Re: Requirements [was Re: Transport level multihoming]

[Ran Atkinson <rja@extremenetworks.com>]

At 17:16 03/04/01, Tony Li wrote:
>Given the choice between remaining compatible with 'legacy' IPv6 stacks or providing a useful multihoming solution, 
>which would you choose?

        I'd choose the useful multihoming solution, though
I'd work very hard to find some kind of graceful transition
schema.  Frankly, if IPv6 multihoming is better than IPv4
multihoming, it creates a tremendous incentive for folks
to move to IPv6 fast.  If IPv6 multihoming is NOT visibly better,
given the current operational realities, then there is a
pretty strong disincentive to move to IPv6.  The cell phone
folks seem quite comfortable with IPv6 being a large-scale
private network, which is not the end-to-end Internet.  Myself,
I fantasise about retaining a deployed end-to-end Internet 
in the future.

        With apologies for being plain spoken, to any of you 
who don't like having operational reality intrude on 
protocol design...

        As to other comments, the SCTP RFC is a fine thing,
but its FAR distant from being a spec on how to do multihoming
by only making transport-layer changes (as my previous issues
list hinted at).  While I usually can infer engineering
from a good architcture document alone, lots of folks on this 
list really really want to see both the architecture document
and the engineering/implementation document before "signing up"
to any particular proposal.  So folk who have ideas, please
write them up...

Cheers,

Ran