Re: Consensus check

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

On Sun, 24 Nov 2002, Tony Li wrote:

> Well folks, a few days ago I threw down the gauntlet
> and challenged the WG chairs to actually drive the
> discussion, get consensus on things, and then write
> down the consensus.

> So far, the gauntlet is still on the floor.  ;-)

I think the chairs were busy with other things. (I think I saw Thomas at
pretty much every session I attended...)

> I'm going to leave it there in case anyone wants
> to pick it up, but at the same time, I'd like to
> move things forward.  Thus, I'd like to truly start
> the architectural discussion at the top level and
> see where we can get to.

Sounds good to me, but maybe we should first see if we can agree this wg
is the right place to do it.

> I'd also like to make
> sure that we're on the same page as to the problem
> that we're solving.

My page would be: at least identify the places where the current
architecture doesn't fit current or reasonably expectable use of TCP/IP.
If possible, come up with an architecture that better handles these.
Then build a multihoming solution with both the new architecture and
existing implementations in mind.

> My read of the mail so far has been that we're
> agreed:

> There are many different problems that could be
> solved under the heading of multihoming, but for
> now, we should focus on site multihoming.

> Agree or disagree?

Sort of. The top of the multihoming pyramid has all kinds of additional
complexities (connecting to the net in different parts of the world, for
instance) while the ground level may have additional restrictions
(dynamic addressing, no cooperation from ISPs). If we can solve
multihoming for everything in between, I'll be very happy.

Also, the two extremes are addressed by incremental solutions in routing
(see my message from last night) and host multihoming/mobility (see
Christians remarks+draft).

Iljitsch