Ted Lemon wrote:
> week ago? I think this is a key point: the end user has no idea
> what ECMP is, what its operational characteristics are, what links
> it might function over, etc. So ECMP would have to self-configure,
> and that includes the sort of stuff Juliusz was talking
> about—noticing interfering versus non-interfering paths, etc. This
> is a research project. I think it would be great if homenet could
> do this work at some point in the future, but it is not something
> that should be part of the base requirements for the homenet,
> because if it were, it would delay availability of a complete
> homenet spec by years.Alia Atlas <[email protected]> wrote: > ECMP or downstream paths is not a research project; it is common used > technology. When the traffic streams desired are larger than can fit > across a single path, it becomes critical. > Figuring out how to handle interfering vs. non-interfering paths is, I > think, orthogonal to ECMP. But, Ted's point is that without a way to identify what are the non-interfering paths are, that is, the non-interfering *WIRELESS* paths, then ECMP won't be useful except on wired links. Wired links are already fast enough, and in the cases where they aren't, and one plugs two GbE cables between two routers, it seems that we don't need ECMP, just automatic link-aggregation, to make it work. And, it would be very nice if: > From a user's perspective, use of multiple paths would be transparent - > except that they'd see better performance through > their network. There can be high-bandwidth demands like data backup or > streaming multiple high-def video. They could be made transparent, but we don't think that they can be made that way for wifi, which is where the major benefit would be. -- Michael Richardson <[email protected]>, Sandelman Software Works -= IPv6 IoT consulting =-
signature.asc
Description: PGP signature
_______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
