TSVAREA meeting Session 2 Wednesday 10
Overview of various WGs from the TSV ADs
Martin Duke (MD): Lots of multipath work going on across transport WGs. Things like scheduling might be common to different proposals, maybe a WG needed?
Lucas Pardue: I support that general idea, QUIC is focusing on a narrow scope and other things could be done elsewhere in parallel
MD: There’s SCTP work happening too, perhaps work a WG. We welcome input on that
some work done, thanks
Spencer Dawkins: you mentioned impact of multipath on congestion control. I think I’m seeing other impacts of multipath that could come up. Maybe low-level scheduling decision vs high-level strategy decisions. “what are you trying to do with this packet?”.
Spencer (separate point, sorry if that wasn’t obvious!): Trying to get my head around people trying to control performance aspects of various interactions between different kinds of traffic within an application. Is this something that we would look to ICCRG for? How do you see that?
MD: If it’s a research group it’s not really my job. If Jana wants to take it to ICCRG, I won’t rip it out of his hands. There are 4 multipath transport right now, they all have some kind of failover. To solve these things, there ought to be a forum for discussion. I’m open on that.
Zahed Sarker: ??
SD: When we were working on HIP, we had an IRTF RG, HIPRG, that worked on research topics like locating HIP endpoints, and we had an IETF WG, HIP, that worked on what was mature enough to engineer. We’ve had various experiences with paired RGs and WGs, so that’s not necessarily a recipe for success, but I wanted to mention it.
Magnus Westerlund: Somebody else, ???, also did a review.
MD: The slides were stale, sorry and thank you to all reviewers.
MD: We’ll see you in the halls!
Zahed: See you in other meetings.