IETF
mptcp@jabber.ietf.org
Wednesday, 27 July 2011< ^ >
ayourtch has set the subject to: Materials at https://datatracker.ietf.org/meeting/75/materials.html
Room Configuration

GMT+0
[08:21:10] yoshifumi.nishida joins the room
[08:21:33] yoshifumi.nishida leaves the room
[08:21:42] yoshifumi.nishida joins the room
[08:21:47] yoshifumi.nishida leaves the room
[12:24:19] costinraiciu joins the room
[12:24:31] costinraiciu leaves the room
[12:38:57] rvdp joins the room
[12:51:34] iljitsch joins the room
[12:52:54] <iljitsch> moguh
[12:53:11] <iljitsch> is the audio up?
[12:53:38] <rvdp> I don't hear anything yet on the various channels
[12:54:30] yoshifumi.nishida joins the room
[12:55:46] Christoph Paasch joins the room
[12:57:06] costinraiciu joins the room
[12:58:08] <iljitsch> I get some really soft stuff in my left ear
[12:58:18] <iljitsch> annoying that this is in stereo
[12:58:48] <Christoph Paasch> yes, have the same problem...
[12:58:51] <costinraiciu> did you try to flip your headset around?
[12:58:55] <iljitsch> although I guess it makes sense for MPTCP
[12:58:56] <costinraiciu> :)
[12:59:15] <iljitsch> I set my computer to play mono
[13:01:30] <iljitsch> so anything happening?
[13:02:09] jlcJohn joins the room
[13:02:34] <iljitsch> or is everyone still having some café?
[13:02:56] <rvdp> I can hear background noises, sounds like people are entering and setting up
[13:03:54] <iljitsch> oh wait
[13:04:16] <iljitsch> the ical feed says everything starts at 15.00 CET
[13:04:21] <iljitsch> but the agenda says 15.10
[13:04:56] <rvdp> I can hear the session (I'm using itunes on mac)
[13:05:19] <iljitsch> phil, you could stand to be a bit louder
[13:05:26] shep joins the room
[13:05:40] gorryf joins the room
[13:06:23] <gorryf> Hi, I am in the meeting room and can pass questions/comments to the meeting, please say if you want me to do this (gorry fairhurst)
[13:06:40] <gorryf> We are bashing the Agenda.
[13:06:53] JavierUbillos joins the room
[13:06:56] <iljitsch> yes the audio is low but otherwise good
[13:07:17] <gorryf> I'll ask Phil to speak more loudly.
[13:07:18] <iljitsch> (and still only left ear)
[13:08:17] <rvdp> works on both ears for me with volume set mid-range, so perfect
[13:08:58] <iljitsch> yes thanks
[13:09:12] <JavierUbillos> iljitsch: FYI: Audio has been "mono" during previous meetings too.
[13:09:24] <iljitsch> you mean left channel only?
[13:10:12] <JavierUbillos> No, "mono", and if you got stereo locally, I'm guessing it's "implementation specific" whether or not you get sound on both or on a single speaker.
[13:10:28] <gorryf> API draft now being presented
[13:11:10] <gorryf> Slide: Next steps
[13:11:27] <iljitsch> inaudible now
[13:11:53] <gorryf> Talk finished - is it ready for LC?
[13:11:57] <gorryf> Mark to present now
[13:12:49] <gorryf> Multipath TCP CC
[13:13:16] <costinraiciu> gorry - volume is really low for us listening in
[13:13:20] <gorryf> Slide: Status
[13:13:21] <costinraiciu> get them closer to the mike
[13:13:23] <iljitsch> can't really hear anything
[13:13:27] <costinraiciu> yes
[13:13:50] <gorryf> I'm doing my best
[13:14:06] <gorryf> Slide: Multipath TCP Spec
[13:14:31] <gorryf> looking for whether we are ready for a WGLC
[13:15:09] <gorryf> Slide: Refresh
[13:15:58] <rvdp> get sound on both ears, but one is louder, now volume set above mid-level, still OK, remote listening from Amsterdam
[13:16:50] <JavierUbillos> Tim: "Does it get there reliably?"
[13:17:06] <gorryf> Mic: Tim
[13:17:11] <JavierUbillos> Tim at the mic
[13:17:32] <rvdp> this is much louder
[13:17:44] <costinraiciu> this is actually normal level
[13:17:52] <costinraiciu> i couldn't hear mark
[13:17:55] <JavierUbillos> Is the presenter mic even on?
[13:18:02] <costinraiciu> they need to play with the mike volume
[13:18:49] <rvdp> I could hear Mark, but it was/is more faint
[13:19:04] <costinraiciu> is there anybody playing the buttons?
[13:19:15] <JavierUbillos> There is now.
[13:19:38] <JavierUbillos> Is it better?
[13:19:38] <Christoph Paasch> Currently the spec says, that the third ack needs to be acknowledged with a fourth ack from the server. (page 14) - before that mptcp is in the state "PRE_ESTABLISHED"
[13:19:51] <JavierUbillos> Is the presenter volume better now?
[13:20:11] <costinraiciu> a bit better
[13:20:30] <JavierUbillos> Tim@mic
[13:20:49] <costinraiciu> we needed the data ack anyways to be checked in the first window
[13:20:52] <iljitsch> this mike is too loud compared to the others
[13:21:22] <costinraiciu> if that isn't received then the key has not been received
[13:21:34] Andrew McGregor joins the room
[13:21:48] <gorryf> Audio: We can't do more (sorry) we don't know which channel feeds what.
[13:21:55] <costinraiciu> gorry : can you relay the above messages?
[13:22:01] <costinraiciu> we needed the data ack anyways to be checked in the first window
[13:22:08] damien.saucez joins the room
[13:22:12] <costinraiciu> if that isn't received then the key has not been received
[13:22:30] <gorryf> Tim: Says more work is needed.
[13:22:46] <gorryf> Jana: ?
[13:23:02] <costinraiciu> gorry please realy this: this is already in the protocol; I don't think it s an issue
[13:23:49] <costinraiciu> data ack can be used for this
[13:24:00] <costinraiciu> costin
[13:24:13] <gorryf> Gorry -> Costin, OK Mark will check...
[13:24:34] <gorryf> Slide: Break before make
[13:25:08] <gorryf> SLide: MP_PRIO
[13:26:15] <gorryf> Slide: Closefown
[13:27:34] <gorryf> Slide: State diagram
[13:28:37] <gorryf> Slide: Open Issues
[13:31:24] <gorryf> Slide: Mobility Fallback
[13:31:36] <gorryf> Mark is asking for comments?
[13:31:56] iljitsch leaves the room
[13:31:59] <gorryf> Tim: remind us why we are in fallback mode?
[13:32:15] iljitsch joins the room
[13:33:17] <gorryf> Tim: The middlebox has rewritten the content?
[13:33:45] <gorryf> Andrew at Mic:
[13:34:06] <gorryf> Andrew: Is a reset OK?
[13:35:00] <gorryf> George: The draft says no more multipath signalling in fallback.
[13:35:15] <gorryf> George: You can not get into this situation.
[13:35:24] <costinraiciu> you can have signalling in the syns
[13:35:35] <costinraiciu> gorry: please pass this through
[13:36:24] <gorryf> Mark's reply - that's why you can do the join, but you can not then sort out what happens afterwards
[13:36:32] <costinraiciu> i heard that - thanks
[13:36:44] <gorryf> georg I'll present a solution later...
[13:37:16] <gorryf> Slide: Teardown
[13:38:07] <gorryf> Philip: Giving up is a heuristic then? Mark: Yes, not interop issue.
[13:38:22] <gorryf> Andrew: What if no interfaces are up?
[13:39:03] <gorryf> Are we done with this document?
[13:39:33] <gorryf> Presentation next: Is it deployable?
[13:42:37] <gorryf> Slide: Middleboxes and new TCP Options in Syn Table
[13:43:58] <gorryf> Slide: What actually...
[13:45:57] <gorryf> Absolute sequence numbers can't be used in a new protocol
[13:46:28] <yoshifumi.nishida> I can't hear audio..
[13:46:36] <yoshifumi.nishida> oh. coming back
[13:48:45] <iljitsch> MY EARS!!!
[13:48:54] <gorryf> Full results are on the URL, data could be put on line, but this would need to be anonymised
[13:49:37] <gorryf> Georg at Mic:
[13:49:55] <gorryf> Where are the middleboxes that do harm?
[13:51:11] <gorryf> End of talk
[13:52:07] <costinraiciu> are georg's slides online?
[13:52:39] <gorryf> Should be on-line now
[13:53:00] <gorryf> Georg:
[13:53:07] <gorryf> Slide: Topics
[13:53:13] <JavierUbillos> http://www.ietf.org/proceedings/81/slides/mptcp-4.ppt
[13:53:45] <gorryf> Slide: Goals and Constraints
[13:54:43] <gorryf> Slide: MPTCP+Wireless
[13:55:48] <gorryf> Slide: MPTCP Applied to Wireless
[13:56:37] <gorryf> Mark at Mic:
[13:57:33] <gorryf> Slide: Issues (skipped slide)
[13:59:02] <gorryf> Mark at Mic:
[13:59:24] <gorryf> Slide: How to pick the best path?
[14:00:19] <gorryf> Slide: Value and Costs
[14:01:57] <gorryf> Slide: low complexity realization
[14:02:07] <gorryf> Slide: Prinicipal Concept
[14:03:24] <gorryf> Slide: Protocol Stack
[14:04:17] <gorryf> Slide: Sig Plane
[14:04:59] <gorryf> Slide: Sender
[14:06:28] <gorryf> Slide: Recv
[14:08:04] <gorryf> Slide: Interop
[14:09:02] <gorryf> Slide: Impl
[14:10:19] <gorryf> Slide: Sig Issues
[14:10:33] shep leaves the room
[14:11:02] <gorryf> Slide: Interface Availability Signalling (title)
[14:11:05] <gorryf> Slide: Interface Availability Signalling
[14:12:00] <gorryf> Slide: Path-Sel Conflict Resolution (title)
[14:12:29] <gorryf> Slide: Path-Sel Conflict Resolution - policy
[14:15:10] <gorryf> Slide: DSS issues (title)
[14:15:20] <gorryf> Slide: +1
[14:16:06] nestor.tiglao joins the room
[14:16:39] <gorryf> Slide: Temp fallback
[14:18:13] <gorryf> Slide: Problem with present fallback mode
[14:18:26] <gorryf> Slide: NO CHECKSUMS
[14:18:32] shep joins the room
[14:19:51] <gorryf> Slide: WITH CHECKSUMS
[14:21:52] <gorryf> Slide: Transparent Proxy (title)
[14:21:55] <gorryf> Slide: Transparent Proxy
[14:23:24] iljitsch leaves the room
[14:23:57] <gorryf> Slide: proposal
[14:26:34] <gorryf> Slide: Summary (Title)
[14:26:37] <gorryf> Slide: Summary
[14:28:31] <gorryf> Questions?
[14:28:38] <gorryf> Mark at Mic:
[14:29:30] <gorryf> One part is revolted by transparent proxies, but this is probably inevitable and we should play nicely with this.
[14:30:04] <gorryf> Mark: ... I think we should explicitly call this a "proxy" flag... does anyone have an opinion?
[14:30:26] <gorryf> Andrew: I agree there are always proxy devices "there"
[14:31:19] <gorryf> Andrew: ... there are good features here (is it important to tag this as "min" complexity, images are getting big for this part of the stack)
[14:31:34] <gorryf> Georg: How about recv buffer reduction?
[14:31:40] <gorryf> Andrew: good idea
[14:31:43] <costinraiciu> receive buffer reduction is orthogonal to the complexity of the implementation
[14:31:49] <gorryf> Mark: Not sure
[14:32:06] <costinraiciu> agree to what mark says
[14:33:32] <gorryf> gorry -> costin: relayed q.
[14:33:37] <costinraiciu> thanks gorry
[14:35:25] <gorryf> Andrew at Mic:
[14:35:53] <gorryf> Andrew: Mac training can be several 100 packets
[14:36:08] <gorryf> David Ward: Satellite PEP?
[14:37:06] <gorryf> Mark at Mic:
[14:37:21] <costinraiciu> that paper assumed there is an oracle that knows the best path
[14:37:32] <costinraiciu> that chooses the best path
[14:37:49] <gorryf> -> costin: - do we need to say this
[14:38:32] <costinraiciu> yes: 15% compared to an optimal path selection algorithm
[14:38:37] <costinraiciu> that can't be achieved in practice
[14:41:03] <gorryf> Question is what is the impact of choosing the wrong path and loosing a segment?
[14:41:14] <costinraiciu> you can also have redundant data if you want to probe and avoid hol blocking
[14:42:14] <gorryf> Philip: Interrupt
[14:42:45] <costinraiciu> thanks gorry - you must be sweating by now with all the realying
[14:42:48] <gorryf> ... It would be nice to have more experiments. I'd like to know final comments
[14:42:49] <costinraiciu> relaying
[14:43:06] <gorryf> Mark at Mic
[14:43:31] <gorryf> ... What needs to change the spec to be sent to IESG?
[14:43:54] <gorryf> <andrew suggests we need some hums>
[14:44:22] <gorryf> Andrew: This is a complexity increase for a small gain!
[14:44:35] <gorryf> ... do not do bulk optimisation,
[14:45:14] <gorryf> Philip: Transparent Proxy - should the spec be changed to allow this - or a side issue?
[14:45:26] <gorryf> Andrew at mic:
[14:46:20] <gorryf> Mark:
[14:47:01] <gorryf> I am in favour of adding a bit to support transparent proxies, 1) address of proxy 2) remote supports mptcp - could decide based on this
[14:47:38] <gorryf> Andrew:
[14:47:49] <gorryf> ... could be a firewall
[14:48:53] <gorryf> No middle boxes yet support MPTCP
[14:49:15] <gorryf> Mark: Andrews point is good (if we succeed)
[14:52:36] <gorryf> Mark:
[14:53:20] damien.saucez leaves the room
[14:53:49] damien.saucez joins the room
[14:54:17] <costinraiciu> accomodating proxies is a good idea
[14:55:33] <gorryf> Philip Decision was to add this to list of issues
[14:56:43] <gorryf> Tim at mic:
[14:57:43] <Christoph Paasch> mptcp.info.ucl.ac.be running v0.6
[14:58:55] <gorryf> Are there other implementations?
[15:00:03] <gorryf> Rechartering discussion now:
[15:02:35] gorryf leaves the room
[15:02:40] Christoph Paasch leaves the room
[15:02:48] costinraiciu leaves the room
[15:03:08] yoshifumi.nishida leaves the room
[15:03:08] damien.saucez leaves the room
[15:03:19] Andrew McGregor leaves the room
[15:03:25] shep leaves the room: Logged out
[15:04:11] JavierUbillos leaves the room
[15:08:22] rvdp leaves the room
[15:08:58] jlcJohn leaves the room
[15:20:15] JavierUbillos joins the room
[15:21:59] JavierUbillos leaves the room
[15:22:01] javvee joins the room
[15:23:38] javvee leaves the room
[15:23:42] javvee joins the room
[15:26:45] javvee leaves the room
[15:26:47] javvee joins the room
[15:28:52] nestor.tiglao leaves the room
[15:37:36] javvee leaves the room
[16:58:55] Andrew McGregor joins the room
[16:59:03] Andrew McGregor leaves the room
[17:08:11] nestor.tiglao joins the room
[17:15:10] nestor.tiglao leaves the room
[17:20:12] javvee joins the room
[17:34:07] javvee leaves the room
[17:34:08] javvee joins the room
[17:38:23] javvee leaves the room