07:52:52] <falk> mallman has more of an investment in seeing the group run well
[07:53:37] <mallman> falk: how about as a truce we just nominate faber?
[07:55:12] <falk> hmm... I like that idea.
[07:55:50] <Ted Faber> Fortunately I'm unelectable.
[07:56:01] <mallman> see, it's all about compromise here in the ietf. i'll get the campaign signs and buttons printed.
[07:56:07] <Ted Faber> Nominate away, boys.
[07:56:45] <Ted Faber> After I throw that baby into the INT meeting, there should be no further problems.
[07:57:08] <falk> baby?
[07:57:15] <Ted Faber> (I don't actually throw any babies)
[07:57:31] --- Ted Faber has left
[07:57:40] <mallman> (i see we're all warming up for the plenary here)
[07:58:02] --- Ted Faber has joined
[08:01:12] --- martin has left
[08:02:03] <rbless> draft-babiarz-tsvwg-rtecn-04 is up
[08:03:08] <rbless> James Polk: How is sender notified to back off?
[08:03:32] <rbless> Kwok: Read draft..
[08:03:56] <rbless> David Black: Not sure that works as advertised...
[08:04:49] <rbless> A: ECn marking is sent to endpoint, ep sends to application control
[08:05:50] <rbless> Collin Perkins: same mechanisms proposed in mmusic WG.(?) There was a strong pushback, because no believe that this works
[08:06:31] <rbless> J. Polk: RTP packets may take different paths...
[08:07:28] <rbless> Babiarz: Application control will terminate session
[08:08:02] --- gorryf has left: Disconnected
[08:08:37] <mattz> who is at mic?
[08:08:56] <rbless> J. Polk: terminate session if you don't know which path was affected...??
[08:09:04] <rbless> M. Westlund?
[08:09:05] <falk> Mattz: Magnus Westerlund
[08:09:07] <rbless> at mike
[08:09:13] <rbless> :-)
[08:09:17] <mattz> thanks
[08:09:47] <falk> Mallman: I say we nominate faber for http://www1.ietf.org/mail-archive/web/ietf/current/msg36668.html
[08:10:24] * falk receives an icy stare from faber
[08:10:28] <Ted Faber> Can we pick on me locally and not here?
[08:11:05] --- Ted Faber has left
[08:12:08] <rbless> Colin Faber: you dont know that the path is consistent. Admission control makes only sense when you know that packets follow same path...
[08:12:23] <rbless> s/Faber/Perkins :-)
[08:13:13] <rbless> Going to be confused by off-topic discussion in here :-)
[08:13:35] <falk> I'll shut up now.
[08:14:10] <rbless> missed several comments...
[08:15:15] --- martin has joined
[08:15:24] <rbless> David Black: Non-RTP as path setup, but RTP as data flow will not work (e.g. path switch due to filters)...
[08:15:34] --- touch has left
[08:15:51] <rbless> Perkins: RTP is the wrong place to do that...
[08:16:01] <rbless> Alison: what about NOOP?
[08:16:43] <rbless> Perkins: broken from RTP point of view. Would be happy if you send RTP data...
[08:17:29] <rbless> Fred Baker: concerned that additional traffic could affect already admitted traffic
[08:17:51] <rbless> A: that's why we want a very lightweight RTP flow (maybe NOOP)
[08:19:21] <rbless> missed comment..
[08:20:28] <rbless> Allison: should go on with this...don't have time now. Please discuss in smaller circle.
[08:21:09] <rbless> Allison: [to David Black] could you start admission control discussion on mailing list?
[08:22:03] <rbless> Sally Floyd next
[08:22:20] --- touch has joined
[08:22:21] <rbless> draft-floyd-ecn-alternates-01
[08:22:40] <rbless> want to go for official WG doc
[08:22:50] --- fred has joined
[08:23:45] <rbless> how do routers know ECN semantics?
[08:24:23] <rbless> RFC 3168 gives the default ECN semantics for all packets, regardless of the DSCP
[08:24:45] <fred> Routers know about ECN at all by configuration. Routers generally don't know about RFC 3168 unless they have some numbers typed in to control RED and are configured to perform ECN. Same with any other semantics.
[08:24:55] --- becarpenter has left: Lost connection
[08:26:46] <rbless> @fred: That was on Sally's slides..
[08:28:24] --- csp has joined
[08:28:55] <rbless> Alternate ECN semantics: Is ECN nonce used? Co-existence with competing traffic? General merits of the alternate ECN semantics?
[08:29:28] <mallman> who was speaking at the mic?
[08:29:44] <lars> bob briscoe
[08:29:53] <mallman> i thought so; thanks
[08:30:10] --- fred has left
[08:30:24] --- fred has joined
[08:31:23] <rbless> Allison: hum for adopt this as WG doc
[08:31:33] <rbless> in favor
[08:31:37] --- becarpenter has joined
[08:32:01] <rbless> -> Informational (no hum against)
[08:32:33] <rbless> Aaron Falk: more a BCP?
[08:33:16] <mattz> afalk: IST?
[08:33:34] <rbless> Next: (Sally) Quick-Start for TCP and IP
[08:33:34] <mattz> (that's a question, not a statement by aaron)
[08:33:41] <falk> ISD (see newtrk)
[08:34:20] <mattz> thanks for the pointer
[08:34:52] <rbless> Quickstart is IP option in the SYN packet, describes sender's desired sending rate
[08:35:27] <rbless> feedback by receiver by SYN/ACK
[08:36:10] <rbless> heading for expiremental..
[08:37:04] <mallman> new ISN is a dumb idea, right?
[08:37:12] --- danwing has joined
[08:37:17] <mallman> (it's my idea, but it probably is dumb)
[08:38:14] <rbless> Are there any TCP implementation that would react badly if retransmitted SYN has different ISN?
[08:38:48] <jishac> answer to that would probably take some digging
[08:39:07] <mallman> we are thinking about a tbit test
[08:39:14] <mallman> which would be easy enough, i guess
[08:39:39] <mallman> i think you'd have to do that to keep this ISN hack in the i-d when it ptogresses
[08:39:56] <rbless> More fundamental question: Use four unused bits in the IP option for Quick start nonce? Goal: discourage receivers from lying about the value of the received rate request
[08:40:08] <mallman> but, if someone could just say "yeah, that's really dumb and here is why" and save me the effort, that'd be good enough.
[08:40:32] <falk> Mark: do you want any of this repreateed at the mic?
[08:40:43] <mallman> no
[08:41:13] <mallman> i can block it on this point myself. ;-)
[08:41:21] <falk> sounds like it
[08:41:32] <mallman> i am more interested in hearing feedback on this nonce.
[08:41:42] <mallman> it isnt great.
[08:42:17] <mallman> but, it's ok, i guess. leaving "reserved" bits in an option is also unappealing to me.
[08:42:39] <rbless> sorry missed some comments from Randal Stewart and M. Mathis
[08:43:20] <rbless> What was the point from Kevin Fall?
[08:43:27] <mallman> to kevin: but, the sender cant send that fast presumably
[08:44:45] --- touch has left
[08:44:57] <rbless> Mark Handley is up next presenting Internet Congestion Control Research Group
[08:45:22] <rbless> [nice drawing, BTW :-)]
[08:45:48] <rbless> proposed charter...:
[08:45:59] <rbless> Limitations of AIMD Congestion Control
[08:47:12] <rbless> => Will need to change the congestion control dynamics of the Internet...
[08:47:39] <rbless> opportunity to do it right (and start now)
[08:48:33] <rbless> Plenty of ideas...need forum for eval and consensus that includes researches and equipment vendors... (IETf not good place to do this)..
[08:49:01] <rbless> proposed charter: http://nrg.cs.ucl.ac.uk/mjh/iccrg
[08:49:34] <rbless> ML: http://oakham.cs.ucl.ac.uk/mailman/listinfo/iccrg
[08:49:45] <rbless> ML by tomorrow (not yet up)
[08:50:50] --- becarpenter has left: Replaced by new connection
[08:50:50] --- becarpenter has joined
[08:50:50] --- becarpenter has left
[08:51:25] --- fred has left: Disconnected
[08:51:56] --- fred has joined
[08:53:13] --- touch has joined
[08:54:29] <rbless> Who is presenting?
[08:55:24] --- oran@jabber.org has joined
[08:55:37] --- csp has left
[08:57:12] <rbless> Topic is high bandwidth delay products and CC. Proposal H-TCP.
[08:57:34] <mattz> presenter is probably doug leith
[08:57:49] <rbless> Main idea: adjust increase rate as function of time since last backoff
[08:58:05] <rbless> Thanks matt
[08:59:27] <rbless> Allison: read individual draft and comment on that
[09:00:06] <rbless> Hum for interest (positive hum)
[09:00:19] <rbless> Bob Briscoe up next
[09:00:57] <rbless> Controlled Load (CL) Service
[09:08:44] --- oran@jabber.org has left: Logged out
[09:11:42] --- fred75657 has joined
[09:15:15] --- touch has left
[09:24:24] <rbless> Fred Baker: Q: rate of ECN in RSVP message, also value with signaling loss?
[09:24:47] <rbless> A: if experience loss then usually due to congestion, so ECN sufficient
[09:26:33] <rbless> Next speaker Mark Watson: FEC Streaming Framework
[09:27:37] --- rbless has left
[09:28:19] --- danwing has left
[09:28:26] --- fred has left: Disconnected
[09:30:22] --- martin has left: Logged out
[09:35:55] --- dotis has left
[09:39:15] --- falk has left: Computer went to sleep
[09:40:13] --- lars has left: Logged out
[09:40:31] --- arifumi@jabber.org has left: Logged out
[09:40:36] --- fred75657 has left
[09:44:28] --- Melinda has left
[09:53:04] --- weddy has left
[09:53:58] --- mattz has left
[09:55:12] --- admcd has left: Disconnected
[09:55:22] --- admcd has joined
[09:56:26] --- admcd has left
[10:05:13] --- mallman has left
[10:11:54] --- jishac has left
[10:18:37] --- touch has joined
[10:18:41] --- touch has left