TSVAREA @ IETF-106 (Singapore) Thursday, Nov 21, 2019 17:40-18:40 - Thursday Afternoon session III Room: Padang Note Taker: Martin Duke * Administrativa - Mirja and Magnus (remote) - Note Well, Blue Sheets, Jabber Scribes, Agenda Bashing - TSV Overview and status - Allison Mankin stepping down from review team. Thanks! - Kyle Rose, Ian Swett, Martin Duke joining. Welcome! - MPTCP will close before Vancouver - 4 new documents in Editor queue, none published - Code of Conduct reminder -- we can all make this a nice place to work!! - WG chairing - Few chairs positions have opened in the last few years -- should we rotate them? - Please speak up if you would like to chair. - Also provide feedback about chairs to ADs. Any feedback? Spencer Dawkins: I wish I had explicitly asked for keepalives when I was AD, to periodically make sure chairs are willing to continue to serve. People don't want to let us down by quitting. Colin Jennings: We should rotate chairs to grow more good new people, not get rid of old people who are bad. - Port registry - Thanks to the experts team; requests come in every week. - We might update the documents about the port registry -- we get requests for closed domains, or for things that don't need fixed ports. - 96 requests in 2017, 98 in 2018 -- 36 and 19 assigned, respectively. Martin Thomson (MT): Telling people no in registries doesn't work, they use it anyway. Mirja Kühlewind: Lots of random protocol requests. MT: If experts are defenders of the internet, that's a useful function. But it's also forcing destructive patterns. Mirja: We have port discovery mechanisms, so they don't need a fixed port. This is different from other areas. Magnus Westerlund: We have enough space for 50 years. Mark Nottingham (mnot): Are these requests or port #s? Magnus: Requests, but 1 per request is standard. David Black: We have supply because there's been a good process. Jana Iyengar: We do receive requests that are updates, so it's not that many numbers. * QUIC update - Mark Nottingham (mnot) - Transport and TLS drafts are in "late stage" processing - Recovery and HTTP drafts about to get there as well. - interop is improving over time. - WGLC at end of year? - but there will be more time for interior testing. - To IESG mid-2020. - New versions, extensions, applications, etc are tracked on the github wiki - v2 -- important to avoid ossification, but what's in it? -> Might need recharter - Lots of drafts for extensions, etc. -> will adopt some but need to recharter. David Black: Another exposed bits draft is out there in tsvwg mnot: Put it on the wiki - Other applications (than HTTP) over QUIC -> WG will not adopt. Send it somewhere else. MT: QUIC is both TCP2 and UDP2, and we can't scale to all the apps that would use it. QUIC WG will be involved early to establish best practices, maybe new documents. mnot: Random tools, etc. bring new application drafts to ADs MT: Use well-established dispatch; also "use" all ADs, not only Transport ADs. Patrick McManus: I am Mr. dispatch, I can help. Gorry Fairhurst: On QUIC over satcom -- has lots of different solutions. Should there be lots of drafts? How to handle? mnot:If this is quicv1, file an issue. For the future, we'll see. Composing into drafts would be good. Spencer: Operations/mgmt draft isn't on the list. Huh? mnot: Yes, they're important, but not getting enough attention right now. Mirja: Those are WG items. Spencer: We had QUIC ops/mgmt considerations in the QUIC working group charter because there was nowhere else to put it. Now that Media Operations (MOPS) has been chartered, they're already talking about measuring troubleshooting QUIC from an operator perspective in their first working group session. Maybe this document should go into MOPS? David Schinazi: I a QUIC-enthusiast: I love it so much and am authoring tons of stuff. Got a lot of "not yet" before. Are you opening the floodgates to innovation? Mirja: yes! Roni Even: Spencer+1. We need changes to v1 for some ops stuff. Who is the quic police? mnot: If you're changing v1, it's an extension or new version. If app-specific, probably not a version. Ted Hardie: We joked of a QUIC area due to all the excitement. Run QUIC dispatch for a couple of IETFs to get the flood in the right spot. A short meeting that concentrated the QUIC police in one room would lead to better decisions. Ted: Lots of creativity right now. people need to understand what they are doing *to* the protocol, not just *with* it. We should write "the hitchhiker's guide to QUIC". Intended for all audiences. Mirja: That's what applicability and mgmt docs are for. Ted: That's not what I'm looking for. Spencer: Imagine if TCP roadmap had started before there were 150 RFCs! Cullen Jennings: Yay for datagram! Art is encapsulating real-time in QUIC, come join us. John Lennox: New apps will require new things, but they may not know what? Ae may need low-latency congestion control. How does QUIC need to change? Mirja: Low-latency congestion control is protocol independent. mnot: The people who can answer this question are in QUICWG. Next year is a good time to talk to them. MT: We have limited capacity to do additional work for next 6 months. Must manage priority. Ted's idea might work. Let's do quic-dispatch! Brian Trammell: I endorse quic-dispatch, not permanently. quic app interface is much richer than others, and undefined. quic wg decided not to do an API. quic-dispatch might provide insight to write that API. mnot: This week we see HTTP/2 used in bad ways, we need to avoid that here. McManus: WG has been unique. Extended focus and volume of work. Composed of three areas! And yet, the work in those three areas is not paced the same way. transport-dominated, but we had just enough of the right voices for it to work. mnot: Lars and I have seen implementers getting exhausted. Everyone is excited, but these people are not infinitely available. Jana: Don't get emotionally attached to protocols. I like quic-dispatch. ICCRG is a good place for congestion control, even if motivated by quic. David Black: More support for quic-dispatch or transport-dispatch. tsvwg is the catch-all and "bursting at the seams" Mirja: great feedback. Night not do it in Vancouver due to other constraints but let's see!