On 02/13/2018 08:30 PM, Erik Kline wrote: > On 13 February 2018 at 15:15, Fernando Gont <fg...@si6networks.com> wrote: >> Hello, Tommy, >> >> On 02/13/2018 08:04 PM, Tommy Pauly wrote: >>> Yes, I think this topic is very relevant to the work that can be done in >>> updated transport APIs, specifically around Path Selection. This is >>> relevant both for initial connection establishment (when we’re racing >>> various transport protocols across different paths, we need to know >>> which local addresses to use), and maintenance of connections that >>> support migration or multipath (which addresses can they use for new >>> flows). The address selection can also be inferred from other properties >>> of the transport connection, such as how long-lived the connection needs >>> to be. >>> >>> Fernando, what’s your plan for your draft? It seems like a great >>> reference for our drafts, and something we need to consider in our list >>> of application preferences when creating transport connections. >> >> Thanks so much for your comments! >> >> I this work is of interest to the TAPS WG, we could try to move this one >> forward in this wg, and also think of a subsequent I-D with >> recommendations (more on this later). >> >> Thoughts? > > When it comes to "recommendations" I'm not in favor of the node-based > recommendations in the NOTE in section 7.3.
This document is moving away from making recommendations, and just focusing on the problem statement. We'll remove the "Note" you referenced. > I think we should let > nodes create and use addresses as they see fit, and leave it to the > apps to figure out what they want to do (given the proper information > via useful APIs). The above said (i.e., the note will be removed in the next rev of this document), I think the "note" still gives control to the node to do what it pleases -- it was suggesting configuration of profiles for different types of networks (which would be a node-configuration setting, as opposed to network-induced configuration) Thoughts? Thanks! Best regards, -- Fernando Gont SI6 Networks e-mail: fg...@si6networks.com PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492 _______________________________________________ Taps mailing list Taps@ietf.org https://www.ietf.org/mailman/listinfo/taps