IETF
mptcp@jabber.ietf.org
Wednesday, April 6, 2016< ^ >
ayourtch has set the subject to: Materials at https://datatracker.ietf.org/meeting/75/materials.html
Room Configuration
Room Occupants

GMT+0
[16:52:34] will ivancic joins the room
[16:54:17] Meetecho joins the room
[16:54:24] Joseph joins the room
[16:56:07] Joseph has set the subject to: Agenda at https://tools.ietf.org/wg/mptcp/agenda?item=agenda-95-mptcp.html
[16:56:24] Adam Hodge joins the room
[16:56:58] Joseph Ishac joins the room
[16:58:09] John Leslie joins the room
[16:58:19] Olivier Bonaventure joins the room
[16:58:25] jch joins the room
[16:58:34] Aastha Gupta joins the room
[17:00:33] Aastha Gupta leaves the room
[17:01:31] Douglas Stamper joins the room
[17:01:43] Matthew Sargent joins the room
[17:02:34] <Joseph Ishac> Audio and Video streams are good (Y)
[17:03:02] hiroyuki goto joins the room
[17:03:05] RISHABH JAIN joins the room
[17:04:38] Adam Hodge leaves the room
[17:05:04] Dave Taht joins the room
[17:05:26] costinraiciu joins the room
[17:06:43] will ivancic leaves the room
[17:06:46] will ivancic joins the room
[17:07:23] RISHABH JAIN leaves the room
[17:07:42] <Joseph Ishac> Hard to hear remotely too
[17:08:52] Victor Holanda joins the room
[17:11:04] RISHABH JAIN joins the room
[17:11:43] kamilla tenorio joins the room
[17:13:00] <Olivier Bonaventure> Heuristics do not affect the interoperability, they affect performance but not interop afaik
[17:13:46] kamilla tenorio leaves the room
[17:14:35] <Olivier Bonaventure> Datacenter : we would be happy to get data beyond the research papers, feel free to contact me offline
[17:16:45] Helen Chen joins the room
[17:17:20] Helen Chen leaves the room
[17:17:48] Olivier Bonaventure leaves the room
[17:18:29] Helen Chen joins the room
[17:19:53] hiroyuki goto leaves the room
[17:20:13] Olivier Bonaventure joins the room
[17:24:07] <costinraiciu> bumping up the version number is a huge decision
[17:24:18] <costinraiciu> it would be nice to require consensus
[17:24:25] <costinraiciu> i don’t think it’s necessarily needed
[17:25:32] safa a joins the room
[17:26:55] <costinraiciu> you can’t do it another way - servers won’t be able to correlate either
[17:27:24] <costinraiciu> is anyone relaying jabber to the mike?
[17:27:45] Matthieu Boutier joins the room
[17:28:14] Matthew Sargent leaves the room
[17:28:40] <Joseph> costinraiciu: I don't believe so.... :(
[17:29:04] <costinraiciu> what is the point of using jabber then?
[17:29:15] <Joseph> Perhaps the intent is to use the "virtual queue"?
[17:29:59] <Joseph> or more likely no one is monitoring the room
[17:30:17] <jch> What was the protocol extension that the last person at the mike mentioned?  "tcpenc"?  Pointer?
[17:30:39] <costinraiciu> tcpinc probably
[17:30:53] <jch> Ah, ty.
[17:30:53] <costinraiciu> (opportunistic encryption for TCP)
[17:34:34] Douglas Stamper leaves the room
[17:34:39] Olivier Bonaventure leaves the room
[17:35:44] <costinraiciu> i am not happy, to be honest - I think this major change to the handshake is not needed
[17:35:44] Olivier Bonaventure joins the room
[17:35:55] <costinraiciu> I am not convinced the benefits are worth it
[17:36:18] Douglas Stamper joins the room
[17:37:34] <Joseph> Seems like such a major change should be treated as a version difference or at least tested experimentally first.  I missed it, has the new handshake been studied in any tests?
[17:38:04] <costinraiciu> i am deeply uncomfortable with the version change
[17:38:13] <costinraiciu> i think it may torpedo any deployments we have now
[17:38:52] <costinraiciu> if you have e2e encryption on top it makes no sense to secure the “routing” of mptcp sub flows
[17:38:56] <jch> Joseph, IMHO this deserves an intervention at the mike.
[17:39:19] <Joseph> jch: unfortunately I am remote :(
[17:40:20] <costinraiciu> MPTCP version number
[17:40:29] <Joseph> These comments span multiple talks
[17:40:34] <costinraiciu> sure
[17:41:44] <costinraiciu> e2e encryption makes it irrelevant to secure the MPTCP sub flow setup
[17:41:54] <costinraiciu> so what’s the point in timing the two up
[17:42:05] <will ivancic> can soneone elaborate concerns on encryption?
[17:42:59] Victor Holanda leaves the room
[17:44:15] RISHABH JAIN leaves the room
[17:45:03] <Olivier Bonaventure> no plan at UCL for the moment on that, maybe experimental draft
[17:45:17] <Olivier Bonaventure> maybe experimental option will be implemented, but not handshake
[17:46:25] <Joseph> Thanks for relaying the questions of the jabber room
[17:47:24] <Dave Taht> hmm.
[17:48:47] <will ivancic> Will take handshake update to list.
[17:51:08] <costinraiciu> we have a load balancer solution that works without protocol changes
[17:51:18] <costinraiciu> (please relay)
[17:51:39] <Rui Paulo> where is it?
[17:51:51] <costinraiciu> it will be presented in berlin
[17:51:53] <jch> Joseph, I think you're absolutely right.  Adding something that's never been tested at this late stage doesn't feel ok.
[17:51:55] <costinraiciu> it is implemented and tested
[17:52:05] <Olivier Bonaventure> send a draft that describes your solution to the list
[17:52:10] <costinraiciu> will do
[17:54:20] Douglas Stamper leaves the room
[17:54:45] Olivier Bonaventure leaves the room
[17:56:04] <Joseph> jch: Precisely. Even if it's only conceived as a minor change on paper.  Testing would at least bound the benefits and allow the option to be weighed properly.
[17:56:12] Olivier Bonaventure joins the room
[18:08:33] <Joseph> (nit: can the slides be made full-screen?)
[18:08:42] <Rui Paulo> are the slides available somewhere?
[18:08:49] <Olivier Bonaventure> https://www.ietf.org/proceedings/95/slides/slides-95-mptcp-2.pdf
[18:08:56] <Rui Paulo> thanks
[18:09:03] <Joseph> Olivier Bonaventure: Thanks
[18:10:07] <Olivier Bonaventure> I always lose video and slide after 5-10, do you lose it as well ?
[18:10:20] <costinraiciu> how do you get the video?
[18:10:37] <Olivier Bonaventure> on meetecho
[18:10:44] <Joseph> They have been steady for me, on since the start of the mptcp session
[18:10:47] <costinraiciu> where’s the link?
[18:11:11] <Joseph> https://buenayreb.conf.meetecho.com/q-meetecho/login.jsp?ietf=mptcp
[18:11:16] <costinraiciu> thanks!
[18:11:42] <jch> iw10!
[18:11:44] <Meetecho> Olivier Bonaventure: here's a flash based alternative for slides+audio, http://conf.meetecho.com/video?s=buenayreb&r=mptcp&c=8895099
[18:11:55] <Meetecho> in case webrtc keeps on failing for you
[18:11:57] Costin Raiciu joins the room
[18:11:58] <Olivier Bonaventure> let me try
[18:12:32] <Joseph> jch, I believe that's default!
[18:12:52] <Joseph> There's a whole thread in tcpm about it.
[18:13:12] <jch> Yeah, but they're in the single digits kilobits.
[18:13:16] <Dave Taht> short MTUs still proved useful.
[18:13:25] Olivier Bonaventure leaves the room
[18:13:26] Rui A. Paulo joins the room
[18:14:40] <Dave Taht> jch: they are doing IW10 at 2.4kbits???? (I was not paying attention earlier)
[18:16:03] <jch> Dave, yes, but in the next slide they reduced it to iw1, and it worked much better, as you'd expect.
[18:16:10] <Joseph> It was an example of tuning, IW1 was used for the other tests
[18:16:26] <Joseph> what jch said :)
[18:16:28] <jch> Now they replaced CUBIC with Balia, and I'm desperately googling for Balia.
[18:17:01] <Rui Paulo> https://tools.ietf.org/html/draft-walid-mptcp-congestion-control-00
[18:17:05] <jch> ty
[18:17:13] <Rui Paulo> https://tools.ietf.org/html/draft-walid-mptcp-congestion-control-04
[18:17:27] <Dave Taht> cool
[18:19:06] Garima Pandey joins the room
[18:21:33] <Garima Pandey> hummm
[18:21:59] <Joseph> Garima Pandey: Audio is clean for me
[18:22:43] Payal Goyal joins the room
[18:22:48] <Garima Pandey> ??
[18:24:09] <Joseph> Sorry, I thought you were trying to signal audio feed issues
[18:24:43] Payal Goyal leaves the room
[18:25:00] <Garima Pandey> no...its ok
[18:27:42] Rahul Aggarwal joins the room
[18:28:41] Garima Pandey leaves the room
[18:29:10] Rahul Aggarwal leaves the room
[18:33:38] <jch> The most detailed description of what I mentioned was described in IFIP Networking 2015, full paper is at http://arxiv.org/pdf/1403.0445v4
[18:34:38] vojislav vucetic joins the room
[18:38:11] <Costin Raiciu> so what is the problem of tunneling IP over MPTCP? How is this solution better?
[18:38:44] costinraiciu leaves the room
[18:42:31] vojislav vucetic leaves the room
[18:44:16] <jch> Costin: context of your question?
[18:55:40] safa a leaves the room
[18:56:36] <jch> Too many words per unit of time, I'm lost :-/
[18:56:47] safa a joins the room
[18:57:41] <Joseph> Ditto
[18:59:41] <Joseph> are the UDP slides online?
[19:01:14] Helen Chen leaves the room
[19:01:23] John Leslie leaves the room
[19:01:31] will ivancic leaves the room: Computer went to sleep
[19:02:00] Dave Taht leaves the room
[19:02:00] Joseph Ishac leaves the room
[19:02:00] Matthieu Boutier leaves the room
[19:02:00] Rui A. Paulo leaves the room
[19:02:01] safa a leaves the room
[19:02:30] Joseph leaves the room
[19:02:46] Costin Raiciu leaves the room
[19:12:38] Meetecho leaves the room
[19:28:28] jch leaves the room
[19:31:10] will ivancic joins the room
[19:57:31] jch joins the room
[19:58:13] jch leaves the room
[20:09:09] Rui Paulo leaves the room
[20:20:16] will ivancic leaves the room: Computer went to sleep
[20:30:37] will ivancic joins the room
[20:39:34] will ivancic leaves the room: Computer went to sleep
[20:56:07] will ivancic joins the room
[21:28:54] will ivancic leaves the room