IETF 106 - DISPATCH WG / ART Area session Sophia room - 10:00 Monday 18 Nov 2019 The chairs would like to express their gratitude to Pete Resnick and Steve Donovan for taking excellent notes, and Jonathan Lennox for acting as Jabber relay. Dispatch Summary The working group discussed two drafts: - SIP Auto-Peer ( draft-kinamdar-dispatch-sip-auto-peer-01 ) The discussion was generally favorable towards progressing the work. Discussion moved on to where the work should be done, and to suggest that we needed to involve people who would actually implement and/or deploy it. Conclusion: The ADs are leaning towards a mini-WG. The next step is to draft a charter. Real-Time Internet Peering Protocol (RIPP, draft-rosenbergjennings-dispatch-ripp-03) Discussion showed interest in the work, but a need to refine the scope. The effort would probably be too large for a "mini" WG. We need to make sure there is HTTP clue involved. Conclusion: Organize a (possibly virtual) BoF. The proponents would like to advance quickly. Cullen to organize a side-meeting Thursday morning. ART Area Summary: The chairs, ADs and BoF chairs plugged various BoFs and other meetings of interest. Bron Gondwana led discussion of various efforts to standardize use of push services for IMAP, etc. There appears to be some interest in the work, but there were also concerns that deployment may be constrained by the mobile OS providers. Only a small number of people responded positively to the question of who would participate in such work. Justin Richer briefly described work on HTTP request signing. More detailed discussion is to occur in secdispatch. However, there is some indication that this is reasonable work for httpbis. Detailed notes follow: --------------------- DISPATCH WG Agenda: https://datatracker.ietf.org/meeting/106/materials/agenda-106-dispatch-03 Materials: https://datatracker.ietf.org/meeting/106/session/dispatch Notetakers: Pete Resnick & Steve Donovan Jabber Relay: Jonathan Lennox Note Well was noted well. Agenda bash. Note that the two Dispatch items have been switched in order. No objections heard. AD Comments: Alexey Melnikov: Welcome Patrick as new DISPATCH co-chair. Chair rotation is to be a regular thing. Thanks to Mary Barnes for her years of service. Barry Leiba: Add thanks to Mary. Thanks to the 5 who accepted nominations for ART AD. Please give your comments. Richard Barnes: Nomcom is open for office hours 9-10 every morning this week. Stop by. SIP Auto Peer presentation https://datatracker.ietf.org/meeting/106/materials/slides-106-dispatch-sip-auto-peer-00 Discussion: Suhas Nandakumar: This is something we've come across. We need to talk about what is needed, and there are places where manual intervention will be required, but this is worth doing. Patrick (from chair): Any more broadbased interest? Cullen Jennings: Are there any vendors in the room who've seen this problem? Gonzalo (Cisco) : We see this enough that it seems worth solving. Ben Campbell (from floor): How fixed will the capability set being offered be over time? - We don't expect changes unless there are radical changes to SIP/RTC. Ted Hardie: Does this belong better in OPS/Management? For instance, YANG stuff. They might have good input. - We're open to talking to them, but we haven't so far. Alexey Melnikov: ADs think that a mini WG might be a good idea. Might land in O&M, might be ART. Robert Sparks: Have you had any conversations with SIPConnect/SIPForum? Is this something they would be receptive to adding? - We see it as complementary - Ben: You should talk to them. Chris Wendt: Is this too little too late? Yes, it makes our life hard, but it's not clear that there's motivation to standardize. - We're hoping that down the line folks will see the time/$ advantage. Mary Barnes: Given how we never could get folks to even agree on basic SIP device configuration, I'm not sure this will go anywhere either. Not that I don't think it's a good idea. Ben: Anyone who thinks we *shouldn't* work on this? (Crickets). Summary: Probably worth doing, but concerns that it might not be used. Need to coordinate with folks who might use it. ADs think mini WG is good, ART or O&M. Brian Rosen: that SIPCORE might be OK. Adam Roach: Could adjust SIPCORE charter. Jon Peterson: Might want to use the MODERN. Robert Sparks: there's no actual SIP there, just nouns about sip/rtp things Mary Barnes: And, I agree with Jon P about MODERN. If stuff works, there is no impetus for change. Barry Leiba: Mary's point is well taken: Might be better to have a mini WG with an OPS advisor to increase visibility. Cullen Jennings: Benoit's been doing some review from a YANG point of view. Conclusion: ADs leaning toward a mini WG. Drafting a charter would be next step. Adam Roach: Put call out here and on SIP list to get interest. (8-10 folks put hands up in this room.) Real Time Internet Peering Protocol https://datatracker.ietf.org/meeting/106/materials/slides-106-dispatch-ripp-01 Ted Hardie: Clarifying question: Are you concerned with cases where there's processing after speech recognition, or speech recognition itself? - If speech recognition is local, less problem. If you're calling out to other service, that's when it counts. Ted: Thanks. Probably want to say in the document that it's both voice recognition *and* callout. Bernard Aboba: Chris Wendt: EKR: This sounds like a bit late to the game. If the plan is to use this to replace things, there's going to be friction. Issue with media going through a centralized point. Jonathan Lennox: Either design this with video from the beginning or declare it out of scope. We tried doing one after the other before, and it didn't work. Jon Peterson: Divide UNI and NNI. Brian Rosen: This sounds like SIP 3.0. This is redesigning multimedia connections. Maybe that's OK, but let's not sidestep into it (like we did with SIP). Mo Zanaty: Video in scope adds significant complexity. EKR: Proposed path forward is BOF. Cullen: yes, should be a working group. Ben: can dispatch to a working group. Cullen: not a miny working group. Mark Nottingham: Suggestion: BCP56bis is in HTTPWG. Advice on how to build things on top of HTTP. Please either follow that document or comment on why it might need to change. Cullen: Agreed. We do need some HTTP clue. Chris Wendt: Need to figure out what kinds of problems we're trying to solve. Brian Rosen: The more you try to limit the scope, someone is going to try to expand it. You could start with a limited scope, but admit that you're the big replacement. Cullen: Yes, it seems inevitable to have a bigger scope. Ben Campbell: Chairs believe scope needs to be big, likely cross area impacts. Talking about a BOF. Cullen: need to find a way to move faster than normal. Barry : Can do things like mailing lists now Pete: Can do a virtual BOF. Robert: Cullen, right now you own the delay. EKR: The reason people want a BOF is because there's not consensus on what needs to be done. Aaron : Can we do something later this week. Cullen: I don't think we could have something together in that time. Alissa Cooper: The 2 years to charter WebRTC is short in comparison to the 8 years to get the documents done. Cullen: One solution would be to keep it outside the IETF. Ted Hardie: Changing the interop strategy (i.e., no need to do backwards compat) makes this a different time frame. Stick to this. Ben: Looking at BOF, will there be side meetings. Cullen: difficult to corrdinate. Mo Zanaty: Could help the WEBtransport effort. Conclusion: Will look to chartering a BOF (perhaps a virtual one soon). There is definitely interest. Clearly note the need for speed. Perhaps will grab a room on Thursday. ART Area portion of meeting Meetings of interest slide. (Note typo in Web Transport BoF item.) Dave Lawrence on ABCD: Recent history on DNS security. Though DoH has specific issues, ABCD is meant to be generic discussion of what kinds of problems should be worked on in IETF. Bernard Aboba on WebTrans: Next step beyond websockets. Proposals are on the table. Discussion of use cases. David S.: Also related to W3C work; they're looking to leverage this. Patrik McManus on WPACK: Web packaging work. Working group forming BoF. Justin Richer: TXAUTH happening this afternoon. "Like OAuth, but hopefully less messy." PUSH Drafts - Bron Gondwana https://datatracker.ietf.org/meeting/106/materials/slides-106-dispatch-dispatch-push-00 Cullen Jennings: Doesn't everybody use Apple's or Google's? Bron: Yes, but no way to communicate to them to tell them to push. Cullen: So doesn't this depend on the two? Bron: You can put something in front of the services, and use the services for those things you can. Cullen: Need to clarify the scope. Neil Jenkins: Clarifying the scope: Need a standardized way to set up the channel with the provider. Seth ?: Is this third party for twilio on my phone, or more generally for mail/calendar/etc.? Bron: The latter. Sean Leonard: Good work to pursue in the IETF. Why not EXTRA? Bron: Sure, why not. Alexey: This is beyond the current charter. Barry: Right people in the room; could just re-charter. Adam Roach: RFC 8599 does this for SIP. Worth looking at. Matt Miller: There needs to be an existing relationship between the service and the push service. That complicates it, but it's necessary. The concern is about SPAM / too many bogus requests. That's what makes the relationship with Apple/Google to make this work. Chris Wendt: Apple has made it extremely restrictive. This is going to be a Terms of Service issue, not a protocol problem. Neil Jenkins: The way to make this work now needs to have the service component that interacts with the service. In the future, you could get the services to support removing the intermediary. Ben Campbell: Will these people work with the IETF? Bron: Yes, I believe so. Question for who would be interested on working on this: Not much (1? in room, 1 in Jabber) HTTP Signing - Justin Richer https://datatracker.ietf.org/meeting/106/materials/slides-106-dispatch-http-signing-00 (Will also be discussed in SECDISPATCH) Cullen Jennings: How does it relate to WebPack? Ted Hardie: They're very distinct. There was discussion whether they should be tied, but no. Justin: They should at least talk to eachother in some instances. Mike Jones: What pieces need to land? Justin: Not sure yet. Where to start is still open. That's the reason it's at SECDISPATCH. Mike: These groups need to talk to eachother. (Incompatibilities potential; would be bad.) Mark Nottingham: This is on the radar in HTTP. We haven't moved on it because of what you've said. Need to include HTTP community in this discussion. Might land in HTTP. Carsten Borman: CoAP has dealt with this as well. John Bradley: Signing is easy, HTTP is hard. Need to work with HTTPWG to make progress. Conclusion: Further discussion at SECDISPATCH, but sounds like HTTP WG might be a good choice. AOB: RIPP side meeting is on Thursday 8:30 am Sophia room. End of meeting.