IETF
rmcat@jabber.ietf.org
Thursday, July 24, 2014< ^ >
martin.thomson has set the subject to: bufferbloat
Room Configuration
Room Occupants

GMT+0
[16:59:38] Geert Van der Auwera joins the room
[17:00:28] Lorenzo Miniero joins the room
[17:02:18] Paul E. Jones joins the room
[17:02:45] <Lorenzo Miniero> Slide 2: Note Well
[17:02:47] <Lorenzo Miniero> Slide 3: Administrativa
[17:02:58] Ted.H joins the room
[17:03:19] <Lorenzo Miniero> Slide 2: Note Well
[17:03:37] <Lorenzo Miniero> Slide 4: Agenda
[17:04:08] <Lorenzo Miniero> Slide 5: Agenda
[17:04:25] Lars joins the room
[17:04:45] <Lorenzo Miniero> Slide 6: WG drafts
[17:05:32] Raffaello Secchi joins the room
[17:07:47] frodek joins the room
[17:08:28] jesup joins the room
[17:10:14] <jesup> Could someone poke Colin to join jabber?
[17:10:34] <jesup> when he sites down
[17:11:45] <Lorenzo Miniero> just as a remainder, a Meetecho room is available to attend this session remotely: http://www.meetecho.com/ietf90/rmcat
[17:12:00] Brian Trammell joins the room
[17:12:18] <jesup> mic: I think delay variation (and perhaps more importantly, delay response to  cross traffic changes) will be seen in responses to the evaluation  criteria.  I don't think we could say much in the requirements - and  keeping delay variation low implies living with higher delay on average;  variation as low-as-possible would mean much higher delay (generally).   So it competes with the primary goal.  Also, delay jitter is normally  absorbed by the jitter buffers - the net result is that if we reduce  transfer jitter, the jitter buffer will contract (but generally not by  as much as we reduce transfer jitter), again leading to longer  end-to-end media delays. Getting the data to the jitter buffer with as  low a delay as possible generally lets the jitter buffer and adaptive  resampling do it's job; and the application can tune those to provide  the level of user-visible delay variation it wants.
So, I think we should keep with our primary goal, and not try to mix  jitter control into it, and let the jitter buffers and other application  tools decide how to process the data.
(I wouldn't say we don't care *at all*, but it's a much lower level of  concern, and I don't think any proposed algorithm would fail to provide  a reasonable delay profile over time - and so I wouldn't make it a  requirement, and spend time arguing over what's "reasonable" when I  don't think it will even come into play.  It's like "Fairness".)
[17:12:32] <Ted.H> Do we have a jabber relay at hand?
[17:12:37] <Lars> we do
[17:12:37] <Ted.H> I can read that if not.
[17:12:41] <Ted.H> Cool
[17:12:42] <jesup> Discussion on the list is ok
[17:12:47] <Lars> but i don't actually want to have the jitter discussion now
[17:12:53] <Lars> can we do this on the list?
[17:12:58] <jesup> lars: let's move it to the list
[17:13:02] <jesup> don't need to read
[17:13:02] <Lars> thanks
[17:13:32] <jesup> Agree, on list
[17:13:33] tuexen joins the room
[17:14:10] csp joins the room
[17:16:48] SCIMBE@gmail.com joins the room
[17:16:51] <Lorenzo Miniero> Slide 7: WG Related Drafts
[17:16:57] Suhas Nandakumar joins the room
[17:20:05] <Lorenzo Miniero> Slide 8: WG Status - Issues (1)
[17:21:03] Ted.H leaves the room
[17:22:58] <jesup> Posted to list on Jitter
[17:27:37] Ted.H joins the room
[17:28:10] <jesup> BTW mirja's mic is horribly distorted
[17:28:34] <csp> Mirja isn’t here…
[17:29:04] <csp> You mean Karen, or the floor mic Xiaoqing was using?
[17:29:07] kmirja joins the room
[17:29:18] <jesup> Karen - told you it was distorted
[17:29:36] <jesup> lars isn't great, may have similar problem  Floor is great
[17:29:51] <csp> think they’re sharing a single mic
[17:30:06] <Lorenzo Miniero> jesup I'm monitoring the audio in Meetecho and it seems ok, may be an issue with the mp3 stream?
[17:30:29] <Lorenzo Miniero> (if that's what you're listening to)
[17:30:36] <Lorenzo Miniero> Slide 9: WG Status - Issues (2)
[17:32:04] <jesup> Lorenzo Miniero: then why would I hear it only on Karen's audio?  (some others are a little poor like Lars.  Odd. )  64K mp3 stream I think
[17:32:30] <Lorenzo Miniero> I'll wait until she speaks again to see if it's distorted here as well
[17:32:35] <jesup> k
[17:32:35] <Lorenzo Miniero> didn't notice anything before
[17:33:12] <SCIMBE@gmail.com> +1 audio quality is poor
[17:33:14] Randell Jesup joins the room
[17:34:04] <Lorenzo Miniero> the mic in the mic line is bad here too, ack
[17:34:07] <jesup> Audio on meetecho is much better
[17:34:22] <jesup> yeah, I ehard that
[17:34:39] <jesup> lars sounds better though on meetecho
[17:37:44] Paul Jones joins the room
[17:38:19] <Lorenzo Miniero> Slide 10: Charter Update Proposal
[17:38:33] <Lorenzo Miniero> Slide 9: WG Status - Issues (2)
[17:39:15] <Paul E. Jones> Yeah, Meetecho is WAY better audio
[17:41:31] <Lorenzo Miniero> Slide 10: Charter Update Proposal
[17:41:52] <Lorenzo Miniero> Slide 11: Milestones
[17:42:35] <Lorenzo Miniero> the IANAPLAN session this morning in this same room was streamed via livestream; IIRC
[17:42:47] <Lorenzo Miniero> maybe some setup theret may have messed up some config for the mp3 stream
[17:42:55] <Lorenzo Miniero> Slide 12: Milestones Update Proposal
[17:43:16] <Lorenzo Miniero> Slide 13: Note
[17:50:08] csp leaves the room
[17:51:47] <Lorenzo Miniero> Presentation is over
[17:52:00] <Lorenzo Miniero> Slide 1: Using RTCP Feedback for Unicast Multimed
[17:52:10] <Lorenzo Miniero> Colin Perkins presenting
[17:52:55] <Lorenzo Miniero> Slide 2: Motivation
[17:52:57] <Lorenzo Miniero> Slide 3: Summary of Draft
[17:58:27] <Lorenzo Miniero> there's a DJ hidden somewhere :)
[17:58:37] <Paul Jones> There is...
[17:58:40] <kmirja> can u us a different mic?
[17:58:55] <kmirja> can u use a different mic?
[17:59:10] <kmirja> please!
[17:59:11] <Lorenzo Miniero> kmirja it's not clear where the issue is coming from
[17:59:20] <Lorenzo Miniero> it's in the room as well
[17:59:53] <Lorenzo Miniero> PS: there's a Meetecho room available if you want to get a video feed from the room as well
[18:00:13] <kmirja> where can i access it?
[18:00:16] <Brian Trammell> we appear to be bleeding audio from a different session
[18:00:31] <Brian Trammell> we've shut off the wireless mics
[18:00:52] <Lorenzo Miniero> http://www.meetecho.com/ietf90/rmcat
[18:01:02] <Lorenzo Miniero> audio improved for remotes as well?
[18:01:10] <jesup> yes
[18:01:31] <jesup> +1 for the draft
[18:01:34] Mirja Kuehlewind joins the room
[18:01:44] Martin Becke joins the room
[18:01:53] <kmirja> thanks better
[18:01:54] <jesup> adopt
[18:02:08] <Suhas Nandakumar> adopt
[18:02:24] kmirja leaves the room
[18:02:34] <Lorenzo Miniero> Presentation is over
[18:03:00] <Lorenzo Miniero> RMCAT evaluation test cases
[18:03:03] Mirja joins the room
[18:03:08] <Lorenzo Miniero> Slide 2: Editorial changes
[18:03:37] <Martin Becke> thx... much better
[18:04:12] <Lorenzo Miniero> Slide 3: Current test order
[18:04:59] <Lorenzo Miniero> Slide 4: Test case changes (1/2)
[18:06:02] <Lorenzo Miniero> Slide 5: Test case changes (2/2)
[18:07:20] Mirja leaves the room
[18:07:40] <Lorenzo Miniero> Slide 6: Misc. issues
[18:14:10] <Lorenzo Miniero> Slide 7: Potential test cases
[18:14:54] <Lorenzo Miniero> Slide 8: Next step
[18:16:25] <jesup> Adopt as WG document
[18:16:54] <Lorenzo Miniero> Slide 9: RMCAT Cellular test cases
[18:16:55] <Lorenzo Miniero> Slide 8: Next step
[18:17:03] <Lorenzo Miniero> Presentation is over
[18:17:27] <Ted.H> It's clearly the case the changes can occur when it is a working group draft, so if you would like to minute "no objections", I will withdraw mine.
[18:17:34] <Lorenzo Miniero> Mo Zanaty: Application Interaction
[18:17:40] <Lorenzo Miniero> Slide 2: Goals
[18:18:18] <Lorenzo Miniero> Slide 3: Changes since -00
[18:19:18] <Lorenzo Miniero> Slide 4: Conceptual Model
[18:20:21] <Lorenzo Miniero> Slide 5: Conceptual Model Additions?
[18:20:31] <jesup> Ask mo to keep the mic closer; audio goes away up and down
[18:20:55] <Lars> ack
[18:20:56] <jesup> I think when he looks one way we can hear him
[18:21:14] <Suhas Nandakumar> agree with Randell
[18:21:23] <jesup> thanks
[18:21:36] <Lorenzo Miniero> Slide 6: Implementation Model
[18:21:48] <Lorenzo Miniero> Slide 7: Interfaces and Interactions
[18:21:54] <Lorenzo Miniero> Slide 8: Config – Codec/RTP/RTCP/CC Interaction
[18:22:18] <Lorenzo Miniero> Slide 9: Codec – RTP/RTCP Interactions
[18:22:21] <Lorenzo Miniero> Slide 10: Codec – CC Interactions
[18:23:02] <Lorenzo Miniero> Slide 11: RTP/RTCP – CC Interactions
[18:23:24] <Lorenzo Miniero> Slide 12: CC – UDP Interactions
[18:23:45] <Lorenzo Miniero> Slide 13: CC – Shared State Interactions
[18:24:05] <Lorenzo Miniero> Slide 14: Next Steps
[18:27:48] <Lorenzo Miniero> Presentation is over
[18:28:12] <Lorenzo Miniero> Michael Welzl, Coupled congestion control
[18:28:21] <Lorenzo Miniero> Slide 2: Context,    
 background    
 
[18:33:04] <Lorenzo Miniero> Slide 3: Context,    
 background    
 /2    
 
[18:35:54] jesup afk for a bit
[18:37:08] <Lorenzo Miniero> Slide 4: Prior    
 draW    
 versions    
 
[18:38:19] <Lorenzo Miniero> Slide 5: What’s    
 new    
 in    
 -­‐03?    

[18:39:11] <Lorenzo Miniero> Slide 6: The    
 Conserva0ve    
 Ac0ve    
 FSE    

[18:40:22] <Lorenzo Miniero> Slide 7: Some    
 simula0on    
 results    
 
[18:41:15] <Lorenzo Miniero> Slide 8: Fairness    
 /    
 priori0za0on    
 (TFR
[18:41:34] <Lorenzo Miniero> Slide 9: 25
[18:43:01] <Lorenzo Miniero> Slide 10: How    
 to    
 evaluate    
 app-­‐lim
[18:43:26] <Lorenzo Miniero> Slide 11: 1    
 app-­‐limited    
 flow,    
 1    
[18:44:03] <Lorenzo Miniero> Slide 12: Using    
 priori0es    
 to    
 “protec
[18:44:40] <Lorenzo Miniero> Slide 13: 2    
 FSE    
 controlled    
 flows    
 
[18:44:50] <Lorenzo Miniero> Slide 14: Q
[18:45:10] <Lorenzo Miniero> Slide 11: 1    
 app-­‐limited    
 flow,    
 1    
[18:47:51] <Lorenzo Miniero> Presentation is over
[18:48:23] <Lorenzo Miniero> Varun Singh: Adaptive FEC for Congestion Control
[18:48:32] <Lorenzo Miniero> Slide 2: IPR
[18:48:44] <Lorenzo Miniero> Slide 3: End-to-End Delay
[18:49:26] Raffaello Secchi leaves the room
[18:49:52] <Lorenzo Miniero> Slide 4: CONCEPT
[18:51:03] Martin Becke leaves the room
[18:51:03] <Lorenzo Miniero> Slide 5: CC Framework
[18:51:20] <Lorenzo Miniero> Slide 6: RFC 6363: FEC Framework
[18:52:27] <Lorenzo Miniero> Slide 7: State Machine
[18:53:20] <Lorenzo Miniero> Slide 8: FEC Scheme
[19:02:02] <Lorenzo Miniero> Slide 9: RTCP Feedback
[19:02:29] Ted.H leaves the room
[19:02:34] <Lorenzo Miniero> Slide 10: Applicability
[19:03:03] tuexen leaves the room
[19:03:29] <Lorenzo Miniero> Slide 11: Next steps
[19:06:02] Brian Trammell leaves the room: Disconnected: session closed
[19:06:08] <Lorenzo Miniero> hi, this session's over
[19:06:14] Geert Van der Auwera leaves the room
[19:06:19] <Lorenzo Miniero> there's a follow up session in about a hour and a half
[19:06:27] <Lorenzo Miniero> we'll cover that as well, so see you in Meetecho later ;)
[19:06:28] Paul Jones leaves the room
[19:06:30] <Lorenzo Miniero> Presentation is over
[19:06:33] Lars leaves the room
[19:06:38] Mirja Kuehlewind leaves the room
[19:06:39] Paul E. Jones leaves the room
[19:07:37] Suhas Nandakumar leaves the room
[19:07:42] Randell Jesup leaves the room
[19:07:50] Lorenzo Miniero leaves the room
[19:18:42] tuexen joins the room
[19:20:18] Lars joins the room
[19:21:24] Ted.H joins the room
[19:22:16] Brian Trammell joins the room
[19:23:34] frodek leaves the room
[19:25:56] Lars leaves the room
[19:31:18] Ted.H leaves the room
[19:37:42] Brian Trammell leaves the room: Disconnected: session closed
[19:43:19] Brian Trammell joins the room
[19:56:06] tuexen leaves the room
[20:07:58] tuexen joins the room
[20:48:19] Brian Trammell leaves the room: Disconnected: Replaced by new connection
[20:48:19] Brian Trammell joins the room
[20:52:54] Brian Trammell leaves the room: Disconnected: session closed
[20:55:16] tuexen leaves the room
[21:05:41] Brian Trammell joins the room
[21:09:25] Brian Trammell leaves the room: Disconnected: session closed
[21:17:17] Brian Trammell joins the room
[21:21:54] Brian Trammell leaves the room: Disconnected: session closed
[21:25:25] Lorenzo Miniero joins the room
[21:26:54] <Lorenzo Miniero> woah, switching rooms on the fly :)
[21:27:05] <Lorenzo Miniero> ready for a new RMCAT session now!
[21:27:23] <Lorenzo Miniero> Available on Meetecho as before: http://www.meetecho.com/ietf90/rmcat_II
[21:30:15] John Leslie joins the room
[21:30:26] Geert Van der Auwera joins the room
[21:32:35] Paul Jones joins the room
[21:32:56] <Lorenzo Miniero> SCReAM: Self-Clocked Rate Adaptation for Multimedia
[21:33:16] Brian Trammell joins the room
[21:33:18] Jonathan Lennox joins the room
[21:33:48] <Lorenzo Miniero> Slide 2: Main features
[21:35:47] <Lorenzo Miniero> Slide 3: other features
[21:36:42] Randell Jesup joins the room
[21:38:37] <Lorenzo Miniero> Slide 4: Why self-clocking?
[21:39:33] Lars joins the room
[21:41:19] <Lorenzo Miniero> Slide 5: Feedback message
[21:43:13] Brian Trammell leaves the room: Disconnected: Replaced by new connection
[21:43:14] Brian Trammell joins the room
[21:43:40] <Lorenzo Miniero> Slide 6: Functional
[21:45:07] <Lorenzo Miniero> Slide 7: Sender queue handling
[21:45:38] Mirja Kuehlewind joins the room
[21:47:55] <Lorenzo Miniero> Slide 8: example
[21:49:07] <Lorenzo Miniero> Slide 9: Conclusion
[21:51:46] <Lorenzo Miniero> Slide 6: Functional
[21:54:45] <Lorenzo Miniero> Slide 7: Sender queue handling
[21:54:49] <Lorenzo Miniero> Presentation is over
[21:55:12] <Lorenzo Miniero> Evaluation of SCReAM and GCC
[21:56:48] <Lorenzo Miniero> Slide 2: undefined
[21:56:49] <Lorenzo Miniero> Slide 3: undefined
[21:56:54] <Lorenzo Miniero> whoops
[21:57:00] <Lorenzo Miniero> Slide 3: Intrduction
[21:57:14] <Lorenzo Miniero> Slide 4: Variable BW Single Flow
[21:57:47] <Lorenzo Miniero> Slide 5: Variable BW Multiple Flows
[21:57:52] <Lorenzo Miniero> Slide 6: Variable BW Multiple Flows
[21:57:58] Jonathan Lennox joins the room
[21:58:03] <Lorenzo Miniero> Slide 7: Competing RMCAT traffic
[21:58:10] Jonathan Lennox leaves the room
[21:58:57] <Randell Jesup> mic: has it been tried with lower limits than 100ms?  100ms of induced delay is problematic
[21:59:05] <Lorenzo Miniero> Xlicd 8: RTT Fairness
[21:59:48] <Lorenzo Miniero> Slide 9. Competing TCP Traffic
[22:00:26] <Lorenzo Miniero> Slide 8: RTT Fairness
[22:02:40] <Lorenzo Miniero> Slide 11: Cellular Test Cases
[22:02:50] <Lorenzo Miniero> Slde 12: Introduction
[22:03:09] <Lorenzo Miniero> Slide 13: Metrics
[22:03:17] <Lorenzo Miniero> Slide 14: General Verdict
[22:06:26] <Mirja Kuehlewind> ledbat is a scavenger protocol; it achieves higher throughput with Codel because traditional TCP cannot utilize the link with low thresholds
[22:09:07] <Lorenzo Miniero> Slide 15: Disclaimer
[22:09:15] <Lorenzo Miniero> Mirja was your comment for the mic?
[22:09:38] <Randell Jesup> what slide?
[22:09:39] <Mirja Kuehlewind> maybe that would be gooc
[22:09:43] <Mirja Kuehlewind> good
[22:09:56] <Lorenzo Miniero> Slide 16: Video Frame Latency
[22:10:04] <Lorenzo Miniero> ok
[22:10:09] <Mirja Kuehlewind> because the congestion control must co-exist with tcp traffic
[22:10:41] <Lorenzo Miniero> Presentation is over
[22:10:55] <Lorenzo Miniero> NADA Update
[22:11:00] <Brian Trammell> do we need a jabber scribe?
[22:11:01] <Lars> thank you for catching the comment, lorenzo
[22:11:08] <Lorenzo Miniero> no problem!
[22:11:19] <Lorenzo Miniero> Slide 2: Outline    
 
[22:11:25] <Lorenzo Miniero> Slide 3: Issues    
 with    
 Currently    
 Docume
[22:12:54] <Lorenzo Miniero> Slide 4:     
 Impact    
 of    
 Forward    
 Propa
[22:14:00] <Lorenzo Miniero> Slide 5: How    
 to    
 Fix    
 This?    
 
[22:15:48] <Lorenzo Miniero> Slide 6: Revised    
 NADA    
 Receiver    
 Behavi
[22:16:49] <Lorenzo Miniero> Slide 7: Revised    
 NADA    
 Sender    
 Behavior
[22:17:52] <Lorenzo Miniero> Slide 8: Diges6ng    
 the    
 Rate    
 Update    

[22:18:10] <Lorenzo Miniero> Slide 9: Diges6ng    
 the    
 Rate    
 Update    

[22:20:39] <Lorenzo Miniero> Slide 10: Diges6ng    
 the    
 Rate    
 Update    

[22:21:56] <Lorenzo Miniero> Slide 11: Diges6ng    
 the    
 Rate    
 Update    

[22:23:16] <Lorenzo Miniero> Slide 12: Simula6on    
 Evalua6ons    
 Based    
 o
[22:23:52] <Lorenzo Miniero> Slide 13: Variable    
 Available    
 Capacity    
 
[22:24:24] <Lorenzo Miniero> Slide 14: Variable    
 Available    
 Capacity    
 
[22:24:26] <Lorenzo Miniero> Slide 13: Variable    
 Available    
 Capacity    
 
[22:25:06] <Lorenzo Miniero> Slide 14: Variable    
 Available    
 Capacity    
 
[22:25:50] <Lorenzo Miniero> Slide 13: Variable    
 Available    
 Capacity    
 
[22:26:34] <Lorenzo Miniero> Slide 14: Variable    
 Available    
 Capacity    
 
[22:26:39] <Randell Jesup> mic: hundreds of seconds to saturate a couple Mbps pipe seems very slow... and the recovery from the spike doesn't seem like it's fast; hard to see at this scale
[22:26:49] <Lorenzo Miniero> Slide 15: Congested    
 Feedback    
 w/    
 Bi-­�
[22:28:27] <Lorenzo Miniero> Slide 16: Congested    
 Feedback    
 w/    
 TCP    

[22:29:29] <Lorenzo Miniero> Slide 17: Compe6ng    
 Flows    
 w/    
 Same    
 R
[22:30:31] <Lorenzo Miniero> Slide 18: RTT    
 Fairness    
 
[22:31:17] <Lorenzo Miniero> Slide 19: RMCAT    
 Flow    
 compe6ng    
 w/    
 a
[22:32:10] Jonathan Lennox leaves the room
[22:32:17] <Lorenzo Miniero> Slide 20: RMCAT    
 Flow    
 compe6ng    
 w/    
 a
[22:32:44] <Lorenzo Miniero> Slide 23: RMCAT    
 Lab:    
 Topology    
 
[22:33:24] <Lorenzo Miniero> Slide 24: Delay    
 Measurement    
 Framework    
 
[22:33:28] <Lorenzo Miniero> Slide 25: Results    
 from    
 Simplified    
 NAD
[22:33:28] <Lorenzo Miniero> Slide 26: Summary    
 of    
 Revised    
 NADA    
 
[22:33:34] <Lorenzo Miniero> Slide 24: Delay    
 Measurement    
 Framework    
 
[22:39:51] <Lorenzo Miniero> done here!
[22:39:51] Lars leaves the room
[22:39:51] Lars joins the room
[22:39:51] Lars leaves the room
[22:39:54] Geert Van der Auwera leaves the room
[22:39:59] <Lorenzo Miniero> I'l lhave to find something to eat or I'll faint on the floor...
[22:40:01] Paul Jones leaves the room
[22:40:03] <Mirja Kuehlewind> thanks
[22:40:05] <Lorenzo Miniero> recordings will be made avalable soon
[22:40:10] <Lorenzo Miniero> on http://ietf90.conf.meetecho.com
[22:40:20] John Leslie leaves the room
[22:40:21] <Lorenzo Miniero> you're welcome Mirja!
[22:41:05] <Lorenzo Miniero> Presentation is over
[22:41:24] <Mirja Kuehlewind> have a nice evening!
[22:41:26] <Mirja Kuehlewind> bye!
[22:41:49] <Lorenzo Miniero> good night! :)
[22:41:50] Mirja Kuehlewind leaves the room
[22:42:06] Randell Jesup leaves the room
[22:42:11] Lorenzo Miniero leaves the room
[22:42:57] Brian Trammell leaves the room: Disconnected: session closed
[22:46:49] Brian Trammell joins the room
[22:47:00] Brian Trammell leaves the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!