IETF
PLUS
plus@jabber.ietf.org
Thursday, July 21, 2016< ^ >
Room Configuration
Room Occupants

GMT+0
[07:42:49] craigt joins the room
[07:49:30] Eliot Lear joins the room
[07:49:39] Barry Leiba joins the room
[07:51:45] Yoav Nir joins the room
[07:54:17] m&m joins the room
[07:55:09] Stephan Neuhaus joins the room
[07:55:27] jm joins the room
[07:55:37] hildjj joins the room
[07:55:56] yuki goto joins the room
[07:56:23] John Leslie joins the room
[07:57:30] Meetecho joins the room
[07:58:08] Lorenzo Miniero joins the room
[07:58:29] tale joins the room
[07:58:51] ted.h joins the room
[07:58:58] <tale> Etherpad for this meeting is at http://etherpad.tools.ietf.org:9000/p/notes-ietf-96-plus?useMonospaceFont=true
[07:58:59] craigt leaves the room: Disconnected: closed
[07:59:16] <Eliot Lear> Good morning.  This is the PLUS BoF and I will be your jabber scribe
[08:00:12] Szilveszter Nadas joins the room
[08:00:27] <Eliot Lear> materials for this bof can be found at https://datatracker.ietf.org/meeting/96/materials.html (search for "plus")
[08:00:32] sftcd joins the room
[08:00:43] gg bb joins the room
[08:01:03] <Eliot Lear> Introductory slides are showing - https://www.ietf.org/proceedings/96/slides/slides-96-quic-0.pdf
[08:02:23] Toru Asahina joins the room
[08:02:34] Lee Howard joins the room
[08:02:48] SzilveszterNadas joins the room
[08:03:14] gg bb_8786 joins the room
[08:03:22] Samuel Jero joins the room
[08:03:40] Martin Stiemerling joins the room
[08:03:42] frodek joins the room
[08:03:47] Catherine Truchan joins the room
[08:04:02] gg bb leaves the room
[08:04:09] hildjj leaves the room
[08:04:20] <Eliot Lear> natasha has opened the meeting
[08:04:21] theresa joins the room
[08:04:26] Yoshiro Yoneya joins the room
[08:04:30] Andrew Sullivan joins the room
[08:04:47] <Eliot Lear> aaron cochairs; remote participation is available
[08:04:51] Kyle Rose joins the room
[08:05:21] Simon Pietro Romano joins the room
[08:05:21] gg bb joins the room
[08:05:22] sureshk joins the room
[08:05:24] <Eliot Lear> slide 2; aaron falk now speaking
[08:05:37] gg bb_8786 leaves the room
[08:05:42] <Eliot Lear> Cast of characters now on the screen
[08:05:51] DanYork joins the room
[08:06:21] spreda joins the room
[08:06:21] Suresh Krishnan joins the room
[08:06:21] <Eliot Lear> a few different viewpoints will be presented
[08:06:30] Simone Ferlin joins the room
[08:06:33] <Eliot Lear> it's a wg forming bof.  slide 5
[08:06:46] <Eliot Lear> normally wg forming bof questions now up
[08:07:01] jsaldana joins the room
[08:07:16] <Eliot Lear> slide 6
[08:07:22] Emile Aben joins the room
[08:07:28] <Eliot Lear> not talking about implementation specifics or protocol specifics
[08:07:34] Carsten Bormann joins the room
[08:07:50] <Eliot Lear> this is a follow-on to the spud bof but we won't talk about that today.  we're aiming for a more abstract discussion
[08:07:53] stere preda joins the room
[08:07:53] dkg joins the room
[08:07:56] <Eliot Lear> slide 7
[08:07:58] <Barry Leiba> I like "ITOOTC".
[08:08:07] <Eliot Lear> consensus questions
[08:08:09] Ole Trøan joins the room
[08:08:22] nygren joins the room
[08:08:52] gg bb leaves the room
[08:08:52] Spencer Dawkins joins the room
[08:08:52] craigt joins the room
[08:09:04] <Eliot Lear> Now Joe Hildebrand- "Impediments to Transport Innovation"
[08:09:06] <tale> Ugh, wifi problems make etherpad very difficult
[08:09:06] <dkg> Eliot Lear: i'm pretty sure that's the quic slides
[08:09:19] <Eliot Lear> Oops
[08:09:30] Lee Howard leaves the room
[08:09:31] <Eliot Lear> https://www.ietf.org/proceedings/96/slides/slides-96-plus-4.pdf
[08:09:39] <dkg> thx :)
[08:09:46] nygren leaves the room: Disconnected: Replaced by new connection
[08:09:48] Brian Trammell joins the room
[08:09:50] Kevin Ma joins the room
[08:09:56] <Eliot Lear> slide 2
[08:10:01] <Eliot Lear> what firewalls do
[08:10:17] gg bb joins the room
[08:10:25] nygren joins the room
[08:10:32] <SzilveszterNadas> @Meetecho: The right of the slide is kind of missing. Not a big problem though.
[08:10:36] m&m leaves the room: Disconnected: Replaced by new connection
[08:10:43] m&m joins the room
[08:10:44] tale leaves the room: Disconnected: Replaced by new connection
[08:10:45] tale joins the room
[08:11:22] gg bb leaves the room
[08:11:37] Spencer Dawkins leaves the room
[08:11:43] <Meetecho> Szilveszter Nadas: possibly a glitch in the capture card that gets the beamer feed, it should be readable though
[08:11:56] aaron joins the room
[08:12:13] <Eliot Lear> slides are available: https://www.ietf.org/proceedings/96/slides/slides-96-plus-4.pdf
[08:12:18] <SzilveszterNadas> ack
[08:12:28] <Eliot Lear> slide 3 - what firewalls do (udp)
[08:12:34] m&m leaves the room: Disconnected: Replaced by new connection
[08:12:38] m&m joins the room
[08:12:42] wseltzer@jabber.org joins the room
[08:13:07] Brian Trammell leaves the room
[08:13:07] Spencer Dawkins joins the room
[08:13:08] Lee Howard joins the room
[08:13:08] <Eliot Lear> [all the things that come along with tcp aren't there…]
[08:13:18] <Eliot Lear> timers vary
[08:13:29] Gabriel Montenegro joins the room
[08:13:43] <Eliot Lear> variety of mechanisms to open up pinholes - not as widely implemented as joe would like
[08:13:55] <Eliot Lear> doing something in the protocol might b enice
[08:14:22] <Eliot Lear> slide 4
[08:14:29] <Eliot Lear> reducing udp blocking and rate-limiting
[08:14:42] Brian Trammell joins the room
[08:15:10] <Eliot Lear> there's an rg that might quantify traffic over time
[08:15:28] <Brian Trammell> that's maprg
[08:15:29] <Eliot Lear> it's the mix of traffic that leads to [enterprise managers wanting to limit udp]
[08:15:48] hta joins the room
[08:15:50] <Eliot Lear> slide 5
[08:16:17] <Eliot Lear> "The goal"
[08:16:28] sftcd1 joins the room
[08:16:33] gg bb joins the room
[08:16:37] Varun Singh joins the room
[08:16:46] <Eliot Lear> there are a few firewall vendors writing code and collaborating…
[08:17:01] <Eliot Lear> aaron asks for clarification questions
[08:17:19] <Martin Stiemerling> No it is not the gate on both sides!
[08:17:19] <Eliot Lear> picture is of checkpoint charlie with wall -> brandenburg gate
[08:17:30] <Eliot Lear> ok Martin- you can clarify
[08:17:37] <Eliot Lear> daniel klein gilmore
[08:17:49] <Eliot Lear> on path attacks?
[08:17:58] <Yoav Nir> *Kahn
[08:18:02] <Eliot Lear> sorry
[08:18:20] <Eliot Lear> natasha:
[08:18:27] hildjj joins the room
[08:18:34] <Eliot Lear> https://www.ietf.org/proceedings/96/slides/slides-96-plus-3.pdf
[08:18:36] <Eliot Lear> slide 1
[08:18:41] <Eliot Lear> ernestine might like PLUS too
[08:18:50] <Eliot Lear> slide 2
[08:19:01] <Eliot Lear> focusing on mobile operators
[08:19:11] <Eliot Lear> spectrum is scarce
[08:19:32] <Eliot Lear> lots of overprovisioning within a cell
[08:19:55] hildjj joins the room
[08:20:05] <Eliot Lear> a number of "IoT" devices will use mobile networks
[08:20:05] hildjj leaves the room
[08:20:10] <Eliot Lear> slide 3
[08:20:22] <Eliot Lear> mobile operators try to make the best use of their resources
[08:20:25] <Eliot Lear> slide 4
[08:20:33] <Eliot Lear> this is not new - there was marnew
[08:20:52] Brian Trammell leaves the room
[08:20:54] <Eliot Lear> there was the accord bof
[08:21:25] Brian Trammell joins the room
[08:21:45] Vaibhav Bajpai joins the room
[08:21:54] <Eliot Lear> noise and signal problems can be misconstrued as  congestion
[08:22:05] <Eliot Lear> (this is now slide 7)
[08:22:07] <Eliot Lear> ?
[08:22:23] <Eliot Lear> sorry slide 4
[08:22:29] <Eliot Lear> (it was slide 7 at accord ;-)
[08:22:34] <Eliot Lear> slide 5
[08:22:44] Brian Trammell leaves the room: Disconnected: closed
[08:22:53] <Eliot Lear> we allocate resource by resource need, based on traffic need
[08:23:00] <Eliot Lear> or traffic characteristics
[08:23:11] <Eliot Lear> bearers is one way we do this
[08:23:33] <Eliot Lear> bearer has associated qos class
[08:24:07] <Eliot Lear> most operators just offer default bearers for data and a separate one for voice
[08:24:07] <Varun Singh> Do apps set bearers? or what in the endpoint is expected to set the bearer?
[08:24:10] Cristel Pelsser joins the room
[08:24:11] <Eliot Lear> slide 6
[08:24:18] Vaibhav Bajpai leaves the room
[08:24:28] Brian Trammell joins the room
[08:24:30] <Eliot Lear> transparent proxies, dpi, etc.
[08:24:32] <Brian Trammell> martin: in an earlier version of these slides it was Brandenburg in '75 and in '00. but the visual difference was too subtle.
[08:24:35] <Eliot Lear> dpi is also dangerous
[08:24:55] <aaron> Varun: today, I think, a customer gets a single bearer for INternet traffic.  all traffic is carried by it.
[08:25:04] <frodek> yes
[08:25:16] <Eliot Lear> don't assume every operator does All of This
[08:25:21] <Eliot Lear> slide 7
[08:25:27] <Eliot Lear> slide 8
[08:25:32] <Varun Singh> The bearers as far as I can tell are not used?
[08:25:42] <Eliot Lear> how can we encrypt and make best use of resources
[08:25:47] <SzilveszterNadas> @Varun: in normal case it is decided by the mobile operator. Even if there is more than 1 BE bearer. Though 1 BE berarer is the most common.
[08:25:47] <frodek> It's used. It's one dedicated for VoLTE
[08:25:53] <Eliot Lear> slide 9
[08:25:58] <frodek> and one for the rest....
[08:26:14] <Varun Singh> I remember having a discussion around this for WebRTC, and it was concluded everything goes over the same bearer as HTTP.
[08:26:16] JoeHallCDT joins the room
[08:26:17] <Eliot Lear> want to classify a flow for technical reasons- want the lowest amount of information possible
[08:26:28] <frodek> I've still not seen one example of operators using more than this
[08:26:34] <Eliot Lear> we would like to trust in the trust model
[08:27:01] <SzilveszterNadas> More than 1 BE bearers could be used, but it is very rare today.
[08:27:17] <SzilveszterNadas> *BE=Inetrnet traffic here
[08:27:27] <Eliot Lear> @frodek when discussion comes up if you are not in the room, when Q/A comes up feel free to ask me to ask for you
[08:27:47] <Eliot Lear> Janna from from google (didn't catch the q)
[08:27:47] <frodek> I'm in the room..
[08:27:54] <Varun Singh> I am not in the room, so following the chat discussion.
[08:28:06] Paul Cosgrove joins the room
[08:28:26] <SzilveszterNadas> q was "clarify last bullet please" I guess
[08:28:35] <Eliot Lear> which bullet
[08:28:47] <SzilveszterNadas> "No trust model based on ..."
[08:28:56] <Eliot Lear> ekr: this assumes there is alignment of interest between user and the carrier
[08:29:03] <SzilveszterNadas> Jana's question I mean
[08:29:15] <Andrew Sullivan> I know that it's sort of silly for me to say this just now, but it's actually a general point
[08:29:30] csp joins the room
[08:29:36] <SzilveszterNadas> @Andrew: what do you mean?
[08:29:39] <Simone Ferlin> can the slides be found somewhere?
[08:29:41] <Andrew Sullivan> the acoustics in this room are terrible.  Please talk slowly and close to mic
[08:29:43] <Eliot Lear> no trust model>
[08:29:46] <Eliot Lear> ?
[08:29:48] <Yoav Nir> There's certainly some interesting game-theoretic issues in there
[08:30:17] <SzilveszterNadas> https://www.ietf.org/proceedings/96/slides/slides-96-plus-3.pdf
[08:30:42] <sftcd1> @chairs: there are people at the mic line waaay over to your left
[08:30:48] <SzilveszterNadas> I meant that is what Jana asked.
[08:30:48] Suzanne joins the room
[08:30:56] <Simone Ferlin> found.
[08:30:58] <craigt> I wonder if 'honesty model' rather than 'trust model' might be a better term here...
[08:30:58] Wolfgang Beck joins the room
[08:31:08] <hildjj> "incentive model"
[08:31:10] JeffH joins the room
[08:31:16] <dkg> Lorenzo
[08:31:19] <Eliot Lear> thanks
[08:31:27] mnot joins the room
[08:31:54] <Eliot Lear> lorenzo: can you share a bit more about which metrics are being used to optimized?
[08:32:05] <Eliot Lear> Natasha: no i've tried, but not possible due to commercial reasons
[08:32:22] <Eliot Lear> transparent proxies exist for a number of reasons -
[08:32:26] jsaldana leaves the room: Disconnected: Replaced by new connection
[08:32:27] Mirja Kühlewind joins the room
[08:32:29] Jose Saldana joins the room
[08:32:41] sftcd1 leaves the room
[08:33:25] sftcd2 joins the room
[08:33:37] <Eliot Lear> jana: users also need to have trust in the trust model
[08:33:45] <Eliot Lear> natasha: agrees
[08:34:50] <sftcd2> "trust model" is so not a useful term here
[08:34:52] <Eliot Lear> api- called canvas to make a 3d object.  sometimes web developers use canvas object just to start the GPU, but it was a bad trust model.
[08:35:04] <Andrew Sullivan> I'm having a very hard time understanding most of this discussion as "clarifying questions"
[08:35:22] <hta> I'm having a hard time understanding this discussion period.
[08:35:24] <craigt> sftcd_: indeed, trust is a conflated term in this forum
[08:35:34] <Eliot Lear> Now Ted Hardie
[08:35:38] <Eliot Lear> Alice and Bob look at Plus
[08:35:47] <hta> It's clear that there isn't a working trust model in the room......
[08:35:54] <Eliot Lear> https://www.ietf.org/proceedings/96/slides/slides-96-plus-2.pdf
[08:36:18] <Eliot Lear> slides 2-5
[08:36:21] <Eliot Lear> (animated)
[08:36:23] Lee Howard leaves the room
[08:36:31] <Eliot Lear> slide 6
[08:36:49] DanYork leaves the room
[08:36:49] <hildjj> talking about this in terms of "aligned incentives" is potentially useful.
[08:36:59] <craigt> i prefer 'honesty' as with an honesty bar etc. etc. you expect to get ripped off ocassionally, but on balance, should be up...
[08:37:12] Brian Trammell leaves the room
[08:37:13] <Suresh Krishnan> @Meetecho: The slides are clipping on the right
[08:37:13] <Eliot Lear> apart from doing "nice" things, they may copy stuff to vast data stores
[08:37:16] <Eliot Lear> enter encryption
[08:37:19] <Eliot Lear> slide 8
[08:37:27] csp leaves the room
[08:37:29] <Eliot Lear> @suresh: slides at https://www.ietf.org/proceedings/96/slides/slides-96-plus-2.pdf
[08:37:29] <Meetecho> Suresh Krishnan: already answered on that
[08:37:38] <Meetecho> the capture card is causing this
[08:37:40] <SzilveszterNadas> @Eliot: sry for not being clear. I will use "MIC:" if I would like a question at mic.
[08:37:50] <Meetecho> it's only a small part of that, it should still be readable
[08:37:52] <Eliot Lear> slide 7
[08:37:58] <Suresh Krishnan> @Meetecho: Thx.
[08:38:00] <Eliot Lear> ok
[08:38:10] <Eliot Lear> slide 9
[08:38:28] <Eliot Lear> there was an inference to create and maintain state in the network
[08:38:37] <Eliot Lear> slide 10 options
[08:38:46] <Eliot Lear> state is being impaired by encryption
[08:39:01] Brian Trammell joins the room
[08:39:04] George Tsolis joins the room
[08:39:05] <Eliot Lear> we could hop by hop in ipv6, for instance at the l3
[08:39:20] Toru Asahina leaves the room: Disconnected: Replaced by new connection
[08:39:20] Toru Asahina joins the room
[08:39:21] <Eliot Lear> could bolt signals onto each udp-based transport
[08:39:27] Lee Howard joins the room
[08:39:45] <Kyle Rose> I know what we need: source routing! j/k
[08:39:55] <Eliot Lear> risk that first one to do this well will [experience wild success]
[08:40:02] <Eliot Lear> or we could do nothing and leave path impaired
[08:40:30] <Eliot Lear> we could be stuck with state timers in UDP is what are used for everything on top and that's not a good thing
[08:40:33] <Eliot Lear> we could encrypt less
[08:40:37] <Eliot Lear> or
[08:40:39] <Eliot Lear> slide 16
[08:40:48] <Eliot Lear> make implicit layer explicit
[08:40:55] Kristopher Price joins the room
[08:41:01] <Eliot Lear> Christian Huitema
[08:41:18] Spencer Dawkins leaves the room
[08:41:28] <JoeHallCDT> CHRISTIAN HUITEMA!!!!!!!
[08:41:40] <Eliot Lear> concerned about use of layers (mic dropped)
[08:41:43] Spencer Dawkins joins the room
[08:42:11] resnick joins the room
[08:42:20] <Eliot Lear> we have tried some of this before with, for example, shim6
[08:42:25] <Eliot Lear> it might be functions rather than layers
[08:42:34] <Eliot Lear> ted: fair point
[08:42:56] <Eliot Lear> alain durand: when we talk about making this explicit we have pcp..
[08:42:57] <tale> Who is at mic?
[08:43:01] <tale> ah alain
[08:43:03] <Andrew Sullivan> Alain Durand
[08:43:12] <resnick> I thought for sure when he started talking about layers of onion, with his French accent, that we were going to start discussing Derrida.
[08:43:27] <Andrew Sullivan> It is possible we have
[08:43:28] <Eliot Lear> Tom Herbert:
[08:43:35] DanYork joins the room
[08:43:43] <Eliot Lear> path layer -> paths layer
[08:43:47] <Eliot Lear> there are multiple paths
[08:43:52] <Andrew Sullivan> Is there a clarifying question here?
[08:44:19] <Andrew Sullivan> (I am pretty sure that "clarifying question" does not usually require >1 sentence of context)
[08:44:28] <Lee Howard> ghee
[08:44:41] <Eliot Lear> Ted: is this model be limited to a single path?  NO
[08:44:43] <hildjj> Please phrase your comment in the form of a question.
[08:45:02] <SzilveszterNadas> Who is at mic?
[08:45:03] <Eliot Lear> Joanna @ Google: concerned about ossification
[08:45:07] <SzilveszterNadas> thx
[08:45:31] Magnus Westerlund joins the room
[08:45:35] <Eliot Lear> concerned about middleboxes making hypothoses about what is happening that themselves will ossify
[08:45:38] <Barry Leiba> (Remoters: please give the jabber scribe a chance before asking)
[08:45:40] <Eliot Lear> Ted: it's worth trying
[08:46:07] <Eliot Lear> it's possible middle boxes could change with "onslaught" of NFV et al
[08:46:13] <Eliot Lear> want to try to improve the situation
[08:46:23] <Eliot Lear> stewart bryant: where does routing fit into this?
[08:46:36] <Eliot Lear> ted: routing system acts on the internet layer alone
[08:46:54] <Eliot Lear> chair: let's get through the proposal
[08:47:25] <Eliot Lear> erik klein: should the information be modifyable?
[08:47:27] <Eliot Lear> Ted: no
[08:48:04] <Eliot Lear> Tiru: what problem is this bof addressing?
[08:48:09] <Eliot Lear> Chair: wait
[08:48:18] <Eliot Lear> Now Brian Trammell
[08:48:23] Hirochika Asai joins the room
[08:48:31] <Eliot Lear> https://www.ietf.org/proceedings/96/slides/slides-96-plus-1.pdf
[08:48:34] <Eliot Lear> slide 1
[08:48:53] <Eliot Lear> slide 3
[08:49:05] <Eliot Lear> endpoints and middleboxes do implicit cooperation all the time
[08:49:29] <Eliot Lear> slide 4
[08:49:49] <Eliot Lear> move implicit -> explicit
[08:51:09] <Eliot Lear> sample verification as an example of the honesty model
[08:51:24] <Eliot Lear> maybe you can't verify all flows at line rate....
[08:51:41] ddolson joins the room
[08:51:50] <Eliot Lear> anything that the path doesn't need to see gets encrypted (including transport hdrs)
[08:52:20] <Eliot Lear> we solve the ossification boundary by moving the boundary
[08:52:22] <Eliot Lear> slide 5
[08:52:25] <sftcd2> I've no idea how an endpoint knows (or agrees with) "what the devices on the path *need* to see
[08:52:33] <Eliot Lear> 3.5 mechanisms
[08:52:53] <ted.h> @sftcd2  it may know it wants the path to maintain state, so it does not have to heartbeat, as a concrete example
[08:53:09] <Eliot Lear> slide 6
[08:53:41] <sftcd2> @ted: isn't that the endpoint knowing what it wants but not what the "path needs"?
[08:54:01] <Eliot Lear> slide 7
[08:54:07] <Eliot Lear> back to slide 6
[08:54:11] Michael Welzl joins the room
[08:54:18] <Eliot Lear> adding a path layer
[08:54:30] <Eliot Lear> slide 7
[08:54:37] <tale> What was ekr trying to say?
[08:54:43] <Eliot Lear> couldn't hear
[08:54:50] <Eliot Lear> slide 8
[08:54:52] m&m leaves the room: Disconnected: closed
[08:55:05] <Mirja Kühlewind> @sftcd2: I guess the path has to take what it gets
[08:55:13] m&m joins the room
[08:55:21] <SzilveszterNadas> @sftcd2: it wants good service, if the path needs something, it might give worse service if it does not get what it needs.
[08:55:23] Kristopher Price leaves the room
[08:55:29] <craigt> He was going to say somthing, then said, "I'll get back to you later"...
[08:55:29] <Eliot Lear> slide 9
[08:55:58] <Eliot Lear> "start" map to syn example
[08:56:03] <Eliot Lear> slide 10
[08:56:06] <Eliot Lear> path to receiver is harder
[08:56:13] <sftcd2> so the  bullet that says "Encrypt everything devices on path don’t need to see" should be "Encrypt everything devices on path don’t decide it wants to see" ?
[08:56:24] <Eliot Lear> sender authorizes the path...
[08:56:36] ddolson leaves the room
[08:56:38] Brian Trammell leaves the room: Disconnected: closed
[08:56:41] <Mirja Kühlewind> no its sender control
[08:56:50] <Mirja Kühlewind> the sender decides what to expose
[08:56:52] <Spencer Dawkins> Stephen, from the milestones in the BOF wiki: Chartering + 18 mo: Submit initial contents of registry of signals to be exposed using PLUS to the IESG
[08:57:05] <Mirja Kühlewind> usually these information are traffic characteristics
[08:57:12] <Eliot Lear> slide 12
[08:57:15] Lee Howard leaves the room
[08:57:21] <SzilveszterNadas> @sftcd: see slide 19
[08:57:23] <Spencer Dawkins> I believe that's where the can of worms is opened
[08:57:33] <Eliot Lear> slide 13
[08:57:43] <Eliot Lear> feedback
[08:57:47] <Eliot Lear> path layer only
[08:57:53] <Eliot Lear> and feedback is encrypted
[08:57:58] <Eliot Lear> path direct to sender
[08:58:13] <Eliot Lear> middle box wants to inform sender?
[08:58:30] ddolson joins the room
[08:58:33] <Mirja Kühlewind> @eliot are there people that don't have meetecho and need this transcript?
[08:58:38] <Spencer Dawkins> Stephen, I also suspect that the repeated use of the word "tussle" is not a coincidence ...
[08:58:58] <Eliot Lear> slide 17
[08:59:06] Lee Howard joins the room
[08:59:12] <Eliot Lear> we get the ports
[08:59:12] <Spencer Dawkins> Mirja, they also do audio-only plus jabber
[08:59:22] <Lorenzo Miniero> 24/62 are on Meetecho, not sure how many of those who aren't are actually in the room
[08:59:33] <Barry Leiba> Mirja: What Eliot's doing is usually requested.  If you think it's not necessary, discuss in IESG
[08:59:39] <Eliot Lear> discussion on spud
[08:59:42] <Andrew Sullivan> That's an unnamed Paul Hoffman at the mic
[08:59:51] <Eliot Lear> (thanks andrew)
[08:59:57] <Eliot Lear> credits now on slide 19
[09:00:00] <JoeHallCDT> well, there is etherpad for notes
[09:00:48] Roman Müntener joins the room
[09:00:49] <JoeHallCDT> http://etherpad.tools.ietf.org:9000/p/notes-ietf-96-plus?useMonospaceFont=true
[09:00:52] <Eliot Lear> can middlebox expose user identifers?  no
[09:00:53] <tale> of questionable quality
[09:01:22] <Eliot Lear> well if someone wants me to continue, say so now
[09:01:35] Roman Müntener leaves the room
[09:01:35] <Eliot Lear> that was ekr
[09:01:47] <Eliot Lear> daniel kahn gilmore
[09:01:52] <tale> *I* want you to continue
[09:01:55] <Eliot Lear> ok
[09:01:59] <tale> It's helping augment etherpad
[09:02:00] Roman Müntener joins the room
[09:02:18] <theresa> I also appreciate it, even though I'm in the room.
[09:02:24] Magnus Westerlund leaves the room
[09:02:27] <Eliot Lear> Dkg: that no makes it seem like it's something the client opts into when it's not the case
[09:02:28] Wolfgang Beck leaves the room
[09:02:30] Gorry Fairhurst joins the room
[09:02:36] <Spencer Dawkins> eliot, I'd feel funny asking you to stop in the middle. a PLUS WG could decide what tools they'll emphasize :D
[09:03:04] <Eliot Lear> goran k from ericsson: is this about coordination?
[09:03:27] <Eliot Lear> goran k: can you list the functions the middlebox to do and what we can expect to do?
[09:03:28] <JoeHallCDT> it would be nice if it weren't in Comic Sans, Eliot… but maybe that's on my side
[09:03:45] <Barry Leiba> I see no Comic Sans here.
[09:03:53] <JoeHallCDT> kk
[09:03:53] <Andrew Sullivan> Me either
[09:03:56] <Spencer Dawkins> JoeHallCDT, same as Barry
[09:04:14] <JoeHallCDT> (might be an Adium thing)
[09:04:18] <Eliot Lear> chrisitan huitema: do you have an estimate of the overhead?
[09:04:27] <Eliot Lear> (sorry- joe, I can't fix it right now)
[09:04:27] Jose Saldana leaves the room: Disconnected: Replaced by new connection
[09:04:27] Jose Saldana joins the room
[09:04:37] <tale> I am amused that Joe set up his client to use comic sans
[09:04:44] <JoeHallCDT> (please continue being useful!)
[09:04:44] <Eliot Lear> 12 bytes of overhead
[09:04:44] <JoeHallCDT> I did not
[09:04:49] <tale> You must have!  User error!
[09:04:50] <Eliot Lear> (it's me)
[09:04:55] <tale> (I kid, I kid.)
[09:05:08] <Eliot Lear> aaron: return from implementation rathole
[09:05:35] <Eliot Lear> christian: many count bits to avoid exceeding threshold.  important to quantify
[09:05:59] Fuyou Miao joins the room
[09:06:22] <Eliot Lear> lorenzo giletti: IETF designing SHOULD before HOW for 25 years (or visa versa)
[09:06:26] magnus joins the room
[09:06:38] <Eliot Lear> Lorenzo: what is the problem statement?  what is this supposed to solve?
[09:06:41] <Eliot Lear> Brian:
[09:07:03] <Eliot Lear> 1. i'd like to be able to encrypt the transport headers, and expose to the path things that operators use
[09:07:32] <Eliot Lear> 2. there are few things that are used to classify traffic - have method to provide more informaiton under endpoint control
[09:08:18] jm leaves the room
[09:08:32] jm joins the room
[09:09:03] <Lee Howard> From https://tools.ietf.org/agenda/96/agenda-96-plus.html
[09:09:07] <Eliot Lear> slide 19
[09:09:11] <Lee Howard> Relevant drafts:
  Requirements for the design of a Substrate Protocol for User Datagrams
  ​https://tools.ietf.org/html/draft-trammell-spud-req-04
  Use Cases for a Substrate Protocol for User Datagrams
  ​https://tools.ietf.org/html/draft-kuehlewind-spud-use-cases-01
[09:09:29] Kristopher Price joins the room
[09:09:31] yuki goto leaves the room
[09:09:33] <Eliot Lear> thanks lee
[09:09:41] yuki goto joins the room
[09:09:42] <dkg> Lee Howard: thanks
[09:09:58] <Mirja Kühlewind> See all relevant information on the bof page: https://trac.tools.ietf.org/bof/trac/
[09:10:08] <Mirja Kühlewind> plus is at the very bottom
[09:10:11] <Eliot Lear> back to slide 10
[09:10:30] <Eliot Lear> and back to slide 19
[09:10:33] <Eliot Lear> sorry 20
[09:11:36] <Eliot Lear> what to share is a choice in this protocol
[09:11:40] <Eliot Lear> slide 21
[09:11:46] <Eliot Lear> can we use ipv6 extension headers?
[09:11:52] <Eliot Lear> hbh headers
[09:12:29] Cristel Pelsser leaves the room
[09:12:56] <ted.h> hbh also intentionally modifiable by path, so mac-type protection not quite the same.
[09:13:25] <Yoav Nir> Don't HBH headers force the packets into the slow path in many routers, leading to negating pretty much all of any benefit in bandwidth utilization we hope to gain?
[09:13:26] <Eliot Lear> falling back on something that works 61% is prayer
[09:13:31] <Eliot Lear> slide 22
[09:13:38] <Eliot Lear> can we use udp options?
[09:13:59] <Eliot Lear> no way to do user space implementation
[09:14:19] <Eliot Lear> Slide: Do we need to choose now?
[09:14:20] Kristopher Price leaves the room
[09:14:21] Kristopher Price joins the room
[09:14:47] <Eliot Lear> slide 24
[09:14:51] <Spencer Dawkins> The TSVWG quote from yesterday was that in recent testing, packets with the evil bit set are more likely to traverse the network than most other IP options ...
[09:14:52] <Eliot Lear> slide 25
[09:15:06] <Eliot Lear> Things we need
[09:15:40] <Eliot Lear> Chair: clarification questions
[09:15:53] <Eliot Lear> Tim Shepherd: would you be adding an RTT?
[09:16:06] Toru Asahina leaves the room: Disconnected: Replaced by new connection
[09:16:06] Toru Asahina joins the room
[09:16:13] Wolfgang Beck joins the room
[09:16:59] <Eliot Lear> Brian: no - that's A Detail ;-) anything that gets exposed before crypto is set up doesn't have a mac.  and then once you have the mac you can repeat the info
[09:17:50] <Eliot Lear> Kyle Rose: on HbH slide: what do you mean by inherent?
[09:17:50] m&m leaves the room: Disconnected: Replaced by new connection
[09:17:51] m&m joins the room
[09:17:56] <Eliot Lear> Brian: 1- probability of packet loss
[09:18:07] <sftcd2> the issue Tim raised seems to me like it may be a killer
[09:18:28] <Eliot Lear> Jana: on slide 20
[09:18:48] <ted.h> @sftcd2  If you are keeping state for 0-RTT, you could keep the state required for the mac as well.
[09:19:00] <Eliot Lear> Jana: does this mean one instantiation is that there is shim with nothing in there?
[09:19:22] <Eliot Lear> Jana: it's a very important point and a very good slide, and that puts control back in endpoints.
[09:19:35] <ted.h> @sftcd2 otherwise you do have to do the set up.
[09:19:43] Spencer Dawkins leaves the room
[09:19:51] <Eliot Lear> Jana: don't call it implicit.  it was coopation.  new way is cooperation
[09:19:56] <JoeHallCDT> and there's always coopetition
[09:20:03] <Eliot Lear> tom herbert: on slide 25
[09:20:52] <Eliot Lear> ????:
[09:21:15] Lee Howard leaves the room
[09:21:43] <Eliot Lear> does the client who is sending a reply from the middlebox?
[09:21:52] <Eliot Lear> brian: yes, but it is something we can define
[09:22:29] <SzilveszterNadas> @Eliot: Is not it Zahed Sarker?
[09:22:33] <Eliot Lear> could be
[09:23:04] <tale> ty Sz
[09:23:09] <SzilveszterNadas> yw
[09:23:12] <Eliot Lear> Joanna @ Google: having trouble understanding how this is all going to work on initial or partial deployment - slight fear- what if the middleboxes give preferential treatment to flows that use these markings
[09:23:20] <Michael Welzl> yes it was zahed
[09:23:30] <Eliot Lear> Brian: phase zero looks like this (slide 20)
[09:23:36] Michael Welzl leaves the room
[09:23:46] <ted.h> Showing slide 9
[09:24:01] <Eliot Lear> slide 10 is years away
[09:24:15] <Eliot Lear> phase 1 looks like quc
[09:24:17] <Eliot Lear> quic
[09:24:19] <sftcd2> so day 0 is where we want to be, and then we work backwards from there to more exposure... hmm
[09:24:53] <Eliot Lear> dkg: slide 19
[09:25:15] Lee Howard joins the room
[09:25:23] <Eliot Lear> dkg: you designed 6 other ways… the other bad ways were none-transparent?
[09:25:41] <Eliot Lear> dkg: so this doesn't solve the problem of non-transparent signal
[09:25:44] <Eliot Lear> Brian: correct
[09:25:51] richsalz joins the room
[09:26:15] <Eliot Lear> Brian: you need the middle box to care about this-  but in phase 0 it just looks like udp traffic
[09:26:19] Wolfgang Beck leaves the room
[09:26:38] <Eliot Lear> dan druta:
[09:26:40] <Eliot Lear> ?
[09:26:58] <sftcd2> that's a clarifying thanks:-)
[09:26:59] <ted.h> @sftcd2  if you don't care about the path impairment or its consequences (e.g. higher heartbeat rates), then you don't ever start this.  But there is actual impairment.
[09:27:07] <Eliot Lear> thanks for the work and for keeping the momentum on this dialog.  agree with jana
[09:27:14] <Eliot Lear> there is conflicting optimization
[09:27:26] <Eliot Lear> this is making collaboration happen
[09:27:51] <Andrew Sullivan> that doesn't sound like a question to me
[09:27:58] <Eliot Lear> murthu?:
[09:28:18] <craigt> Muthu
[09:28:22] <Mirja Kühlewind> it was a clarification, without a question... :-(
[09:28:26] <Eliot Lear> is this a udp problem or an encryption problem and if it is an encryption problem which one?
[09:28:39] <craigt> can'y read the surname, apols
[09:28:51] Carsten Bormann leaves the room
[09:28:56] <craigt> Tiru Konda (Cisco)
[09:28:58] <Eliot Lear> Tiru from Cisco:
[09:29:25] <Eliot Lear> how will this get deployed faster than previous mechanisms like pcp?
[09:29:45] <Eliot Lear> Brian: this is in band and PCP was out of band.  you're talking to middle boxes that you don't know exists
[09:30:15] <resnick> Good, I was going to ask that question but was too lazy to go to the mic. :-)
[09:30:17] <Eliot Lear> Tiru: PCP-proxy [might be useful in this context]
[09:30:26] <Andrew Sullivan> If you're going to the mic, you _really need_ to speak slowly and loudly into the mic.  The acoustics here are bad
[09:30:53] <Eliot Lear> Tiru: PCP has security mechanisms to [establish explicit trust]
[09:31:53] <Eliot Lear> Brian: how to deal with case of [non-participating PCP/fw]? this is declarative as opposed to imperative.
[09:32:49] <Eliot Lear> Ok Brian will sit down
[09:33:04] ted.h leaves the room
[09:33:13] <Eliot Lear> Now back to the chair's deck
[09:33:17] <Eliot Lear> slide 8
[09:33:21] nygren leaves the room
[09:33:32] <Suzanne> Well done Aaron on tagging the proponents :-)
[09:34:05] <sftcd2> seems to me the extesible registry of things to expose is a key scope point of the charter
[09:34:23] <Eliot Lear> https://www.ietf.org/proceedings/96/slides/slides-96-plus-0.pdf
[09:35:04] <craigt> About 16 ppl waiting at the mics
[09:35:10] <Eliot Lear> jana: make 3rd line "optional signaling control"
[09:35:43] Michael Welzl joins the room
[09:35:59] Gorry Fairhurst leaves the room
[09:36:11] <Eliot Lear> jana: there is an end to end shim with the possibility of adding information that the senders may add and the receiver may read.
[09:36:15] <Eliot Lear> jana: that's a good thing
[09:36:22] <Eliot Lear> jana: but there's something fundamental going on here
[09:36:50] <Eliot Lear> jana: there have been proposals where devices in the network have participated in this network (xtp? comes to mind).
[09:37:28] <Eliot Lear> jana: my concern is that we won't be talking about these things if the network expects the clients to signal its b/w requirements...
[09:38:15] Wolfgang Beck joins the room
[09:38:16] <Eliot Lear> Ted: you are concerned about the network adding requirements that would not make sense.  We're starting with flow semantics is because those semantics are clear.  The registry for this model should be as small as can be.
[09:39:22] ddolson leaves the room
[09:39:45] <Eliot Lear> Mirja: we're not talking about all specific use cases because people would want to go to the mic and talk about theirs…  just focus on the very basic use case.
[09:40:09] <Eliot Lear> if there are any jabber comments for the mic please indicate with "MIC:"
[09:40:19] <Eliot Lear> now slide 9
[09:40:23] <sftcd2> I don't get why we'd want an "as small as possible" registry - if there's a short list of stuff that we don't want expanded just document those in an RFC
[09:41:04] <Eliot Lear> eliot (speaking as self): +1 to sfcd2
[09:41:24] <SzilveszterNadas> @sftcd2: +1
[09:41:24] <Eliot Lear> alain durand:
[09:41:52] <Eliot Lear> i thought we were talking about inband versus out-of-band
[09:41:56] <Michael Welzl> what is that very basic use case?
[09:42:21] <Eliot Lear> alain: how will this impact other things, like existing ICMP messages...
[09:42:30] <Eliot Lear> david black:
[09:42:50] Samuel Jero leaves the room
[09:42:51] <Eliot Lear> brian how does discussion of phases relate to scope of working group and the initial target?
[09:43:24] <Eliot Lear> brian: we're fixing the ossification problem by re-ossifying and we get to do that once
[09:44:19] <Eliot Lear> not phases of work of protocol design but phases of deployment
[09:44:37] Carsten Bormann joins the room
[09:44:37] <Eliot Lear> Sebotha:
[09:45:01] <Eliot Lear> might firewalls start viewing non-signaling endpoints as suspicious
[09:45:14] <Eliot Lear> Joe: in some of these networks all udp content is susupicious now
[09:45:39] <Eliot Lear> part of this is getting the incentive structure correct.  so operators need to understand the ramifications of abusing what information they get
[09:45:40] <resnick> Just notified secretariat that the mics in here are set too quietly.
[09:46:25] <Kyle Rose> @resnick: Thank you
[09:46:38] <craigt> Zahed Sarker at the mic
[09:46:41] <Eliot Lear> Zahed: how many [vendors] are willing to do this?  [Recounts ECN experience]
[09:46:44] Kristopher Price leaves the room
[09:47:00] sjero joins the room
[09:47:12] <Eliot Lear> Ted Hardie:
[09:47:28] <Eliot Lear> the more people who recognize the path impairment, the more who will be willing to deploy
[09:47:31] <Martin Stiemerling> One question: How does the charter recognize and value existing IETF work in this space, namely MIDCOM, NSIS, PCP ,etc?
[09:47:51] <Martin Stiemerling> espcially NSIS as this is in-path/path-coupled?
[09:47:55] <Eliot Lear> we can't give you a solid estimate now becuase the work isn't done
[09:48:11] <Martin Stiemerling> that is a question to the mic, please?
[09:48:12] <Eliot Lear> mirja: this is not a tiny change.
[09:48:16] <Eliot Lear> ok
[09:48:20] <Martin Stiemerling> thx
[09:48:24] ddolson joins the room
[09:48:56] <SzilveszterNadas> @Meetecho: video is frozen. Even when pressing refresh
[09:49:02] Lee Howard leaves the room
[09:49:13] <Meetecho> Szilveszter Nadas: fixing that
[09:49:28] <Eliot Lear> Zaher: should i be hopeful?
[09:49:39] <SzilveszterNadas> Now evertything is gone
[09:49:40] <aaron> Zahed
[09:49:44] <resnick> Take the hums now so that when you have no idea what they mean when people hum “no”, you can have them come up to the mic and discuss.
[09:49:44] mls.ietf@gmail.com joins the room
[09:49:48] mls.ietf@gmail.com leaves the room
[09:50:02] <Kyle Rose> Multipass
[09:50:04] Gabriel Montenegro_2331 joins the room
[09:50:08] <Eliot Lear> Eric Vyncke: to we take into account that this is Unicast/multicast?
[09:50:13] <SzilveszterNadas> @Meetecho: voice is also gone
[09:50:19] <Meetecho> technical difficulties, it should be restarting right now
[09:50:24] <SzilveszterNadas> ack
[09:50:29] Simone Ferlin_3661 joins the room
[09:50:33] Martin Stiemerling leaves the room
[09:50:34] gg bb leaves the room
[09:50:34] Gabriel Montenegro leaves the room
[09:50:35] Martin Stiemerling joins the room
[09:51:00] Mirja Kühlewind leaves the room
[09:51:04] Lorenzo Miniero leaves the room
[09:51:05] Simone Ferlin leaves the room
[09:51:07] <Martin Stiemerling> thx!
[09:51:10] Paul Cosgrove leaves the room
[09:51:10] Lorenzo Miniero joins the room
[09:51:17] <Martin Stiemerling> will do :)
[09:51:20] <Eliot Lear> Ted hardie: send charter text, martin, to the list
[09:51:25] <Meetecho> everything should be back now
[09:51:25] Simon Pietro Romano leaves the room
[09:51:30] <Eliot Lear> Harald:
[09:51:32] Simon Pietro Romano joins the room
[09:51:32] yuki goto leaves the room
[09:51:33] <Meetecho> if you still don't et anything, try rejoining
[09:51:37] Catherine Truchan leaves the room
[09:51:37] <Meetecho> *get
[09:51:38] Catherine Truchan joins the room
[09:51:40] yuki goto joins the room
[09:51:55] <Eliot Lear> is it already decided whether the extra information goes into the same packet as the payload?
[09:52:05] Paul Cosgrove joins the room
[09:52:19] <Eliot Lear> brian: no but if your are in a situation where it doesn't, you have to build a transport protocol
[09:52:36] <Eliot Lear> harald: is this a new version of RSVP and i'm worried about your mtu
[09:52:45] Roman Müntener leaves the room
[09:52:46] hildjj leaves the room
[09:52:50] cheshire joins the room
[09:52:58] <Eliot Lear> Brian: we're worried about that too but we're using that as a feature to increase the parsiminy of the signaling
[09:53:20] Hirochika Asai leaves the room
[09:53:20] Szilveszter Nadas leaves the room
[09:53:34] <Eliot Lear> Christian H: I have a really bad feeling.  10% increase?
[09:53:43] Szilveszter Nadas joins the room
[09:53:52] <Eliot Lear> Joe H: it's possible this will not be additional overhead, but the mechanisms may already be there.
[09:54:06] <Eliot Lear> Christian: that's a terrible idea.
[09:54:10] Wolfgang Beck leaves the room
[09:54:16] <Martin Stiemerling> +1 to harald
[09:54:42] Martin Stiemerling leaves the room
[09:54:43] <Eliot Lear> Brian: not a shim layer, but a definition of the first few bits of a transport protocol
[09:54:44] Martin Stiemerling joins the room
[09:54:50] Lee Howard joins the room
[09:55:22] <Martin Stiemerling> and I also do not see how plus will fix the deployment challenges NSIS (and others) stumbled over. But thanks in general for the very short answer ;)
[09:55:25] <Eliot Lear> Christian: use cases mix stuff that is widely recognized and stuff that is a really bad idea
[09:55:36] <Martin Stiemerling> (this is just for the record )
[09:55:59] ddolson leaves the room
[09:56:06] <Eliot Lear> Brian: thinking of restricting vocabulary
[09:56:58] <sftcd2> that registry idea makes no sense at all and definitely has the danger Christian mentioned
[09:57:10] JoeHallCDT leaves the room
[09:57:21] <Eliot Lear> Christian: don't make it extensible
[09:57:25] <SzilveszterNadas> @sftcd2: please explain more teh second part
[09:57:43] <Eliot Lear> Lorenzo:
[09:57:57] <sftcd2> if there exists a registry then people will ask to define dodgy entries (e.g. "over 18")
[09:58:02] Ole Trøan leaves the room
[09:58:44] <Eliot Lear> we will ossify on a tcp-like semantic with only one ???.  if we want to go to a world with multipath is common as useful to users, we must support multipath.
[09:58:52] <Martin Stiemerling> and you can use registry values w/o asking for them, right?!
[09:59:06] <Martin Stiemerling> and demand usage of them in deployments.
[09:59:09] stere preda leaves the room
[09:59:10] <sftcd2> squatting has indeed happened
[09:59:11] stere preda joins the room
[09:59:24] <Eliot Lear> is multipath in scope or out of scope?
[09:59:49] Carsten Bormann leaves the room
[10:00:02] <Eliot Lear> Joe H: i do believe it is possible to get better multipath approaches, but they have not been described here.
[10:00:09] <Eliot Lear> but we've had an "aha moment"...
[10:00:53] <Eliot Lear> Chair: should multipath be in scope?
[10:00:59] <Eliot Lear> Mirja: yes
[10:01:06] ddolson joins the room
[10:01:09] <Eliot Lear> by acclaimation
[10:01:11] <Eliot Lear> Erik: Nygren:
[10:01:22] <Eliot Lear> (a momentary interrupt for bluesheets)
[10:01:48] <Eliot Lear> can the charter be more explicit as to what the experiment is?
[10:01:52] Carsten Bormann joins the room
[10:01:59] <Eliot Lear> what is the bare minimum that everyone really agrees on?
[10:02:32] ddolson leaves the room
[10:02:36] <Eliot Lear> [worried about registries]
[10:02:45] <Eliot Lear> What are we ossifying onto?
[10:03:03] <Eliot Lear> Is experiment how quickly do people ossify onto an experiment?
[10:03:21] <sftcd2> yeah, is there no protocol draft at all?
[10:03:26] Kevin Ma leaves the room
[10:03:28] <SzilveszterNadas> @sftcd2: you said both " don't get why we'd want an "as small as possible" registry" and "if there exists a registry then people will ask to define dodgy entries" - kind of contradictory for me
[10:03:36] <Eliot Lear> Tom Herbert:
[10:03:39] <SzilveszterNadas> @sftcd2: could you elaborate?
[10:03:49] nygren joins the room
[10:04:00] <Eliot Lear> i know about a middle box I might be able to trust it
[10:04:10] <Eliot Lear> but otherwise less so
[10:04:13] <sftcd2> as christian said: if this is extensible, it's dangerous  => don't want any registry at all
[10:04:15] <Eliot Lear> what is the scope of the signaling?
[10:04:22] <Eliot Lear> ted hardie:
[10:04:28] <Eliot Lear> tell the path what you need from the path
[10:04:30] <SzilveszterNadas> @Meetecho: video is gone again :)
[10:04:38] <Eliot Lear> tom herbert:
[10:04:55] <Eliot Lear> in mobility, maybe only the 1st hop is interesting
[10:05:20] <SzilveszterNadas> @sftcd2: I see. Are you completely against the idea, or are you supporting some parts?
[10:05:24] <Eliot Lear> Ted: this is multipath and under sender control
[10:05:44] <Eliot Lear> Eliot: question for the record- what would happen if the 1st hop stripped this path layer?
[10:05:45] <sftcd2> I'm not expressing a for/against opinion of the overall right now
[10:06:03] <hta> can we use the evil bit instead of a new shim layer?
[10:06:11] <Eliot Lear> Stewart Bryant: is this a new layer or what layer would it be in?
[10:06:13] <Stephan Neuhaus> :D
[10:06:49] <Eliot Lear> Mirja: routing is out of scope but there is interaction with the network layer
[10:07:02] <resnick> Mirja didn’t understand the question.
[10:07:03] <Eliot Lear> Mirja: and I think we have them in the room.
[10:07:19] <resnick> And hence didn’t answer Stewart.
[10:07:32] <Eliot Lear> Wendy Seltzer: support tom and christian.  a more restrictive language would be more valuable.  fewer side channel issues fewer privacy issues
[10:07:59] <Eliot Lear> Wendy: charter had opinionated language about how necessary this work is
[10:08:03] <Martin Stiemerling> @Pete, questions are sometimes not really answered...
[10:08:04] Hirochika Asai joins the room
[10:08:16] <Eliot Lear> Randy Stewart:
[10:08:25] <Eliot Lear> in both user space and kernel space...
[10:08:38] Sebastian Kiesel joins the room
[10:08:41] <Eliot Lear> we did this within SCTP
[10:08:45] <Eliot Lear> and we had good results
[10:09:11] Suresh Krishnan leaves the room
[10:09:12] <Eliot Lear> question: i quic going on and i see this going on.  how does plus interact with quic?
[10:09:22] <Eliot Lear> Ted H: quic wouldn't have to wait for this
[10:09:45] <Eliot Lear> if quic had been done later it might have used these facilities...
[10:10:04] <Eliot Lear> brian: these two efforts will coevolve.
[10:10:35] sjero leaves the room
[10:10:46] Hirochika Asai leaves the room
[10:10:49] <Eliot Lear> joanna @ google: what if the transport the transport ossifying even further is TCP? do i have to wrap a shim layer in udp on top?  tcp?
[10:11:04] jm leaves the room
[10:11:05] jm joins the room
[10:11:25] <Eliot Lear> joe: idea was to [prove/test] this using an existing transport
[10:11:38] <JeffH> new networking notion: "ossifiability"
[10:12:03] <Eliot Lear> @JeffH: not new. that's the waist of the hour glass
[10:12:29] <SzilveszterNadas> HUM YES
[10:12:30] <Stephan Neuhaus> hummmmm
[10:12:31] <Paul Cosgrove> hummmmm
[10:12:33] <Eliot Lear> humm for useful
[10:12:35] <Eliot Lear> disagree
[10:12:52] <Eliot Lear> 5 minutes for why you disagree
[10:12:56] Martin Stiemerling leaves the room
[10:12:58] Martin Stiemerling joins the room
[10:13:07] hildjj joins the room
[10:13:17] ddolson joins the room
[10:13:27] ddolson leaves the room
[10:13:29] <Eliot Lear> paul hoffman: wasn't clear what was being hummed against
[10:13:37] <Eliot Lear> Andrew: same issue
[10:14:26] Jose Saldana leaves the room: Disconnected: Replaced by new connection
[10:14:32] <Eliot Lear> dkg: as explicit cooperation.  no analysis over who has the power- the network or the endpoint.  incentives may not align and could lead to coersion
[10:14:54] jsaldana joins the room
[10:15:29] <Eliot Lear> dkg: not worse with than without, but if we've made a mechanism that requires opt in [that's cooption]
[10:16:36] Martin Stiemerling leaves the room
[10:16:38] <Eliot Lear> natasha: we have a network management issue.  there is a small amount of information we need and we want [to align incentives]
[10:17:24] <Eliot Lear> lorenzo: if you don't do this, then i will deny service
[10:17:46] <Eliot Lear> customer can then complain out of band
[10:18:08] <Eliot Lear> if we build a clean framework to communicate how this happened, that's way worse
[10:18:19] <Eliot Lear> mirja: how do you enforce this?
[10:18:46] <Eliot Lear> Christian: the moment the IETF publishes a standard, it changes the balance of power
[10:18:58] Gorry Fairhurst joins the room
[10:19:18] <Eliot Lear> Brian: i can already force you to use the lower 64
[10:19:29] Michael Welzl leaves the room
[10:19:54] ddolson joins the room
[10:20:00] <Eliot Lear> Mirja: we have to decide in the wg
[10:20:05] <sureshk> @Eliot: Not sure. At least in the networks Natasha is talking about
[10:20:11] <nygren > Should we have a hum on whether people would be more comfortable with this if it was much more narrowly scoped?
[10:20:14] <sureshk> the host can use whatever in the lower 64 bits
[10:20:24] <sureshk> The whole /64 belongs to the host
[10:20:40] <Eliot Lear> Ted Hardie: flow state semantics only?
[10:21:12] Gorry Fairhurst leaves the room
[10:21:14] <Eliot Lear> Chair: if the scope were to support multipath and to be restricted to flow state semantics and nothing else, would that be agreeable?
[10:21:31] <Andrew Sullivan> if you think it's impossible, say no
[10:21:40] ddolson leaves the room
[10:21:46] <Eliot Lear> mnot:
[10:21:55] ted.h joins the room
[10:22:16] <Eliot Lear> by standardizing this we can make new things possible.  concerned about how this will be surfaced to users
[10:22:51] Sebastian Kiesel leaves the room
[10:22:53] <ted.h> Actually, I opposed them
[10:23:04] <Eliot Lear> EKR:
[10:23:07] craigt leaves the room: Disconnected: timeout during writing
[10:23:21] Carsten Bormann leaves the room
[10:23:45] <sftcd2> can the flow semantics be transport independent?
[10:23:56] <Lee Howard> Chair instructions to everybody else: please speak quickly.  To EKR: slooowly
[10:24:24] <Eliot Lear> aside from the middle box people, is anyone interested in transmitting these bits?
[10:25:14] <resnick> Did someone catch this person’s name?
[10:25:22] <sureshk> Georg Mayer
[10:25:26] <resnick> thx
[10:25:32] <sureshk> No probs Pete
[10:25:40] <Eliot Lear> (Georg Mayer from 3GPP): there are things that people can be afraid of but we are really in need of such mechanisms.  we now have 15 years of experience.  we should build on this to fulfill requirements but address people's fears and we need a starting point.
[10:26:02] <SzilveszterNadas> If middlebox vendors have no other chance, will not they do things like breaking encryption instead?
[10:26:21] <Eliot Lear> Ian ? Google:
[10:26:43] <Eliot Lear> Which parts are middlebox vendors excited about?
[10:26:44] <ted.h> Ian Swett
[10:27:09] <Eliot Lear> some of the ideas are unattainable
[10:27:14] Barry Leiba leaves the room
[10:27:20] <Eliot Lear> there's still a tls handshake
[10:27:40] <Eliot Lear> tom polly:
[10:27:42] Brian Trammell joins the room
[10:28:06] <Eliot Lear> We (Apple) there are some cases where we definitely want to do this, and we have concrete cases where we would definitely find this stuff useful
[10:28:23] <Eliot Lear> colin perkens
[10:28:55] <Eliot Lear> (garbled first part) and helping me find a transparent path
[10:29:04] ddolson joins the room
[10:29:16] <Eliot Lear> yoav:
[10:29:17] Fuyou Miao leaves the room
[10:29:22] Gabriel Montenegro_2331 leaves the room
[10:29:33] <Eliot Lear> what i would like to know: when does the flow start, flow end, and continue
[10:29:49] <Eliot Lear> when can i deallocate my state?
[10:29:49] ddolson leaves the room
[10:30:04] <Eliot Lear> concerned about extensibility
[10:30:21] Andrew Sullivan leaves the room
[10:30:22] <Eliot Lear> Chair: and now for a word from our AD
[10:30:26] <Eliot Lear> Spencer:
[10:30:30] Andrew Sullivan joins the room
[10:30:47] DanYork leaves the room
[10:30:58] Lee Howard leaves the room
[10:31:06] <Eliot Lear> Thanks for coming!
[10:31:06] frodek leaves the room
[10:31:07] Brian Trammell leaves the room
[10:31:08] aaron leaves the room
[10:31:08] Yoav Nir leaves the room
[10:31:09] ted.h leaves the room
[10:31:13] Kyle Rose leaves the room
[10:31:17] Eliot Lear leaves the room
[10:31:20] Simon Pietro Romano leaves the room
[10:31:22] <tale> ciao
[10:31:22] Suzanne leaves the room
[10:31:24] Yoshiro Yoneya leaves the room
[10:31:24] jsaldana leaves the room
[10:31:25] nygren leaves the room
[10:31:25] resnick leaves the room
[10:31:26] Toru Asahina leaves the room
[10:31:44] SzilveszterNadas leaves the room
[10:31:50] cheshire leaves the room
[10:31:51] magnus leaves the room
[10:31:57] Lorenzo Miniero leaves the room
[10:31:58] yuki goto leaves the room
[10:31:58] Catherine Truchan leaves the room
[10:31:58] John Leslie leaves the room
[10:31:58] Varun Singh leaves the room
[10:31:58] Simone Ferlin_3661 leaves the room
[10:31:58] George Tsolis leaves the room
[10:31:59] Emile Aben leaves the room
[10:32:09] wseltzer@jabber.org leaves the room
[10:32:17] Paul Cosgrove leaves the room
[10:32:27] Stephan Neuhaus leaves the room
[10:32:32] stere preda leaves the room
[10:32:38] Szilveszter Nadas leaves the room
[10:32:40] JeffH leaves the room
[10:32:43] hta leaves the room
[10:33:12] tale leaves the room: Disconnected: closed
[10:34:22] richsalz leaves the room
[10:34:57] Meetecho leaves the room
[10:38:30] mnot leaves the room
[10:38:39] hildjj leaves the room
[10:43:09] sftcd2 leaves the room
[10:43:09] sureshk leaves the room
[10:45:09] dkg leaves the room
[10:47:49] Andrew Sullivan leaves the room
[10:48:04] theresa leaves the room
[10:48:19] m&m leaves the room: Disconnected: closed
[10:53:12] Lee Howard joins the room
[10:57:40] jm leaves the room
[10:57:44] jm joins the room
[10:59:00] spreda leaves the room: Disconnected: closed
[10:59:06] Lee Howard leaves the room
[10:59:48] Lee Howard joins the room
[10:59:49] Lee Howard leaves the room
[11:07:05] jm leaves the room
[11:07:28] jm joins the room
[11:13:31] hildjj joins the room
[11:15:03] jm leaves the room
[11:15:07] jm joins the room
[11:19:39] jm leaves the room
[11:20:01] jm joins the room
[11:45:02] jm leaves the room
[11:45:14] jm joins the room
[11:47:36] hta joins the room
[11:48:58] Carsten Bormann joins the room
[11:50:13] jm leaves the room
[11:50:37] jm joins the room
[11:55:25] hta leaves the room
[11:56:39] frodek joins the room
[11:57:09] jm leaves the room
[11:57:09] hildjj leaves the room
[11:57:10] jm joins the room
[11:57:30] m&m joins the room
[11:57:36] jm leaves the room
[11:58:44] ted.h joins the room
[11:59:26] jm joins the room
[12:00:18] hta joins the room
[12:00:29] Lee Howard joins the room
[12:01:20] cheshire joins the room
[12:01:25] cheshire leaves the room
[12:01:46] jm leaves the room
[12:01:48] jm joins the room
[12:02:09] frodek leaves the room
[12:02:50] sureshk joins the room
[12:02:53] tenghardt joins the room
[12:03:01] sureshk leaves the room
[12:04:25] tenghardt leaves the room
[12:05:53] frodek joins the room
[12:09:38] sjero joins the room
[12:09:51] sjero leaves the room
[12:09:57] Kyle Rose joins the room
[12:10:57] resnick joins the room
[12:11:23] mnot joins the room
[12:12:04] aaron joins the room
[12:12:23] aaron leaves the room
[12:14:56] nygren joins the room
[12:15:26] cheshire joins the room
[12:19:20] jm leaves the room
[12:19:21] jm joins the room
[12:20:14] mnot leaves the room: Disconnected: Replaced by new connection
[12:20:16] nygren leaves the room
[12:20:18] mnot joins the room
[12:30:39] ted.h leaves the room
[12:32:33] Suzanne joins the room
[12:33:12] Suzanne leaves the room
[12:34:57] Kyle Rose leaves the room
[12:39:07] jm leaves the room
[12:41:12] jm joins the room
[12:43:46] mnot leaves the room
[12:44:29] m&m leaves the room: Disconnected: Replaced by new connection
[12:44:31] m&m joins the room
[12:44:43] m&m leaves the room
[12:45:44] tale joins the room
[12:47:50] frodek leaves the room
[12:48:15] jm leaves the room
[12:48:16] jm joins the room
[12:57:03] jm leaves the room
[12:57:04] jm joins the room
[13:00:08] jm leaves the room
[13:00:08] jm joins the room
[13:00:26] jm leaves the room
[13:00:27] jm joins the room
[13:01:12] jm leaves the room
[13:01:14] jm joins the room
[13:01:15] jm leaves the room
[13:01:19] jm joins the room
[13:19:35] Lee Howard joins the room
[13:20:08] Lee Howard leaves the room
[13:31:38] Lee Howard joins the room
[13:32:06] Lee Howard leaves the room
[13:37:16] ddolson joins the room
[13:38:05] Carsten Bormann leaves the room
[13:54:15] resnick leaves the room
[13:55:26] cheshire joins the room
[13:55:39] cheshire leaves the room
[13:58:29] hta leaves the room
[14:06:32] Lee Howard leaves the room
[14:09:22] cheshire leaves the room
[14:09:32] jm leaves the room
[14:09:33] jm joins the room
[14:10:09] Carsten Bormann joins the room
[14:12:23] jm leaves the room
[14:12:25] jm joins the room
[14:13:56] hildjj joins the room
[14:15:39] hildjj leaves the room
[14:19:56] ddolson leaves the room
[14:20:09] cheshire joins the room
[14:27:28] Lee Howard joins the room
[14:35:28] DanYork joins the room
[14:35:32] DanYork leaves the room
[14:39:29] ddolson joins the room
[14:41:23] spreda joins the room
[14:48:46] spreda leaves the room
[15:03:46] ddolson leaves the room
[15:08:00] ddolson joins the room
[15:33:30] ddolson leaves the room
[15:40:03] Lee Howard leaves the room
[15:40:39] jm leaves the room
[15:40:40] jm joins the room
[15:44:39] jm leaves the room
[15:44:39] jm joins the room
[16:12:44] ddolson joins the room
[16:20:59] cheshire leaves the room
[16:22:08] Carsten Bormann leaves the room
[16:25:49] tale leaves the room: Disconnected: closed
[16:32:47] tale joins the room
[16:32:52] ddolson leaves the room
[16:35:01] ddolson joins the room
[16:35:04] cheshire joins the room
[16:36:35] tale leaves the room
[17:05:48] hta joins the room
[17:06:37] Carsten Bormann joins the room
[17:15:42] Carsten Bormann leaves the room
[17:31:59] Carsten Bormann joins the room
[17:37:05] ddolson leaves the room
[17:43:18] cheshire leaves the room
[17:43:23] cheshire joins the room
[17:43:30] cheshire leaves the room
[17:43:32] jm leaves the room
[17:43:33] jm joins the room
[17:49:27] Carsten Bormann leaves the room
[17:49:32] jm leaves the room
[17:49:33] jm joins the room
[18:07:18] hta leaves the room
[18:18:24] jm leaves the room
[18:18:25] jm joins the room
[18:19:50] jm leaves the room
[18:19:52] jm joins the room
[18:30:42] jm leaves the room
[18:30:43] jm joins the room
[18:31:56] jm leaves the room
[18:31:58] jm joins the room
[18:38:34] jm leaves the room
[18:38:35] jm joins the room
[18:45:24] jm leaves the room
[18:47:23] jm joins the room
[18:55:52] jm leaves the room
[18:55:54] jm joins the room
[18:57:17] jm leaves the room
[18:57:19] jm joins the room
[18:57:50] jm leaves the room
[18:59:53] jm joins the room
[19:08:12] jm leaves the room
[19:08:13] jm joins the room
[19:15:22] jm leaves the room
[19:15:26] jm joins the room
[19:18:06] jm leaves the room
[19:18:08] jm joins the room
[19:19:58] jm leaves the room
[19:19:59] jm joins the room
[21:51:14] ddolson joins the room
[22:23:39] hta joins the room
[22:42:30] Lee Howard joins the room
[22:44:48] ddolson leaves the room
[22:49:04] ddolson joins the room
[22:49:20] ddolson joins the room
[22:49:20] ddolson leaves the room
[22:50:51] Lee Howard leaves the room
[23:09:10] hta leaves the room
[23:30:18] ddolson leaves the room
Powered by ejabberd - robust, scalable and extensible XMPP server Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!