Re: Transport level multihoming

[Brian E Carpenter <brian@hursley.ibm.com>]

Greg Maxwell wrote:
> 
> On Tue, 3 Apr 2001, Brian E Carpenter wrote:
> 
> > Greg,
> >
> > Regardless of whether transport level multihoming can be
> > achieved (which is a long running debate, and I agree
> > that SCTP makes it a bit more plausible), the operational
> > requirement for IP level multihoming isn't going to go away
> > at least for the next 10 or 15 years, so we have to solve it.
> 
> Bah.
> 
> I don't agree.
> 
> It is totally feasible to implement a self-multihomed-transport-protocol
> with the exactly the same level of API change as is required for IPv6
> support. 

Er, the IPv6 socket API changes are fully defined and largely implemented,
and people are already adapting their middleware accordingly. And
they tell me the changes are annoying and fiddly, but not fundamental.
And the TCP and UDP applications are *not* going away for the next
10 or 15 years however much you might wish them to.

> All the needs to be changes is the paramaters for settings
> what addresses to bind to, beyond that, configuration knobs like
> prefered address are optional (the default should be decided via RTT
> or TTL, etc).
> 
> SCTP may not currenly offer this level of API change simplicity, but it is
> certantly possible.

It would be interesting to see a description of an API for SCTP that
offers TCP and UDP emulation to the apps, accompanied by coexistence and
transition mechanisms equivalent to those produced by ngtrans for the
network layer.

> I think there is a lot of potential common-intrest here, it would be
> foolish to ignore it.

True, but like it or not this will not take away the need for IP level
multihoming. In fact I think transport layer multihoming would be better
served by a separate discussion list.

   Brian