Re: administrivia (on avoiding injury)

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

On Wed, 11 Apr 2001, Jim Bound wrote:

> I think so greg but if we can figure it out quickly we should. 
> 
> I am feeling like all want to ship something useful quick? I can go in
> that mode easily?  Then we work on stuff progressively?  
> 
> Is that our model?

If we are to decide that transport multihoming is the right thing, we must
do is soon there are too many architectural differences between network
level and transport level multihoming. If we allow practically any network
level multihoming we begin on the slipperly slope of deaggregation.

There is a lot of mential mass we would need to move, as it's hard to
get people to cope with IPs not being host identifiers.

We need to figure out whats required of multihoming. *Draft*
Figure out what transport can and can't do. *Draft?*
Decide if transport is best; if so: 
Figure out what architecture is needed to best capture the most benifits 
       of transport multihoming. *Draft*
Figure out what is the fastest way to make that architecture deployable.
	*draft*

The deployable solution could probably be slightly inferior to the world
of IPv4 multihoming today if we concurrently provide 'the better way'
(potentially extensive TCP options or SCTP, etc).

Most users of IPv6 will come *long* after the early adopters and *will
probably never use non-transport-multihoming enabled hosts.

I think this is a big factor, as I intend to live a long time with the
Internet that results. :)