Is this because of the video? The room seems empty!
", "time": "2024-03-21T03:01:20Z"}, {"author": "Reese Enghardt", "text": "I think it's because people are just getting back from lunch
", "time": "2024-03-21T03:01:40Z"}, {"author": "Christian Huitema", "text": "Realize that I cut my dinner short for this!
", "time": "2024-03-21T03:03:53Z"}, {"author": "Victor Vasiliev", "text": "I think the current text of the draft does not contradict what Christian says? It says \"SHOULD be evaluated\", right?
", "time": "2024-03-21T03:16:15Z"}, {"author": "Gorry Fairhurst", "text": "Yes - I think that's what we wanted to say.
", "time": "2024-03-21T03:18:26Z"}, {"author": "Gorry Fairhurst", "text": "What is needed is for people to take a coffee/beverage and see if the final text is acceptable...
", "time": "2024-03-21T03:18:52Z"}, {"author": "Christian Huitema", "text": "Real time control protocols nowadays do things like simulcast of three different definitions of the media, e.g. low def, mid def, high def. That gives them quite a bit of range of what they can do.
", "time": "2024-03-21T03:19:06Z"}, {"author": "Jonathan Lennox", "text": "Should data center (tiny rtt) be included among the special cases?
", "time": "2024-03-21T03:21:12Z"}, {"author": "Rodney Grimes", "text": "I think we need to consider the datacenter CC issues
", "time": "2024-03-21T03:22:10Z"}, {"author": "Eric Kinnear", "text": "https://github.com/ietf-wg-ccwg/rfc5033bis/issues/98
", "time": "2024-03-21T03:22:52Z"}, {"author": "Gorry Fairhurst", "text": "We'll add an issue - and we ought to work out which aspects are realte dto the general Internet and which do not.... this sounds like a new section would be good.
", "time": "2024-03-21T03:25:13Z"}, {"author": "altanai", "text": "As a thought i see that the document doesn't address oscillation cycles that sometimes happen due to reaction to CC
", "time": "2024-03-21T03:27:24Z"}, {"author": "altanai", "text": "think thats referred to as a boom n bust cycle
", "time": "2024-03-21T03:28:02Z"}, {"author": "Eric Kinnear", "text": "Christian: Clarification question or should we wait?
", "time": "2024-03-21T03:34:11Z"}, {"author": "Victor Vasiliev", "text": "I don't think this specific issue appears in non-window-based algorithms
", "time": "2024-03-21T03:39:28Z"}, {"author": "Victor Vasiliev", "text": "BBR typically has the opposite problem (app-limited results in underestimating bandwidth)
", "time": "2024-03-21T03:40:04Z"}, {"author": "Rodney Grimes", "text": "I feel the underlying principal that should be, and I believe actually is, described is \"application limitied\" bandwidth use, and the specific details are how that effects cwnd or pacing values in a CC. @christian, would that satisfy your want?
", "time": "2024-03-21T03:42:15Z"}, {"author": "Martin Duke", "text": "+1 to Lars
", "time": "2024-03-21T03:44:52Z"}, {"author": "Magnus Westerlund", "text": "I wonder a bit which aspect @Lars Eggert want to declare historic of the RMCAT output. Yes, none of the actual defined experimental algorithms have achieved large scale deployment. However, the real-time targeting algorithm in google webrtc library implementation is very largely deployed.
", "time": "2024-03-21T03:49:26Z"}, {"author": "Gorry Fairhurst", "text": "Like for the other specs, the purpose of the ID is to avoid continued growth of cwnd - I suspect this doesn't need much text, but is more than an Errata.
", "time": "2024-03-21T04:02:58Z"}, {"author": "Ian Swett", "text": "target_inflight?
", "time": "2024-03-21T04:09:53Z"}, {"author": "Christian Huitema", "text": "Ingemar's data show that L4S works well in presence of wireless throughput variability. That's good news, because L4S reacts in 1RTT. It should mean that other protocols that also react in 1RTT would also work.
", "time": "2024-03-21T04:10:33Z"}, {"author": "Ingemar Johansson", "text": "I am not 100% sure that I understood Ian's comment on SCReAMv2 , please clarify
", "time": "2024-03-21T04:17:27Z"}, {"author": "altanai", "text": "@Ingemar Johansson I think that msg is an alternative name for congestion window
", "time": "2024-03-21T04:20:38Z"}, {"author": "Martin Duke", "text": "@ingemar a concise definition of cwnd is \"max flight size\". Redefining cwnd to something that is not the flight size is very confusing
", "time": "2024-03-21T04:22:42Z"}, {"author": "Ingemar Johansson", "text": "OK, Guess one need to find something else \"reference window\", perhaps ?..
", "time": "2024-03-21T04:23:51Z"}, {"author": "Lucas Pardue", "text": "+1 sticking stuff in an active WG draft
", "time": "2024-03-21T04:41:51Z"}, {"author": "Christian Huitema", "text": "I did implement BBRv3 from the spec in Picoquic. And yes, there are a couple places where one has to improvise.
", "time": "2024-03-21T04:42:09Z"}, {"author": "Ingemar Johansson", "text": "Chairs mic is down
", "time": "2024-03-21T04:44:55Z"}, {"author": "altanai", "text": "Anyone else experiencing audio losses ?
", "time": "2024-03-21T04:44:58Z"}, {"author": "Reese Enghardt", "text": "@meetecho can you reset our audio?
", "time": "2024-03-21T04:45:01Z"}, {"author": "Mike English", "text": "Lost audio from the room and also the mic cam
", "time": "2024-03-21T04:45:16Z"}, {"author": "Lorenzo Miniero", "text": "Looking into it
", "time": "2024-03-21T04:45:24Z"}, {"author": "Eric Kinnear", "text": "Thank you
", "time": "2024-03-21T04:45:31Z"}, {"author": "Gorry Fairhurst", "text": "@Christian: where is the spec for BBRv3 that you implemented.
", "time": "2024-03-21T04:45:43Z"}, {"author": "Gorry Fairhurst", "text": "?
", "time": "2024-03-21T04:45:50Z"}, {"author": "Mike English", "text": "Lorenzo Miniero said:
\n\n\nLooking into it
\n
Good now, thanks!
", "time": "2024-03-21T04:46:32Z"}, {"author": "Lars Eggert", "text": "Pro tip: The first step to standardize is would be to agree on ONE version :-)
", "time": "2024-03-21T04:47:02Z"}, {"author": "Christian Huitema", "text": "Queue is closed, but since Cullen mentioned my work, we did have to add a mechanism to BBRv3 to handle the \"Wi-Fi suspension\" issues -- Wi-Fi radio getting in suspended mode for 100ms or sometimes more. Basically entering a \"suspended\" state when we find that, and doing some variation of \"careful resume\" when the suspension stops.
", "time": "2024-03-21T04:47:21Z"}, {"author": "Christian Huitema", "text": "I implemented the BBRv2 draft + the deltas delineated in the IETF117 presentation.
", "time": "2024-03-21T04:48:00Z"}, {"author": "Gorry Fairhurst", "text": "@Christain - thanks.
", "time": "2024-03-21T04:48:32Z"}, {"author": "Christian Huitema", "text": "On the real time issue, the one part that I really don't like is \"exit startup if delivery rate does not include for 3RTT\". Something like Hystart/Hystart++ would not create such big queues for so long.
", "time": "2024-03-21T04:50:08Z"}, {"author": "Jason Livingood", "text": "@EricKinnear Would be great to speak more loudly
", "time": "2024-03-21T04:51:14Z"}, {"author": "Neal Cardwell", "text": "Lars: Yes, the idea would be to focus all attention on BBRv3.
", "time": "2024-03-21T04:51:29Z"}, {"author": "Jason Livingood", "text": "(better now thx Eric)
", "time": "2024-03-21T04:51:35Z"}, {"author": "Neal Cardwell", "text": "(IMHO)
", "time": "2024-03-21T04:51:39Z"}, {"author": "Ian Swett", "text": "The gap between BBRv1 and BBRv2 is substantial, but BBRv2 to v3 is much smaller, FWIW
", "time": "2024-03-21T04:52:06Z"}, {"author": "Ingemar Johansson", "text": "Christian Huitema said:
\n\n\nOn the real time issue, the one part that I really don't like is \"exit startup if delivery rate does not include for 3RTT\". Something like Hystart/Hystart++ would not create such big queues for so long.
\n
Yes, that can potentially give quite large queues in cellular access queues
", "time": "2024-03-21T04:52:08Z"}, {"author": "Reese Enghardt", "text": "@meetecho it appears our screen shows 4 in the queue but it's only 1 in my Meetecho (Lars)
", "time": "2024-03-21T04:55:34Z"}, {"author": "Reese Enghardt", "text": "Maybe something got out of sync earlier
", "time": "2024-03-21T04:55:40Z"}, {"author": "Lorenzo Miniero", "text": "Thanks for the heads up, checking
", "time": "2024-03-21T04:56:14Z"}, {"author": "Martin Duke", "text": "maybe we should charter congestion control dispatch
", "time": "2024-03-21T04:56:35Z"}, {"author": "Gorry Fairhurst", "text": "Personally, I saw ICCRG as being a great home for analysis of mechanisms; performance analysis; etc - wgere we could do a deeper dive into analysis. If the goal is a fully specified interop spec, then an IETF WG seems the right place.
", "time": "2024-03-21T04:59:00Z"}, {"author": "Neal Cardwell", "text": "Yes, test suites would be great.
", "time": "2024-03-21T05:01:06Z"}, {"author": "Reese Enghardt", "text": "Yes, we had a test suite in ICCRG and some ideas on more work in that space
", "time": "2024-03-21T05:01:29Z"}, {"author": "Christian Huitema", "text": "What would really great is something like the QUIC interop setup, with participants bringing a docker image of their implementation and the infrastructure testing those implementations against themselves and each other.
", "time": "2024-03-21T05:02:23Z"}]