My vote is to publish VXLAN ala port 4789. Note its not on the list but the most deployed in real product. That would be practical.
Dino P.S. And this is probably the 4th time I have stated this. And as time goes on, VXLAN is even more deployed. > On Jul 14, 2016, at 1:26 PM, Jesse Gross <[email protected]> wrote: > > I agree as well. We’ve had this question outstanding for the past couple > years and haven’t had much luck on picking/merging. Given the timing, I think > it’s effectively impossible to do so now. It seems like #1 is the pragmatic > choice. > > On 7/14/16, 1:00 PM, "nvo3 on behalf of Lucy yong" <[email protected] on > behalf of [email protected]> wrote: > > Agree with Anoop’s analogy. > > Lucy > > From: nvo3 [mailto:[email protected]] On Behalf Of Anoop Ghanwani > Sent: Thursday, July 14, 2016 12:46 PM > To: Tom Herbert > Cc: Bocci, Matthew (Nokia - GB); NVO3 > Subject: Re: [nvo3] Mail regarding NVO3 data plane drafts > > #2 is going to be nearly impossible (or it would have happened earlier, and > if it were possible, why would we even bother publishing the other two?), so > that makes it an easy choice. > > Anoop > > On Thu, Jul 14, 2016 at 10:01 AM, Tom Herbert <[email protected]> wrote: > On Thu, Jul 14, 2016 at 9:21 AM, Bocci, Matthew (Nokia - GB) > <[email protected]> wrote: > > WG, > > > > The NVO3 working group has adopted three data plane encapsulations: > > - VXLAN-GPE, > > - Geneve, > > - GUE (although the draft is moving to the Intarea WG, we > > anticipate that NVO3 will still reference this). > > > > We have discussed this situation with Alia and we feel that there is little > > benefit to the community in publishing all three as standards track RFCs. > > > > We would note that the discussion on the drafts has been relatively light > > since their adoption. There has not been serious discussion about their > > relative pros/cons (if any), or about the actual usefulness of their > > extensibility or differentiators. > > > > This leaves two options: > > > > 1) Publish all of them as informational or experimental, potentially moving > > one of them to standards track in the future based on > > implementation/deployment. > > 2) Pick one now based on technical and/or implementation/deployment > > criteria. > > > > I would like to propose a third option. Create a design team in nvo3 > to come up with the goal of proposing one data plane protocol that > consolidates the best features of the three. Outside of extensibility > there is fundamentally little difference amongst these, and the > different models of extensibility (flag-fields, TLVs, NSH) could be > fit into one protocol. Also, the encapsulation design considerations > (draft-ietf-rtgwg-dt-encap-01) provides a good reference for creating > such an encapsulation protocol. > > Thanks, > Tom > > _______________________________________________ > nvo3 mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/nvo3 > > _______________________________________________ > nvo3 mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/nvo3 _______________________________________________ nvo3 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nvo3
