[10:20:16] alex-meetecho joins the room
[10:36:21] spencerdawkins joins the room
[10:48:05] Meetecho joins the room
[10:50:00] Carsten Bormann joins the room
[10:50:01] Ekaterina Volodina joins the room
[10:50:01] Lucas Pardue joins the room
[10:50:01] Yue Yin joins the room
[10:50:01] Hirochika Asai joins the room
[10:50:01] Spencer Dawkins joins the room
[10:50:01] Tobia Castaldi joins the room
[10:50:01] Dirk Kutscher joins the room
[10:50:01] Markus Amend joins the room
[10:50:19] Li Yizhou joins the room
[10:50:36] Lucas Pardue leaves the room
[10:50:53] Alessandro Toppi joins the room
[10:51:26] cabo joins the room
[10:51:32] Ekaterina Volodina leaves the room
[10:51:36] Ekaterina Volodina joins the room
[10:52:11] Dirk Kutscher leaves the room
[10:52:13] Dirk Kutscher joins the room
[10:52:24] cabo has set the subject to: https://codimd.ietf.org/notes-ietf-108-loops
[10:52:45] Frode Kileng joins the room
[10:53:03] Richard Wilhelm joins the room
[10:53:04] Xingwang Zhou joins the room
[10:53:08] Rohit Abhishek joins the room
[10:53:13] Martin Duke joins the room
[10:53:27] frodek joins the room
[10:53:37] Jianglong joins the room
[10:53:38] Richard Wilhelm leaves the room
[10:53:42] Erfan Mozaffariahrar joins the room
[10:54:04] Julien Maisonneuve joins the room
[10:54:43] Dirk Kutscher leaves the room
[10:54:55] Dirk Kutscher joins the room
[10:55:12] Ekaterina Volodina leaves the room
[10:55:12] Ekaterina Volodina joins the room
[10:55:49] Magnus Westerlund joins the room
[10:55:52] Lucas Pardue joins the room
[10:56:00] Akbar Rahman joins the room
[10:56:08] Erfan Mozaffariahrar leaves the room
[10:56:09] Rick Taylor joins the room
[10:56:14] Brian Trammell joins the room
[10:56:27] Tommy Pauly joins the room
[10:56:43] David Black joins the room
[10:56:57] Ekaterina Volodina leaves the room
[10:56:59] Magnus Westerlund too joins the room
[10:57:25] Marie-Jose Montpetit joins the room
[10:57:28] Ekaterina Volodina joins the room
[10:57:37] Cullen Jennings joins the room
[10:57:46] Kannan Jayaraman joins the room
[10:57:52] Cullen Jennings_401 joins the room
[10:58:06] Ekaterina Volodina leaves the room
[10:58:08] Gorry Fairhurst joins the room
[10:58:08] David Black leaves the room
[10:58:08] Carlos Bernardos joins the room
[10:58:15] Shuai Zhao joins the room
[10:58:20] Shogo Asaba joins the room
[10:58:22] Gorry Fairhurst leaves the room
[10:58:23] Ekaterina Volodina joins the room
[10:58:25] David Black joins the room
[10:58:27] Gorry Fairhurst joins the room
[10:58:34] Colin Perkins joins the room
[10:58:56] Nicolas Kuhn joins the room
[10:59:00] Alexandra Kelly joins the room
[10:59:05] Praveen Mada joins the room
[10:59:15] Mirja Kühlewind joins the room
[10:59:23] Matt Green joins the room
[10:59:30] Georgios Karagiannis joins the room
[10:59:47] David Oran joins the room
[10:59:47] Patrick McManus joins the room
[10:59:52] Shogo Asaba leaves the room
[10:59:55] Adam Wiethuechter joins the room
[10:59:56] Chi-Jiun Su joins the room
[11:00:06] Martin Thomson joins the room
[11:00:37] Hannu Flinck joins the room
[11:00:38] Gunter Van de Velde joins the room
[11:00:41] Nicolas Kuhn leaves the room
[11:00:46] Wim Henderickx joins the room
[11:00:48] Nicolas Kuhn joins the room
[11:00:48] Joerg Ott joins the room
[11:00:52] Jorge Rabadan joins the room
[11:00:54] Fengwei Qin joins the room
[11:01:00] Philip Eardley joins the room
[11:01:06] DaveO joins the room
[11:01:09] Chonggang Wang joins the room
[11:01:14] Nicolas Kuhn leaves the room
[11:01:28] Nicolas Kuhn joins the room
[11:01:30] Mirja Kühlewind leaves the room
[11:01:36] Mirja Kühlewind joins the room
[11:01:52] Maksim Proshin joins the room
[11:01:53] Laura Al Wardani joins the room
[11:02:05] Mirja joins the room
[11:02:16] avezza joins the room
[11:02:17] Dirk Hugo joins the room
[11:02:32] Adam Wiethuechter leaves the room
[11:02:34] Joseph Lewis joins the room
[11:02:36] Adam Wiethuechter joins the room
[11:02:43] Stuart Card joins the room
[11:02:53] Jordi Paillissé Vilanova joins the room
[11:02:58] Ronald in 't Velt joins the room
[11:03:00] Qin Wu joins the room
[11:03:06] Greg White joins the room
[11:04:05] Lixia Zhang joins the room
[11:05:20] Joerg Ott leaves the room
[11:05:30] Joerg Ott joins the room
[11:05:42] chi.jiun.su joins the room
[11:06:12] Fengwei Qin leaves the room
[11:06:18] Gorry Fairhurst leaves the room
[11:06:19] Nicolas Kuhn leaves the room
[11:06:22] Yue Yin leaves the room
[11:06:23] Gorry Fairhurst joins the room
[11:06:35] Xuesong Geng joins the room
[11:07:06] Fengwei Qin joins the room
[11:07:08] Nicolas Kuhn joins the room
[11:07:14] Yue Yin joins the room
[11:07:15] Adam Wiethuechter leaves the room
[11:07:17] Adam Wiethuechter joins the room
[11:07:20] <Joerg Ott> are the correct slides being shared or is it just me?
[11:07:27] Erfan Mozaffariahrar joins the room
[11:07:34] <Martin Duke> it's just you
[11:07:38] Nicolas Kuhn leaves the room
[11:07:38] Nicolas Kuhn joins the room
[11:07:45] <Joerg Ott> weird...
[11:07:50] <Martin Thomson> I'm looking at -02: https://www.ietf.org/proceedings/108/slides/slides-108-loops-combined-chairproponent-slides-02
[11:07:52] Fengwei Qin leaves the room
[11:07:53] Nicolas Kuhn leaves the room
[11:08:00] Yue Yin leaves the room
[11:08:04] Joerg Ott leaves the room
[11:08:04] Jonathan Lennox joins the room
[11:08:11] Zhuangyan joins the room
[11:08:13] Joerg Ott joins the room
[11:08:13] daniel.bernier@bell.ca joins the room
[11:08:15] <Tommy Pauly> Slides look fine
[11:08:15] <Martin Thomson> I can't tell if that matches what is displayed here, because it's broken.
[11:08:29] Nicolas Kuhn joins the room
[11:08:31] Gorry Fairhurst leaves the room
[11:08:32] <Brian Trammell> it seems to match what's in -02
[11:08:33] Gorry Fairhurst joins the room
[11:08:44] Nicolas Kuhn leaves the room
[11:08:44] <Martin Thomson> Sorry, meetecho is broken.
[11:08:47] Fengwei Qin joins the room
[11:08:47] <Martin Thomson> Not the slides.
[11:08:50] Nicolas Kuhn joins the room
[11:08:55] <Joerg Ott> now I see them
[11:08:57] <Tommy Pauly> Carry on =)
[11:08:58] <Markus Amend> Everyting is fine
[11:08:59] Yue Yin joins the room
[11:09:00] Ekaterina Volodina leaves the room
[11:09:03] Ekaterina Volodina joins the room
[11:09:13] <Brian Trammell> ah; carsten is on the slide labelled 10
[11:09:14] Yue Yin leaves the room
[11:09:25] Yali Wang joins the room
[11:09:28] <Brian Trammell> (I'll put slide numbers here for people who have lost meetecho sync)
[11:09:28] <Brian Trammell> 11
[11:09:29] Nicolas Kuhn leaves the room
[11:09:41] <Brian Trammell> 12
[11:09:49] Ekaterina Volodina leaves the room
[11:09:49] Nicolas Kuhn joins the room
[11:09:51] Ekaterina Volodina joins the room
[11:09:55] Yue Yin joins the room
[11:10:06] Ekaterina Volodina leaves the room
[11:10:14] Nicolas Kuhn leaves the room
[11:10:19] Reji Thomas joins the room
[11:10:20] Nicolas Kuhn joins the room
[11:10:34] Ekaterina Volodina joins the room
[11:10:37] Nicolas Kuhn leaves the room
[11:10:39] Nicolas Kuhn joins the room
[11:10:50] Joseph Lewis leaves the room
[11:10:54] Craig Taylor joins the room
[11:11:06] jmaisonn@jabber.uk joins the room
[11:11:10] <Brian Trammell> 13
[11:11:37] <Brian Trammell> 14
[11:11:38] Stephan Emile joins the room
[11:11:54] Yali Wang leaves the room
[11:12:09] Reji Thomas leaves the room
[11:12:18] Christine McCarthy joins the room
[11:12:27] Lars Eggert joins the room
[11:12:56] <Brian Trammell> 15
[11:13:14] Nicolas Kuhn leaves the room
[11:13:21] Nicolas Kuhn joins the room
[11:13:24] Diego Lopez joins the room
[11:13:27] <Brian Trammell> 16
[11:13:39] <Brian Trammell> is there anyone who still can't see the meetecho slides?
[11:13:54] Nicolas Kuhn leaves the room
[11:13:55] Nicolas Kuhn joins the room
[11:13:55] <Joerg Ott> yes, works now for me — reload helped
[11:14:09] Nicolas Kuhn leaves the room
[11:14:15] Nicolas Kuhn joins the room
[11:14:27] Vincent Roca joins the room
[11:14:32] <Brian Trammell> 17
[11:14:32] Nicolas Kuhn leaves the room
[11:14:33] Nicolas Kuhn joins the room
[11:14:36] Jianglong leaves the room
[11:14:40] Jianglong joins the room
[11:14:43] Nicolas Kuhn leaves the room
[11:15:01] Joseph Lewis joins the room
[11:15:18] <Brian Trammell> 18
[11:15:19] Yolanda Yu joins the room
[11:15:27] Jorge Rabadan leaves the room
[11:15:29] Nicolas Kuhn joins the room
[11:15:30] Peng Liu joins the room
[11:15:32] Luis Contreras joins the room
[11:15:37] Jorge Rabadan joins the room
[11:16:00] Nicolas Kuhn leaves the room
[11:16:14] Pete Resnick joins the room
[11:16:21] Brian joins the room
[11:16:28] Lixia Zhang leaves the room
[11:17:23] Yue Yin leaves the room
[11:17:26] Pete Resnick (the other one) joins the room
[11:17:29] Yolanda Yu leaves the room
[11:17:29] Yolanda Yu joins the room
[11:17:32] Nicolas Kuhn joins the room
[11:17:35] Yue Yin joins the room
[11:17:50] Nicolas Kuhn leaves the room
[11:18:14] Nicolas Kuhn joins the room
[11:18:30] <Brian Trammell> ...21
[11:18:38] <Brian Trammell> and props for the graphic
[11:18:45] <Martin Thomson> meetecho sync restored
[11:19:19] Bhavit Shah joins the room
[11:19:54] Niket Agrawal joins the room
[11:19:58] Jana Iyengar joins the room
[11:20:11] Andre Bondi joins the room
[11:20:11] <Colin Perkins> ECN is presumably traditional ECT(0) marking, right?
[11:20:26] Xuesong Geng leaves the room
[11:20:31] Gorry Fairhurst leaves the room
[11:20:35] Gorry Fairhurst joins the room
[11:20:36] <spencerdawkins> Colin, I believe so.
[11:20:40] Xuesong Geng joins the room
[11:20:42] Joseph Lewis leaves the room
[11:21:03] ZHAOQIANYING joins the room
[11:21:19] <Martin Duke> I would think ECT(1) low-latency traffic would be a poor fit for anything but FEC, anyway
[11:21:49] <David Black> Is there a dependency on the TSVWG tunnel congestion feedback draft here?
[11:21:50] Nicolas Kuhn leaves the room
[11:21:52] Peng Liu leaves the room
[11:21:54] Nicolas Kuhn joins the room
[11:22:08] <Gorry Fairhurst> That draft has really stalled though...
[11:22:16] <Brian Trammell> yes but this is a good point - LOOPS (if chartered) should interop with L4S reasonably
[11:22:31] Nicolas Kuhn leaves the room
[11:22:40] <Martin Thomson> carsten did basically just say "L4S later", which seems reasonable
[11:22:41] Gorry Fairhurst leaves the room
[11:22:45] Nicolas Kuhn joins the room
[11:22:46] <Brian Trammell> yep
[11:22:50] Gorry Fairhurst joins the room
[11:22:56] <Gorry Fairhurst> The audio is beraking up for me really badly...
[11:23:04] Peng Liu joins the room
[11:23:07] Peng Liu leaves the room
[11:23:07] Peng Liu joins the room
[11:23:10] <Magnus Westerlund too> Audio sounds fine for me.
[11:23:13] <Joerg Ott> ok here
[11:23:18] Jana Iyengar leaves the room
[11:23:20] <Brian Trammell> audio ok here too
[11:23:21] <Gorry Fairhurst> :-)
[11:23:22] <David Black> +1 on focusing on what currently works/is deployed
[11:23:24] <Martin Thomson> audio has always been solid
[11:23:26] <Meetecho> Gorry Fairhurst: there's a button to reconnect audio (circling arrows, low/right corner)
[11:23:31] Craig Taylor leaves the room
[11:23:37] Craig Taylor joins the room
[11:24:31] <Gorry Fairhurst> Meetecho: that worked!
[11:24:36] Vincent Roca leaves the room
[11:24:49] <spencerdawkins> I think turning off video helps, too ...
[11:25:42] Shwetha Bhandari joins the room
[11:25:58] Gang Yan joins the room
[11:26:27] Steffen Lindner joins the room
[11:27:03] <Martin Thomson> maybe you could try spinning around three time anticlockwise
[11:27:04] Vincent Roca joins the room
[11:27:25] Jana Iyengar joins the room
[11:28:07] <spencerdawkins> At 264 second RTT, you should be running DTN anyway :-)
[11:28:37] Lars Eggert leaves the room
[11:28:44] Lars Eggert joins the room
[11:28:46] Gorry Fairhurst leaves the room
[11:29:07] Chunshan Xiong joins the room
[11:29:08] Gorry Fairhurst joins the room
[11:29:18] Kannan Jayaraman leaves the room
[11:29:25] Joerg Ott leaves the room
[11:29:34] Gorry Fairhurst leaves the room
[11:29:40] Gorry Fairhurst joins the room
[11:29:42] <Ronald in 't Velt> CBOR encoded :)
[11:30:14] Lixia Zhang joins the room
[11:30:23] Wim Henderickx leaves the room
[11:30:34] Wim Henderickx joins the room
[11:30:39] Gorry Fairhurst leaves the room
[11:30:40] Lars Eggert leaves the room
[11:30:49] Lars Eggert joins the room
[11:30:52] Gorry Fairhurst joins the room
[11:30:57] Peng Liu leaves the room
[11:31:04] Lars Eggert leaves the room
[11:31:10] <Martin Thomson> whoa: is this a routing protocol?
[11:31:10] Lars Eggert joins the room
[11:31:18] <cabo> No.
[11:31:18] Gorry Fairhurst leaves the room
[11:31:23] Gorry Fairhurst joins the room
[11:31:27] <cabo> It is used in an application that uses a roi
[11:31:29] Peng Liu joins the room
[11:31:30] <Martin Thomson> it sounds like we're talking about path selection
[11:31:32] <cabo> Routing protocol.
[11:31:40] Peng Liu leaves the room
[11:31:40] Peng Liu joins the room
[11:31:51] Nabil Bitar joins the room
[11:32:00] <cabo> Huawei is dong path selection, but the LOOPS aspect is not.
[11:32:03] <Martin Thomson> OK, this is describing an application.
[11:32:05] <Rick Taylor> Is it just me, but how does LOOPS differ from DetNet?
[11:32:11] <DaveO> Probably ought to be specific that tunnel selection by an ingress is out of scope. Ditto for ensuring the traffic gets routed to the right ingress
[11:32:21] <cabo> Good question.  Should have made some slides.
[11:32:33] Gorry Fairhurst leaves the room
[11:32:36] Gorry Fairhurst joins the room
[11:32:55] <David Black> @Rick - DetNet prevents losses, LOOPS reacts/recovers for the most part
[11:33:15] <cabo> Detnet is for more nailed down traffic that can avoid losses
[11:33:22] <cabo> LOOPS is for more general traffic
[11:33:26] Craig Taylor leaves the room
[11:33:32] Craig Taylor joins the room
[11:33:55] <Martin Thomson> so the ingress observes that traffic is going past an egress (using $methods) and associates those packets with the control session it has with that egress?
[11:34:06] <cabo> DaveO: Noted.
[11:34:16] <Martin Thomson> or is data being routed explicitly into a tunnel?
[11:34:29] <cabo> The SDN controller might do that
[11:34:41] <cabo> E.g., traffic from customer A goes there, from customer B doesn't.
[11:34:45] Fengwei Qin leaves the room
[11:34:45] Fengwei Qin joins the room
[11:35:02] spencerdawkins leaves the room: Disconnected: closed
[11:35:02] Spencer Dawkins leaves the room
[11:35:05] Spencer Dawkins joins the room
[11:35:22] Fengwei Qin leaves the room
[11:35:23] Fengwei Qin joins the room
[11:35:31] spencerdawkins joins the room
[11:35:34] Gorry Fairhurst leaves the room
[11:35:37] Gorry Fairhurst joins the room
[11:35:53] <DaveO> Another query: are these supposed to be instrumented with tunnel traceroute, or are they opaque to endpoints?
[11:36:06] Spencer Dawkins leaves the room
[11:36:10] <Jana Iyengar> I thought the use case showed routing, but that is not part of LOOPS.... but he is saying that "CPE1 enables LOOPS", which suggests routing is part of it
[11:36:15] Spencer Dawkins joins the room
[11:36:20] Fengwei Qin leaves the room
[11:36:21] <Jana Iyengar> Martin: +1
[11:36:25] Yolanda Yu leaves the room
[11:36:36] <cabo> CPE can include path selection as well as LOOPS
[11:36:38] Fengwei Qin joins the room
[11:36:53] <Jana Iyengar> Why is CPE1 even part of LOOPS?
[11:37:00] Yolanda Yu joins the room
[11:37:02] Yolanda Yu leaves the room
[11:37:08] <Jana Iyengar> Shouldn't that be just on the boxes in the middle?
[11:37:11] <Gorry Fairhurst> Is there a classifier somehow in the network path?
[11:37:13] <cabo> It might be the ingress
[11:37:18] Yolanda Yu joins the room
[11:37:26] <cabo> Maybe?  All orthogonal to LOOPS
[11:37:35] Gorry Fairhurst leaves the room
[11:37:37] Gorry Fairhurst joins the room
[11:37:44] Yolanda Yu leaves the room
[11:37:45] <Jana Iyengar> Gorry: +1
[11:37:52] Chunshan Xiong leaves the room
[11:37:56] Yolanda Yu joins the room
[11:38:02] Chunshan Xiong joins the room
[11:38:10] Oleg Pekar joins the room
[11:38:11] <Martin Thomson> so there is a model I can see where a router selects a route for which it knows of an egress point, but it doesn't sound like this is what is happening
[11:38:11] Yolanda Yu leaves the room
[11:38:28] Yolanda Yu joins the room
[11:38:43] <DaveO> (the above might be more a Geneve question than specific to LOOPS)
[11:38:51] Jianglong leaves the room
[11:38:58] Jianglong joins the room
[11:38:58] <David Black> Think path selection is orthogonal to LOOPS - use LOOPS on less-than-wonderful path after it's selected.
[11:39:02] <Martin Thomson> that is, the router - after having selected a route - determines whether that route has an egress node present that it can collaborate with - and so applies loops
[11:39:34] ZHAOQIANYING leaves the room
[11:39:38] <Martin Thomson> David Black: I had assumed from the introduction that route selection is orthogonal, but I'm starting to wonder how
[11:39:56] ZHAOQIANYING joins the room
[11:40:00] <Rick Taylor> @David Black - Re DetNet - I may misunderstand, but the difference between LOOPS and DetNet is the definition of 'Reliability' or 'Determinism'  - Both WGs are concerned with using FEC and duplication to increase Reliability, traditionally 5 9's, but it could be 2 9's for these use-cases
[11:40:02] <Colin Perkins> I had thought that LOOPS was just a fancy tunnel, that recovers some loss. The use cases seem more general. Are they conflating LOOPS and services that use LOOPS, or is LOOPS more general?
[11:40:04] <Brian Trammell> my understanding is that loops is what you do when you can't select a better path
[11:40:09] <DaveO> The above of course assumes you have access to enough routing information to see that far down the path…
[11:40:18] <Martin Thomson> the introduction implied that this is about multiple hops in the network, but you can't be sure where things will end up if there are multiple hops
[11:40:26] Jianglong leaves the room
[11:40:34] Jianglong joins the room
[11:40:38] <Brian Trammell> (and you figure out it's not better through measurement
[11:40:45] jmagallanes leaves the room
[11:40:45] jmagallanes joins the room
[11:40:48] Yolanda Yu leaves the room
[11:40:54] Yolanda Yu joins the room
[11:40:58] Oleg Pekar leaves the room
[11:41:09] Yolanda Yu leaves the room
[11:41:09] <Gorry Fairhurst> "you can't select" means the router on the path uses a heuristic to decide whether this is needed?
[11:41:14] Niket Agrawal leaves the room
[11:41:27] <Brian Trammell> that's my assumption
[11:41:37] <Brian Trammell> where the heuristic might be "i was configured to"
[11:41:43] <Jana Iyengar> So, can I assume that classification is also not part of loops?
[11:41:44] Yolanda Yu joins the room
[11:41:52] <Martin Thomson> but this depends on route selection?
[11:41:54] <David Black> @Rick - significant difference is pre-reserved capacity for DetNet.  FEC is not fundamental to DetNet.
[11:42:07] <Martin Thomson> I am selecting a route that involves a tunnel to the egress node?
[11:42:21] <Jana Iyengar> Right, so question: "How do you *know* that the egress will be on path"?
[11:42:27] <DaveO> And incidently, the interaction between FEC and multipath could be "interesting" as LOOPs could be used a a back up FEC path.
[11:42:43] <Dirk Kutscher> Would be great to see a concise description of what exactly should be standardized and what not.
[11:43:01] <DaveO> (you send some packets on the direct path and FEC pckets on the LOOPS tunnel)
[11:43:01] <Jana Iyengar> And what the assumptions are about the context.
[11:43:23] <Martin Thomson> I got the impression that CE marked packets would stay CE marked, but yes, that is interesting
[11:44:13] <Martin Thomson> gorry, you are cutting out a little
[11:44:16] <Rick Taylor> @David Black - I think DetNet are looking to address problems beyond pre-reserved paths.  I'm not suggesting that DetNet takes on this work, but I do think there should be cross-polination
[11:44:25] <Lucas Pardue> gorry needs LOOPS
[11:44:36] <Martin Thomson> what you get is a whole lot of reordering
[11:44:52] ZHAOQIANYING leaves the room
[11:44:55] ZHAOQIANYING joins the room
[11:45:27] <Martin Thomson> circuit breaker is in the draft, I seem to recall
[11:45:36] <cabo> And in the charter
[11:46:10] Huaru Yang joins the room
[11:46:14] Gorry Fairhurst leaves the room
[11:46:18] Gorry Fairhurst joins the room
[11:46:25] Yolanda Yu leaves the room
[11:46:30] <Colin Perkins> thks
[11:46:37] Vincent Roca leaves the room
[11:46:40] Vincent Roca joins the room
[11:48:29] <Colin Perkins> It presumably either adds reordering or latency
[11:48:31] <Martin Duke> IIUC this is a not a tunneling design; it is an add-on to tunnels
[11:48:43] <Martin Thomson> the question then becomes something I can't answer: how do you know that the application in use can tolerate that degree of reordering?
[11:49:04] <David Black> Martin's clarification indicates that ECN has to be in use, so TSVWG tunnel congestion feedback draft is not related, as that uses ECN in outer header for traffic that does not use ECN (inner header is not-ECT).
[11:49:09] <Mirja Kühlewind> if it doesn't it shouldn't be worse than loss...
[11:49:36] <Gorry Fairhurst> I think it relies on reordering to prevent PTO/RACK retx "too quickly" for loops.
[11:49:42] <Jana Iyengar> Martin: Exactly
[11:50:05] Andre Bondi leaves the room
[11:50:29] <Mirja Kühlewind> reordering should not have a worse effect than loss
[11:50:31] <Patrick McManus> I'm just a bystander but istm that mostt of the work that has been done here has been on applying loops to geneve tunnels.. is that right?
[11:50:39] <Colin Perkins> Possibly one for the charter discussion, but an informational draft that explains the envelope where this helps and where it hurts might be useful
[11:50:39] Jianglong leaves the room
[11:50:40] ZHAOQIANYING leaves the room
[11:50:40] <Martin Duke> @MT you end up with a spuriuous retrans
[11:50:41] Diego Lopez leaves the room
[11:50:45] Diego Lopez joins the room
[11:50:54] ZHAOQIANYING joins the room
[11:51:08] <Martin Duke> which means LOOPS adds a little downstream load, but the receiver gets the data faster
[11:51:08] <Martin Thomson> so if this segment is ahead of a bandwidth-constrained segment, and this displacement effect occurs, that adds latency
[11:51:15] <cabo> Martin Duke: Yes
[11:51:21] Jianglong joins the room
[11:51:24] <DaveO> Another FEC question - is the intent that DEC recovery be done at each LOOPs hop, or is the more general case of concatenated LOOPs tunnels covered?
[11:51:31] Diego Lopez leaves the room
[11:51:32] <Martin Duke> and of course FEC should be less unordered
[11:51:33] Diego Lopez joins the room
[11:51:35] <DaveO> DEC->FEC
[11:51:42] <Martin Thomson> this is making an assumption that all data eventually is the goal of all applications
[11:52:38] <Martin Thomson> FEC on the loop doesn't have this problem of course, but what I've heard is that generic FEC is hard to justify
[11:52:49] <cabo> DaveO: LOOPS hops don't talk with each other at the moment
[11:52:52] <Gorry Fairhurst> MT: I think it is making some treatment by default.
[11:53:04] <Martin Thomson> Concatenated loops is a great question
[11:54:40] <Martin Thomson> that's a question I think that I can answer: wifi repair and resequencing is not great, but as the additional delay is minimal it rarely contributes significantly to performance degradation
[11:55:01] <Martin Duke> Maybe ECT(1) should signal that the application wants LOOPS :-)
[11:55:06] <Martin Thomson> do that over a longer span (like 50ms) and you will start to pay costs
[11:55:26] <Mirja Kühlewind> @Martin duke: no
[11:55:35] <DaveO> cabO: ok, that leaves some of the biggest benefits of FEC off the table, but just note so in the scope (ditto for the multipath quesiton - if you have parallel LOOPs tunnels you don't deal with splitting FEC across them or striping between the tunnel and the direct path
[11:55:35] Gorry Fairhurst leaves the room
[11:55:37] <Lixia Zhang> I feel so stupid: ACK/NACK/retrans. requires unique packet ID, what is used for that?
[11:55:41] <David Black> Huawei Hong Kong - Shenzhen example seems relevant to Jana's Q
[11:55:47] Erfan Mozaffariahrar leaves the room
[11:55:58] Gorry Fairhurst joins the room
[11:56:10] <DaveO> @Lixia: Geneve tunnel header
[11:56:13] Diego Lopez leaves the room
[11:56:14] Diego Lopez joins the room
[11:56:17] <spencerdawkins> I THINK the point is that these are managed paths between routers, not end-to-end, so managers won't do stupid things.
[11:56:30] Diego Lopez leaves the room
[11:56:33] Diego Lopez joins the room
[11:56:51] ZHAOQIANYING leaves the room
[11:56:55] ZHAOQIANYING joins the room
[11:57:19] Xuesong Geng leaves the room
[11:57:23] Kazuho Oku joins the room
[11:57:53] <Gorry Fairhurst> MT: If you have WiFi doing retx...then the WiFi link has made decisions on when/how to retx; if LOOPS wotks over that, it effectively changes that design ... without understanding the L2.
[11:58:04] ZHAOQIANYING leaves the room
[11:58:18] ZHAOQIANYING joins the room
[11:58:27] <DaveO> @spencer, then do you think a measurement protocol is (a)in scope, and/or (b) mandatory for compliance?
[11:58:32] Xuesong Geng joins the room
[11:58:35] <Martin Thomson> gorry: this is my point, but I was saying that WiFi gets away with some abuses because it does so over very short distances, whereas LOOPS will not
[11:58:43] <Lixia Zhang> @DaveO: thanks. That seems answering Jana question: given it's config'ed tunnel, it *is* a "link" :-)
[11:59:11] Xuesong Geng leaves the room
[11:59:14] Rick Taylor leaves the room
[11:59:16] Xuesong Geng joins the room
[11:59:18] <Martin Duke> jana's question about when it helps or hurts should probably be part of the charter
[11:59:25] <Gorry Fairhurst> MT: agree.
[11:59:38] <cabo> Spencer Dawkins: This.
[11:59:50] Gorry Fairhurst leaves the room
[11:59:53] Gorry Fairhurst joins the room
[12:00:03] <Jana Iyengar> spencerdawkins: That's a strong assumption
[12:00:05] <Martin Thomson> It seems to me that a router that has to decide between a LOOPS tunnel to A or a LOOPS tunnel to B (beyond A) is inherently causing LOOPS to be involved in routing decisions.  Just as choosing between a LOOPS tunnel and a non-LOOPS tunnel based on ECT(0) markings on incoming packets.
[12:00:16] Jonathan Lennox leaves the room
[12:00:18] <DaveO> @Lixia, yes, LOOPS seems to be just a congestion/loss mitigation scheme for Geneve tunnels. That's(luckily) a pretty narrow scope.
[12:00:21] <Lixia Zhang> But the questions on interactions apply (that even happened to true link layers as we saw with multihop wireless...)
[12:00:26] <cabo> Martin Duke: So that would be some discussion that goes into the document?
[12:00:33] <Martin Thomson> So I guess I'm challenging the assertion that LOOPS is orthogonal to routing choices.
[12:01:05] <Martin Duke> @cabo I would encourage LOOPS products to specifically address interactions with loss detection and latency measurement at the endpoints
[12:01:10] Mirja leaves the room
[12:01:16] <David Black> @Martin - but does LOOPS have to solve routing problem, or merely present improved links for routers to deal with?a
[12:01:38] <Nicolas Kuhn> should loops consider how fair it is when it shares capacity with non loops connections?
[12:01:41] <cabo> MT: Who asserted that?  LOOPS doesn't have to know about routing choices. Routing choices may want to
[12:01:51] <Martin Thomson> David Black: I see your point.  That's not entirely unreasonable.
[12:01:58] ZHAOQIANYING leaves the room
[12:02:00] <Lucas Pardue> i thought a routing loop was a bad thing ;)
[12:02:07] Maksim Proshin leaves the room
[12:02:15] <Martin Duke> @david black IMO LOOPS is something that happens after overlay routing decisions are made
[12:02:32] ZHAOQIANYING joins the room
[12:02:34] ZHAOQIANYING leaves the room
[12:02:34] ZHAOQIANYING joins the room
[12:02:41] <Martin Thomson> In other words, if this is viewed as providing new types of links that can be selected, then that might be OK.
[12:02:42] <David Black> +1 - matches my view of how this happens.
[12:02:44] <spencerdawkins> @Lucas - groan. We're going to hear more LOOPS jokes than QUIC jokes ...
[12:02:55] <Gorry Fairhurst> NK: It needs to consider the sharing with non-LOPS flows, which is also trciky since LOOPS may be high rate aggregates!
[12:03:06] <Martin Thomson> I wonder how a flow that only marks some packets ECT(0) will end up doing here.
[12:03:16] <Lucas Pardue> yes, lots of retransmittion of loops jokes
[12:03:27] <David Black> @MartinT - Dr. it hurts when I do this ...
[12:03:35] Tom Jones joins the room
[12:03:45] <Gorry Fairhurst> MT: The drafts says Non-ECT is simply dopped in some way when needed.
[12:03:46] <spencerdawkins> @David - exactly.
[12:04:05] <Martin Thomson> David Black: this happens
[12:04:06] Reji Thomas joins the room
[12:04:29] <spencerdawkins> To be fair, the proponents have been doing more updates to the charter than to the drafts. We should probably focus there.
[12:04:38] <Martin Thomson> Gorry: I need to read that text...
[12:04:48] <Jana Iyengar> Can someone please find MP?
[12:05:08] <Stephan Emile> Sorry I make a mistake
[12:05:25] <Martin Thomson> MP_NOT_FOUND!
[12:05:39] <Martin Thomson> EVERYONE gets MP_NOT_FOUND
[12:05:43] Stephan Emile leaves the room
[12:05:45] <Jana Iyengar> Oh, so that's what MP is!
[12:05:46] <Lixia Zhang> it seems to me no study has been done regarding potential interaction with higher layers?
[12:05:49] <Jana Iyengar> Microphone
[12:05:52] <Brian Trammell> ahhh
[12:05:58] <Jana Iyengar> Lixia: Exactly
[12:06:01] jmaisonn@jabber.uk leaves the room
[12:06:04] Stephan Emile joins the room
[12:06:06] <Brian Trammell> i thought meetecho was making a comment about multipath in the charter
[12:06:10] <Colin Perkins> Huh - I've been wondering about that all week :)
[12:06:19] <Stuart Card> There has been _discussion_ of cross-layers interactions, I don't know about _study_.
[12:06:19] <Jana Iyengar> Yeah, wow. Thanks, Carsten!
[12:06:27] <Brian Trammell> multipath? once you say multipath, the bof is over
[12:06:31] <Adam Wiethuechter> Don't feel back Jana, it took me a bit to understand what MP meant too....its way to early in the AM for this
[12:06:57] <Lucas Pardue> "letter not loaded in any tray"
[12:07:15] <Pete Resnick (the other one)> @Lixia / @Jana : Are you suggesting an Experimental protocol?
[12:07:29] <Adam Wiethuechter> This predates me but "lp0 on fire"
[12:07:31] <Martin Thomson> Lucas "PC Load Letter"
[12:07:42] Jiao Kang joins the room
[12:07:46] <Lucas Pardue> that's it
[12:08:16] <Jana Iyengar> @Pete: No, I'm asking if the LOOPS proponents have thought about this.
[12:08:20] <Xingwang Zhou> To Lixia &amp; Jana, LOOPS interacts with up layer TCP with ECN after loss recovery
[12:08:20] Jordi Paillissé Vilanova leaves the room
[12:08:25] <Lixia Zhang> @Pete: no protocol should be designed before one plays with the concept first to understand what happens.  experiments, if done right, always expose unknowns
[12:08:26] <Brian Trammell> (CW: 1990s printers)
[12:08:55] <Martin Duke> @jana yes we've been talking about interaction with endpoint loss detection/RTT measurement/CC
[12:09:06] <Martin Duke> that's one of the reasons ECN appears here at all
[12:09:09] <spencerdawkins> Speaking as an individual, I'd be comfortable with Experimental.
[12:09:43] <Martin Duke> i concur that some endpoints/applications will not benefit much, and some may be mildly degraded
[12:09:43] <spencerdawkins> That's a fine conversation for the ADs to have with the group, of course.
[12:10:10] <Gorry Fairhurst> Or even a research group - since the is potential for harm as well as good.
[12:10:22] ZHAOQIANYING leaves the room
[12:10:25] <Jana Iyengar> @Martin Duke: It's not clear to me that it will be mild. All that depends on the context of deployment.
[12:10:31] <Brian Trammell> Colin is still here, yes?
[12:10:34] ZHAOQIANYING joins the room
[12:10:40] <Colin Perkins> Yup
[12:11:00] <Martin Duke> @jana that might be. a good question for the group if benefit &gt; cost, or if that's worth an experiment
[12:11:01] <David Black> TSVWG FEC work has concluded - RFCs published, no current/ongoing activity.
[12:11:14] <Jana Iyengar> The biggest problem here is going to be once of misaligned incentive/control. Network operators might deploy LOOPS, but they have no idea what it is doing to application traffic, since they have no way to measure impact to applications.
[12:11:15] Gorry Fairhurst leaves the room
[12:11:15] Gorry Fairhurst joins the room
[12:11:17] <Brian Trammell> (in case we move the charter discussion in the RG direction, I'd like you to be able to speak to that :) )
[12:11:25] Fengwei Qin leaves the room
[12:11:27] <Lixia Zhang> @Xingwang : what I'm thinking is not the defined interactions (e.g. both react to ECN bit as defined), but rather, unexpected interactions, e.g. Jana mentioned the impact on BBR
[12:11:51] <Lucas Pardue> NWCRG is looking to wind up in the near future IIUC
[12:11:55] Fengwei Qin joins the room
[12:12:33] Reji Thomas leaves the room
[12:12:34] <DaveO> Qeustion: on measurement - does this belong as general enhancement ot Geneve?
[12:12:42] <Martin Duke> can someone post a link to the charter in jabber?
[12:13:10] <Brian Trammell> https://github.com/loops-wg/charter
[12:13:13] <Brian Trammell> is the working copy
[12:13:57] <Martin Thomson> Thanks Brian
[12:14:41] <cabo> noted
[12:14:46] Umberto Fattore joins the room
[12:15:07] <Li Yizhou> ECN is used for feedback. ECN is the one closest to "ready-for-enginnering". That's why MVP is proposed.
[12:15:39] <Jana Iyengar> So higher order point: loss detection and recovery goes all the way back to the days of the TCP and IP split. The e2e paper makes it an explicit example of what the network does not need to since not all traffic needs it. The reasons we do it at the link level is that that for paths that are *known* to be lossy when there's no congestion, and for links that have RTTs that are tiny as compared to the e2e RTT, it makes a ton of sense to do local recovery. This is specific to a link, but I'll push back against the assumption that this is generalizable to arbitrary multiple IP hops.
[12:15:45] <Martin Thomson> So I was going to argue for not doing this on the basis that it makes decisions about repair that might run counter to endpoint preferences, but Colin might have convinced me to retract that with "people are going to do this anyway, might as well have a good applicability statement"
[12:15:48] Yanyuan QIN joins the room
[12:16:09] hannuflinck@sure.im joins the room
[12:16:37] <Martin Thomson> Jana: to be fair, the e2e paper makes a subtly different argument about the necessity of endpoint repair, which makes in-network repair much less valuable
[12:16:44] <DaveO> @Jana: it depends on whether your mental model is tunnel==link or not
[12:16:46] <Jana Iyengar> @Martin: That applicability statement would actually be useful. That's not what I hear as being proposed.
[12:17:00] <Martin Thomson> Jana: yeah, without an applicability statement, this is harmful
[12:17:12] <cabo> SO let's add that to the charter
[12:17:16] Dawei Fan joins the room
[12:17:43] <Lixia Zhang> @Yizhou: there are undefined interactions. For people who'd do it anyway: I'm only suggest doing the work more scientifically, study the whole system before diving to do products.
[12:17:46] <Martin Thomson> waiting for someone to mention DSCP
[12:18:08] <David Black> +1 on applicability - would help a lot with Jana's concern about arbitrary multiple IP hops.
[12:18:17] Nabil Bitar leaves the room
[12:18:21] Nabil Bitar joins the room
[12:18:38] <Martin Thomson> don't do ECN seems like a poor mechanism for opt-out
[12:19:04] <Gorry Fairhurst> My Mic failed so in the chat: RFC3135 focussed on leeting the user have control over whether the "enhanced" its traffic. How does this work when an operator has a TE-team in the underlay... I was asking if this is something an endpoint might have visibility over (besides disabling ECN)?
[12:19:06] <DaveO> Ce we come back to my quesiton about tunnel traceroute for Geneve?
[12:19:31] <Brian Trammell> DaveO: is that MIC:
[12:19:32] <Brian Trammell> ?
[12:19:52] <Brian Trammell> (and +1 MT)
[12:20:20] <Martin Thomson> Think about video traffic: some of it is loss tolerant, others not.  Distinguishing between those creates excellent signals for traffic analysis.
[12:20:22] <DaveO> Perhaps we should strongly indicate that traceroute should not be blocked by operators employing Loops?
[12:20:45] <Martin Thomson> but if you can know what needs repair and what doesn't this works much better
[12:21:19] <Brian Trammell> stick it in the PLUS header
[12:21:19] <Jana Iyengar> DSCP!
[12:21:23] <Tom Jones> :D
[12:21:27] <Martin Thomson> chaching!
[12:21:42] <Jana Iyengar> Martin Thomson : You win
[12:21:43] <David Black> If using this to patch an underlay "pothole" what's the point of endpoint interaction?
[12:21:51] <DaveO> @Brian: start with RFC3609
[12:21:57] <Colin Perkins> So, DSCP isn't deployable, so we're back to ECT(1) to signal it...?
[12:22:15] <Gorry Fairhurst> DCSP :-), if a DSCP tells you what you wish to know.
[12:22:36] <Gorry Fairhurst> CP: Does ECT(1) signal don't add latency... or latency is important?
[12:22:47] Yolanda Yu joins the room
[12:22:50] <Colin Perkins> ECT(1): the multi-function bit
[12:23:02] Shwetha Bhandari leaves the room
[12:23:11] <Mirja Kühlewind> DSCP2.0 :-) -&gt; new wg to make DiffServ deployable
[12:23:14] <Mirja Kühlewind> or maybe an rg
[12:23:49] Fengwei Qin leaves the room
[12:23:56] <Martin Thomson> endpoint signaling is a hill many people have died on, not just PLUS
[12:24:04] <cabo> Network operators can do stupid things, and we should help them not do that, so I'm all for an applicability statement.
[12:24:07] Yolanda Yu leaves the room
[12:24:08] Fengwei Qin joins the room
[12:24:33] <Colin Perkins> I've said before: QIRG can do a quantum superposition of different meanings of the ECT(1) bit and solve all these signalling problems
[12:24:57] <Mirja Kühlewind> sounds good Colin. When will they deliver that?
[12:25:27] <cabo> (Not true in WiFi, BTW)
[12:25:28] <David Black> Also need a quantum tunneling protocol to get DSCPs across operator boundaries ...
[12:26:12] <Colin Perkins> @mirja next April?
[12:26:34] Fengwei Qin leaves the room
[12:26:39] <DaveO> should we select quantum entanglement as the FEC aproach?
[12:26:50] Fengwei Qin joins the room
[12:27:00] <Brian Trammell> man spencer's picking on all my bofs today isn't he?
[12:27:31] <Gorry Fairhurst> +1 with Jana .. BUT, I do see the value in somehow saying something about the problem space, because others wiull try to repeat the story.
[12:27:32] Adam Wiethuechter leaves the room
[12:27:36] Adam Wiethuechter joins the room
[12:27:47] <Martin Thomson> spencer: I might have to find a recording of dkg in Hawaii saying "are you saying poor people don't deserve privacy too?"
[12:28:03] <Lucas Pardue> queeg - quantum usability extensensibility of ECN for Good
[12:28:13] <Martin Thomson> VOTE
[12:28:24] <Jana Iyengar> There's another issue I'd like to note: operators measure traffic aggregate metrics to determine improvement, and that can run contrary to endpoint measurements of application metrics
[12:28:41] Fengwei Qin leaves the room
[12:28:51] Fengwei Qin joins the room
[12:28:51] <Colin Perkins> @lucas: excellent!
[12:28:56] Philip Eardley leaves the room
[12:29:00] <Mirja Kühlewind> so provide better measurements to the operators?
[12:29:06] Fengwei Qin leaves the room
[12:29:08] Fengwei Qin joins the room
[12:29:09] Jiao Kang leaves the room
[12:29:23] Fengwei Qin leaves the room
[12:29:29] <Lucas Pardue> smoke me a kipper, I'll be back for WPACK
[12:29:34] <Gorry Fairhurst> Some may have TE teams who cna fix things.
[12:29:39] <Martin Thomson> how about we provide you with the contents of our hard drives, Mirja?
[12:29:40] Fengwei Qin joins the room
[12:29:46] <Matt Green> @Lucas doesn't that also imply the existance of one for Evil?
[12:30:19] <Mirja Kühlewind> would certainly be interesting to know which information you collect on a per-user basis ;-)
[12:30:37] <Adam Wiethuechter> @Matt so queee....an elongate "what?"
[12:30:47] <Lucas Pardue> Is the evil bit more succesfulyl deployed than ecn...
[12:30:55] <Martin Thomson> Mirja, we publish what we collect
[12:30:57] <DaveO> @Jana: yes but I'm struggling with why these should be at odds if the operator is not a transit. For transit, i think the question is quite muddled.
[12:30:59] Jiao Kang joins the room
[12:31:21] Gorry Fairhurst leaves the room
[12:31:24] Gorry Fairhurst joins the room
[12:31:30] <Tommy Pauly> Is this "should we do applicability statement, assuming we do the work"?
[12:31:33] <Mirja Kühlewind> not sure how you define success for the evil bit
[12:31:33] <Martin Thomson> the way that the tool causes the buttons to jump around is really bad
[12:31:35] <Lars Eggert> i don't understand how spencer's question makes the first question any clearer
[12:31:48] <Tommy Pauly> I'm assuming this is conditional on doing the work
[12:31:49] <Jana Iyengar> Dave: I don't understand why it matters if it's a transit or not
[12:31:51] Daniel Gillmor joins the room
[12:31:58] Mach Chen joins the room
[12:32:00] <Jana Iyengar> I don't think this is conditiona
[12:32:03] <Lars Eggert> we can only do an applicability statement *if* we agree to work on this
[12:32:04] <Mirja Kühlewind> I guess if no-one is setting the evil bit, no-one is evil?
[12:32:15] Adam Wiethuechter leaves the room
[12:32:19] Adam Wiethuechter joins the room
[12:32:20] Yolanda Yu joins the room
[12:32:24] <Jana Iyengar> applicability statement doesn't need a WG, does it?
[12:32:31] <Tommy Pauly> That's the question
[12:32:32] <Martin Thomson> Alissa played musak
[12:32:33] <Lars Eggert> well, it needs IETF consensus
[12:32:36] <Tobia Castaldi> :)
[12:32:49] <Pete Resnick (the other one)> @Jana: Some ADs really don't like AD sponsored docs.
[12:32:50] Fengwei Qin leaves the room
[12:32:53] Stephan Emile leaves the room
[12:32:53] Fengwei Qin joins the room
[12:33:08] Fengwei Qin leaves the room
[12:33:08] Stephan Emile joins the room
[12:33:10] <Mirja Kühlewind> we have tsvwg
[12:33:12] Fengwei Qin joins the room
[12:33:13] Gang Yan leaves the room
[12:33:30] <Pete Resnick (the other one)> @Mirja: True.
[12:33:33] <DaveO> @Jana: because if not transit the operator is likely also hosting the endpoints and often does actually understand the application metrics through SLAs.
[12:33:36] <Gorry Fairhurst> Yea .. another transport/siugnal/ops issue - welcome at tsvwg!
[12:33:47] Stephan Emile leaves the room
[12:34:02] Stephan Emile joins the room
[12:34:13] <Colin Perkins> "Is it useful to look into the applicability of technologies like this" is a very different question to "is it useful to write an applicability sttement for this"
[12:34:18] Stephan Emile leaves the room
[12:34:18] Stephan Emile joins the room
[12:34:34] <Tommy Pauly> Right, to say "don't use this approach in any of these scenarios"
[12:34:39] Stephan Emile leaves the room
[12:34:48] Stephan Emile joins the room
[12:34:49] <Tommy Pauly> That's independent of defining any standards
[12:35:11] <Pete Resnick (the other one)> @Spencer: Mirja mentioned above that an A/S could be a TSVWG thing.
[12:35:27] Yolanda Yu leaves the room
[12:35:27] Yolanda Yu joins the room
[12:35:30] <Jana Iyengar> @Gorry: +1 to tsvwg
[12:35:51] <spencerdawkins> @Pete/Jana - that would work for me.
[12:36:10] <Jana Iyengar> @Pete: yes, that SGTM
[12:36:27] <Pete Resnick (the other one)> Don't blame me. Mirja's suggestion. ;-)
[12:36:32] <Martin Thomson> An applicability statement as a precondition for a working group seems reasonable to me.
[12:36:35] <spencerdawkins> Applicability statements are PS, for what that's worse.
[12:36:48] <spencerdawkins> "worth"
[12:36:56] <Martin Thomson> I don't know if this is "Bring me a rock", but it is clear that an applicability statement would be useful.
[12:37:12] <Pete Resnick (the other one)> @spencer: People use "A/S" to mean many things. Could be a BCP.
[12:37:22] <Jana Iyengar> Martin: I thikn there's interest in working on the AS together
[12:37:26] Zongpeng Du joins the room
[12:37:28] <spencerdawkins> @Martin - I don't think that's "bring me a rock".
[12:37:35] <Gorry Fairhurst> Tobe clear I said: (That is it could discuss this topic, that doesn't necessarily mean publish a doc.)
[12:37:53] Jiao Kang leaves the room
[12:37:56] Yolanda Yu leaves the room
[12:37:56] Yolanda Yu joins the room
[12:38:10] Jiao Kang joins the room
[12:38:19] <spencerdawkins> @Pete, yes, and if it's guidance for people who are working on products now anyway, it needs lots of eyes.
[12:38:28] Jianglong leaves the room
[12:38:32] ZHAOQIANYING leaves the room
[12:38:44] <Jana Iyengar> push bacl
[12:38:47] ZHAOQIANYING joins the room
[12:39:21] Jianglong joins the room
[12:39:28] Dawei Fan leaves the room
[12:39:43] Pete Resnick leaves the room
[12:39:51] <Gorry Fairhurst> There is a CC topic; but also a much broader problem (I seem to agree with jana today).
[12:40:06] <David Black> @Gorry@Jana +1
[12:40:16] Umberto Fattore leaves the room
[12:40:17] Steffen Lindner leaves the room
[12:40:33] <Nicolas Kuhn> should we try to have sdwan box providers involved ?
[12:40:33] Martin Thomson leaves the room
[12:41:28] Gorry Fairhurst leaves the room
[12:41:52] <Lars Eggert> "forte" which way?
[12:41:54] <DaveO> One final input: I would be useful to have a document that says what the necessary preconditions are for doing work and what constraints need to be in place.
[12:42:05] <Lars Eggert> we had "forte" for *and* against
[12:42:19] <Lars Eggert> thanks for clarifying
[12:42:48] <DaveO> For two examples: (a)measurment, (b)visibility to parties other than the tunnel endpoints
[12:42:50] <spencerdawkins> "Forte" wasn't just Carsten, or even the proponents :-)
[12:43:14] David Oran leaves the room
[12:43:16] Laura Al Wardani leaves the room
[12:43:18] <Martin Duke> Thanks to the chairs, and the proponents
[12:43:18] DaveO leaves the room
[12:43:25] Markus Amend leaves the room
[12:43:26] Nicolas Kuhn leaves the room
[12:43:27] Frode Kileng leaves the room
[12:43:28] Jorge Rabadan leaves the room
[12:43:29] Lars Eggert leaves the room
[12:43:34] Vincent Roca leaves the room
[12:43:35] Matt Green leaves the room
[12:43:35] Chonggang Wang leaves the room
[12:43:36] David Black leaves the room
[12:43:37] Diego Lopez leaves the room
[12:43:38] Luis Contreras leaves the room
[12:43:39] Chi-Jiun Su leaves the room
[12:43:39] Tommy Pauly leaves the room
[12:43:39] Adam Wiethuechter leaves the room
[12:43:39] Greg White leaves the room
[12:43:42] Jiao Kang leaves the room
[12:43:44] ZHAOQIANYING leaves the room
[12:43:46] Carlos Bernardos leaves the room
[12:43:48] Georgios Karagiannis leaves the room
[12:43:49] Kazuho Oku leaves the room
[12:43:50] Ekaterina Volodina leaves the room
[12:43:50] Colin Perkins leaves the room
[12:43:51] Hannu Flinck leaves the room
[12:43:53] Spencer Dawkins leaves the room
[12:43:53] Magnus Westerlund leaves the room
[12:43:54] Martin Duke leaves the room
[12:43:58] Dirk Hugo leaves the room
[12:43:59] Julien Maisonneuve leaves the room
[12:44:00] Cullen Jennings_401 leaves the room
[12:44:01] Fengwei Qin leaves the room
[12:44:03] Yanyuan QIN leaves the room
[12:44:04] Xuesong Geng leaves the room
[12:44:05] Peng Liu leaves the room
[12:44:08] frodek leaves the room
[12:44:09] Brian Trammell leaves the room
[12:44:10] Cullen Jennings leaves the room
[12:44:16] Mirja Kühlewind leaves the room
[12:44:16] Mach Chen leaves the room
[12:44:19] Li Yizhou leaves the room
[12:44:20] Dirk Kutscher leaves the room
[12:44:28] <Jana Iyengar> @Gorry, I'm happy for such days when we agree :-)
[12:44:29] Ronald in 't Velt leaves the room
[12:44:31] Jianglong leaves the room
[12:44:33] Meetecho leaves the room
[12:44:39] Hirochika Asai leaves the room
[12:44:40] Wim Henderickx leaves the room
[12:44:47] Stuart Card leaves the room
[12:44:52] Lixia Zhang leaves the room
[12:44:53] Huaru Yang leaves the room
[12:44:54] Tom Jones leaves the room
[12:44:59] Zhuangyan leaves the room
[12:45:07] alex-meetecho leaves the room
[12:45:12] Tobia Castaldi leaves the room
[12:45:12] Gunter Van de Velde leaves the room
[12:45:12] Qin Wu leaves the room
[12:45:12] Carsten Bormann leaves the room
[12:45:12] Patrick McManus leaves the room
[12:45:12] Jana Iyengar leaves the room
[12:45:12] Shuai Zhao leaves the room
[12:45:12] Daniel Gillmor leaves the room
[12:45:12] Yue Yin leaves the room
[12:45:12] Xingwang Zhou leaves the room
[12:45:12] Stephan Emile leaves the room
[12:45:12] Marie-Jose Montpetit leaves the room
[12:45:12] daniel.bernier@bell.ca leaves the room
[12:45:12] Akbar Rahman leaves the room
[12:45:12] Nabil Bitar leaves the room
[12:45:12] Alexandra Kelly leaves the room
[12:45:12] Lucas Pardue leaves the room
[12:45:12] Christine McCarthy leaves the room
[12:45:12] Rohit Abhishek leaves the room
[12:45:12] Alessandro Toppi leaves the room
[12:45:12] Yolanda Yu leaves the room
[12:45:12] Craig Taylor leaves the room
[12:45:12] Bhavit Shah leaves the room
[12:45:12] Praveen Mada leaves the room
[12:45:12] Chunshan Xiong leaves the room
[12:45:12] Zongpeng Du leaves the room
[12:50:19] Magnus Westerlund too leaves the room
[13:01:25] spencerdawkins leaves the room
[13:05:45] hannuflinck@sure.im joins the room
[13:07:44] hannuflinck@sure.im leaves the room
[13:12:40] hannuflinck@sure.im leaves the room
[13:43:39] Qin joins the room
[13:47:31] Qin leaves the room
[14:06:30] avezza leaves the room
[15:09:15] Pete Resnick (the other one) leaves the room
[15:59:51] cabo leaves the room
[16:03:41] cabo joins the room
[16:03:51] cabo leaves the room
[16:18:31] jmagallanes leaves the room
[16:26:48] jmagallanes joins the room
[16:27:53] jmagallanes leaves the room
[17:01:52] Qin joins the room
[17:10:59] Brian leaves the room: Disconnected: closed
[17:32:34] Qin leaves the room
[17:37:59] cabo joins the room
[17:52:47] Brian joins the room
[17:52:52] Brian leaves the room: Disconnected: closed
[18:10:55] Brian joins the room
[18:18:37] Brian leaves the room
[18:19:23] cabo leaves the room
[19:09:25] chi.jiun.su leaves the room
[19:51:15] chi.jiun.su joins the room
[19:52:14] chi.jiun.su leaves the room
[20:50:57] cabo joins the room
[22:59:56] cabo leaves the room