Re: Next question...

[Iljitsch van Beijnum <iljitsch@muada.com>]

On Mon, 25 Nov 2002, Tony Li wrote:

> Have we reached consensus that we need to deal with multihoming
> policies?

Not sure what you mean by that.

> And do we agree that doing so at the per-host level
> doesn't scale?

The problem isn't so much that it can't scale, but that it will be hard
to manage in large networks. If necessary, this can be fixed but this
makes the whole thing very complex. Complex gets you shouted at during
meetings.  :-(

My opinion: per-host multihoming is a good thing for two reasons:

1. Having two connections to the local network provides an additional
   level of protection against reachability problems, regardless of
   site multihoming
2. It will decrease demand for site multihoming so not-so-scalable
   solutions in this area have a longer life time

Another opinion: to achieve our scalability goals in a clean way, we
need another architecture for TCP/IP.

And a third: creating the above will take too long, so we need a
short-term routing based solution in the mean time.

:-)

Iljitsch