Brian E Carpenter wrote:
> 
> This doesn't resolve the problem of ambiguous subnet prefixes
> when routing domains merge. So it doesn't go far enough IMHO.

That's because dealing with uniqueness and merging is a deployment issue,
not an architectural issue.  The proposal specifies what the routing system
and applications need to know about these restricted deployment addresses
for correct general operation.

Achieving the uniqueness criteria is actually rather easy.  HOW to do it is
up to the person deploying the site.  Workable solutions include generating
unique /64 subnet IDs ( draft-hinden-ipv6-global-site-local-00.txt,
draft-white-auto-subnet-00.txt ), generating a /48 based on the MAC of the
root router, or allocating a unique /48 from a registry.

Of course, if you're in the mood to shoot yourself in the foot, you could
number your site from fec0::/48.


I think we need to divide the problem space into two parts:

- Cleaning up site-local and scopes:
  - Enforcing the global unrouteability of the FEC0::/10 prefix.
  - Otherwise removing scope processing from address architecture.

- Document recommended ways of allocating FEC0::/10 addresses to
'near-enough' guarantee uniqueness.

The first change makes FEC0:://10 the same as any other filtered address,
with one little exception - everybody agrees that this is to be filtered.

A well-known non-globally-routeable prefix allows applications to make
decisions based on scope: "I'd rather use / not use filtered addresses".  As
a default, I'd recommend the rule "If I have an address in FEC0:://10 and
the destination offers FEC0:://10, try that first."  Assuming longest-match
destination address selection, this will work for both the situation where
FEC0:://10 exists and doesn't.  Of course, some applications will override
this and say "no, don't give me an FEC0:://10".  The convenient property is
that applications can care if they want to, and otherwise can remain
ignorant.

The second point deals with the ambiguity problem.

-- 
Andrew White                [EMAIL PROTECTED]
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to