RE: ingress filteing problem

[Erik Nordmark <Erik.Nordmark@sun.com>]

> Agreed. For this reason, I believe that there are two tasks ahead of us.
> The first one is to solve ingress-filtering for the "IPv6 site which
> has two prefixes and two ISPs", and possibly three or four; the goal
> here is to make sure that new connections "just work". The second task
> is to define additional support mechanisms for multi-homing; the
> emphasis will be on "taking advantage of multi-homing" so things work
> "better", e.g., ensure that connections survive a re-homing event.

Agreed.

Perhaps we should add something to think-about and the architectural analysis
of that appears to be focused on the second task, to ask what impact different
approaches to that task might have on different approaches to the first task.

For instance, some approaches to the second task might need a solution
to ingress filtering. This prersumably means we want a good, long-term solution
for ingress filtering etc. Other approaches to the second task might instead
use locator rewriting which probably means we don't need to heavily optimize
the ingress filtering solution.

  Erik