IETF
tsvwg
tsvwg@jabber.ietf.org
Thursday, 8 November 2012< ^ >
becarpenter has set the subject to: TSVWG mtg at IETF 82
Room Configuration

GMT+0
[17:58:15] jpc joins the room
[17:59:19] jpc leaves the room
[20:02:28] Brian Carpenter joins the room
[20:04:24] erge joins the room
[20:15:27] gorryf joins the room
[20:16:15] frodek joins the room
[20:16:23] JoelHalpern joins the room
[20:18:50] nestor.tiglao joins the room
[20:19:17] rscheff joins the room
[20:19:49] <rscheff> agenda
[20:20:05] <gorryf> This is TSVWG ... if you are remote participant please tell us of audio problens
[20:20:34] wesley.m.eddy joins the room
[20:20:43] martin.stiemerling joins the room
[20:20:59] martin.stiemerling has set the subject to: TSVWG mtg at IETF 85
[20:21:54] <rscheff> diff serv slides
[20:22:19] jpc joins the room
[20:22:31] <rscheff> diff serv topics
[20:22:36] <gorryf> David Blakc - slideset 1a
[20:23:01] <rscheff> important diffserv rfcs
[20:23:36] <gorryf> Matt: It would be nice to see a roadmap of the DS specs
[20:23:44] <gorryf> James: there are at least 6 more
[20:24:42] <rscheff> slide 5 - diffserv history
[20:24:47] Eliot Lear joins the room
[20:26:19] <rscheff> slide 6 - topics why are we here
[20:27:40] <rscheff> slide 7 - diffserv interconnection problem statement
[20:30:15] <gorryf> Joel reviewing the liaison and what we should do next as a WG
[20:33:17] jpc leaves the room
[20:33:50] <rscheff> slide 7 - what should the ietf do
[20:34:44] <gorryf> Matt: There is a missing component ...
[20:35:14] <gorryf> David: The scope is limited to interconnect - this is not related to end systems.]
[20:35:24] <gorryf> Jmaes Polk:
[20:35:29] <rscheff> Updating DiffServ Service Class Guidelines
[20:35:52] <rscheff> http://www.ietf.org/proceedings/85/slides/slides-85-tsvwg-13.pptx
[20:36:48] jlcJohn joins the room
[20:37:41] <gorryf> Can people hear james?
[20:37:56] <jlcJohn> +
[20:39:05] <rscheff> slide 2 - who's affected by diffserve guidelines
[20:41:15] jpc joins the room
[20:43:20] <gorryf> Lars Eggert: I thought DS was clear - do you care more about latency or bandwidth? (You can only choose one)
[20:44:04] <Eliot Lear> who is the woman that is speaking?
[20:44:09] <gorryf> Mirja: I think it is a binary choice - you can queue to get both at the same time.
[20:44:22] <gorryf> Mirja - Chair of RMCAT
[20:44:25] <Eliot Lear> thanks
[20:44:27] <erge> Rüdiger: whats the difference betweeen two QoS classes under standard operational conditions?
[20:45:19] <gorryf> Mirja: Mappinga dont exactly fit apps
[20:46:06] <gorryf> To Rüdiger: - Can you explain your question and we will ask at the Mic...
[20:46:49] <erge> Rüdiger: QoS classes shouldn't be fitted to applications. Look at class properties at transport levele: congestion aware or not congestion aware.
[20:46:59] <gorryf> Lars: If it's low latency it goes in a quque that you don't buffer deeply
[20:48:33] <gorryf> James: I don't think there should be a hard coding. (David: We went into this in the wrong way, you just mark and condition).
[20:48:55] <rscheff> slide 3 - confusion about diffserv guidelines 1/2
[20:49:12] frodek leaves the room
[20:49:25] Magnus joins the room
[20:52:01] <rscheff> slide 4 - confusion about diffserv guidelines 2/2
[20:56:35] <gorryf> Matt: Do you distinguish RT and Streaming video (I don't believe for RT video)
[20:57:08] <gorryf> James: Telepresence is an example - you see pixelation.
[20:57:20] <gorryf> Matt: On the public Internet video is different
[20:57:59] <erge> Rüdiger: Agree to Matt's comment, my companies video/audio cracks also say, customer complaints increase significantly if lipsynch is lost.
[20:58:14] <Eliot Lear> i'm streaming right now and the delay is an issue for me. of course i suspect the real problem is the encoder, but i don't have the data
[20:59:27] <gorryf> Martin: the 1st part was about SC and these were confusing - we now talk about adding 2 classes...
[21:00:13] <rscheff> slide 5 - service class vs treatment aggregate
[21:01:51] <rscheff> slide 6 - purpose of the drafts, I
[21:04:07] <rscheff> slide 7 - purpose of the drafts, II
[21:06:10] <rscheff> slide 8 - particulars within RFC4594 update
[21:08:13] <gorryf> Lars: we can't extend this forever, at some point you need to deploy INTSERV...
[21:08:56] <gorryf> Matt: We should have put margin notes in the document to say what the values really mean
[21:11:35] <erge> Rüdiger: SP networks can deliver QoS based on DiffServ. The problem isn't transport in the network or a lack of classes. It's usually the edge, the service architecture and the application causing trouble.
[21:11:58] <gorryf> - will relay after bob
[21:12:37] <gorryf> Bob Briscoe: operators identify importance of services using DS
[21:13:01] <Eliot Lear> what does this have to do with the preso?
[21:13:04] <Eliot Lear> or the draft?
[21:13:04] frodek joins the room
[21:13:39] <gorryf> mirja: deciding service class is hard
[21:13:53] <Eliot Lear> sorry- long streaming delay
[21:19:20] martin.stiemerling leaves the room: Replaced by new connection
[21:19:22] martin.stiemerling joins the room
[21:20:37] frodek leaves the room
[21:20:39] <rscheff> slideset 1a - what should the ietf do
[21:23:12] <rscheff> slideset 2,3,4 - slide 19
[21:23:29] <gorryf> Paul Jones: I think that over time - there is value on guidance on using values
[21:24:17] <gorryf> There is a distinction between MM conferening and RT interactive
[21:26:24] <erge> Rüdiger: Is there experience with QoS transport offered by Service Providers showing it is not supporting any of the mentioned applications properly?
[21:26:28] <gorryf> Is this going to require new PHB or router changes?
[21:27:30] <gorryf> No - because this does not impact the carrier cores
[21:31:33] <gorryf> To Rüdiger-> I'm not sure this is the issue, since this isn't about treatment aggregate
[21:31:40] <jlcJohn> hum+
[21:32:50] <gorryf> Chairs identify there is at least one new edge marking that we should have done - i.e. multimedia traffic in 4594 needs updated,
[21:33:31] <gorryf> Pat: AV bridging in 802.1 has been a work item.
[21:35:51] <rscheff> slideset 1a, - slide 6
[21:39:21] <Eliot Lear> crickets
[21:39:31] <erge> Rüdiger: What are you out for, a class which offers low delay and low jitter for long packets and high loads? That's the AF41 Streaming class.
[21:40:30] <erge> Rüdiger: sorry, I'm a bit lost now - please formulate on the list.
[21:42:11] <erge> Rüdiger: DiffServ interconnection offers four classes. One could have 3 PHBs.
[21:42:37] <Eliot Lear> liaison can be found at https://datatracker.ietf.org/liaison/1165/
[21:44:18] <rscheff> http://www.ietf.org/proceedings/85/slides/slides-85-tsvwg-7.pdf
[21:44:31] <rscheff> Diffserv Recommendations for LLN class
of traffic
[21:44:34] <Eliot Lear> bye everyone.
[21:44:34] <rscheff> slide 2
[21:44:39] Eliot Lear leaves the room
[21:45:54] <rscheff> slide 3 - bierf on lln
[21:46:24] <rscheff> sldee 4 motivation
[21:47:29] <rscheff> slide 5 scope
[21:49:14] <rscheff> slide 6 lln traffic classes
[21:49:50] jpc leaves the room
[21:49:50] <rscheff> slide 7 proposed recommendation
[21:50:18] JoelHalpern leaves the room: offline
[21:51:15] <rscheff> slide 6
[21:51:19] <rscheff> slide 8 deployment scenarion (industrial automation)
[21:51:39] <rscheff> slide 7 proposed recommendation
[21:53:30] <rscheff> slide 8 deployment scenarion (industrial automation)
[21:55:41] Brian Carpenter leaves the room: offline
[21:57:18] nestor.tiglao leaves the room
[22:00:21] <rscheff> RSVP support for PCN
[22:00:59] <rscheff> http://www.ietf.org/proceedings/85/slides/slides-85-tsvwg-14.pptx
[22:01:06] <rscheff> 8) RSVP MULTI_INSTANCE Object
[22:01:32] <rscheff> slide 2 - scope of new ID
[22:04:19] <rscheff> slide 4 - comparing old with new proposal
[22:06:42] <rscheff> slide 5 - new option issue
[22:08:51] <rscheff> slide 4
[22:12:46] martin.stiemerling leaves the room
[22:12:55] gorryf leaves the room
[22:13:15] erge leaves the room
[22:13:24] wesley.m.eddy leaves the room
[22:13:53] rscheff leaves the room: Computer went to sleep
[22:17:01] Magnus leaves the room: Replaced by new connection
[22:17:01] Magnus joins the room
[22:18:06] Magnus leaves the room
[22:18:09] Magnus joins the room
[22:19:14] Magnus leaves the room: Replaced by new connection
[22:19:15] Magnus joins the room
[22:55:54] wesley.m.eddy joins the room
[23:33:29] wesley.m.eddy leaves the room
[23:47:10] Magnus leaves the room: Replaced by new connection
[23:47:10] Magnus joins the room
[23:48:33] Magnus leaves the room: Replaced by new connection
[23:48:34] Magnus joins the room
[23:55:11] Magnus leaves the room: Replaced by new connection
[23:55:12] Magnus joins the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!