Re: I-D ACTION:draft-ietf-shim6-applicability-01.txt

[Daniel Roesen <dr@cluenet.de>]

On Wed, Jun 14, 2006 at 10:30:13AM +0200, Brian E Carpenter wrote:
> >Could you please explain the "site model" of shim6 you are referring to?
> 
> It's pretty clear - if a site has N ISPs it will have N prefixes,
> so each host in the site will have M(<=N) addresses, one under each
> prefix, and those M addresses will form the locator set for
> shim6. None of the addresses in the locator set can be under
> a prefix that isn't one of the N.

And now replace "site" with "host-attached LAN(s)". Where's the "site"
factor gone now?

shim6 is host multihoming. Only shim6ifying each and every device in the
site makes the site multihomed, but only implicitly. And even that only
under the assumption, that all remote communication partners do also
support shim6, otherwise the multihoming is impaired.

Calling a brick a "house building solution" just because it can be
part of a wall is slightly over the top...

But I understand that I won't gain any support for this view here, the
discussion seems to lead nowhere, so I'll rest my case.


Regards,
Daniel

-- 
CLUE-RIPE -- Jabber: dr@cluenet.de -- dr@IRCnet -- PGP: 0xA85C8AA0