Re: multihoming (was Re: draft-ietf-nat-protocol-complications-02.txt)

[Keith Moore]

> So one of IPv6's multihoming approaches is no worse than IPv4, 
> while another appears to be significantly better. 

...in terms of its impact on the routing system.  it's not clear
that having multiple addresses per host is significantly better
for applications in general.

my guess is that some applications will do well with multiple-address
style multihoming, while others will not.

but even if multiple-address style multihoming isn't a suitable 
replacement for all applications of single-address multihoming,
it's still nice to have as an option.

Keith