Re: Requirements [was Re: Transport level multihoming]

[Greg Maxwell <gmaxwell@martin.fl.us>]

On Tue, 3 Apr 2001, xxvaf wrote:

> The danger is that if the population of "legacy hosts" is large enough
> (i.e. if most hosts don't support applications written to use the
> transport-layer multihoming API), then there will still be large demand
> for multhoming in the routing system. Remember, people getting connected
> to the Internet don't particularly care if their actions degrade global
> aggregation - they just want to "optimize" connectivity for their little
> bit of the world. In your draft, you may want to think about how you make
> transport-layer multihoming more attractive than current practice.

I have several ideas to address this, some of which should be possible
without significant changes to applications (obviously of the most benefit
here), and others which will require addtional awareness in the
application.

Obviously, a successful perposal would be one that provided the most
benefit for the fewest changes.

I'll be sure to address this. Thanks.