RE: Fwd: [ipv6mib] So, where were we?

["C. M. Heard" <heard@pobox.com>]

>>>>> Margaret Wasserman wrote:
> [C. M. Heard wrote:]
> >As it stands the above is insufficient, even for a minimalist scheme,
> >because a given prefix may be reachable via multiple next hops.
> 
> Right.  And, different next hops may be used for traffic that has
> different diffserv codepoints.

>>>>> On Wed, 9 Oct 2002, Dave Thaler replied:
> My understanding is that TOS-based next-hops are deprecated,
> and that different diffserv codepoints should never change
> the routing, only the way it's handled by boxes on the path.

Certainly TOS-based next hops using the interpretation of the TOS
field published in RFC 1349 are deprecated;  there were never two
independent interoperable implementation of TOS-based routing, so
that part of OSPF was removed from the specification upon going to
full standard, and RFC 1439 was reclassified as Historic.  Going
forward there is certainly no need to keep RFC 2096's ipCidrRouteTos.
It is always zero in most implementations anyway.

However, I don't recall that anything published by the diffserv WG
ever said one way or another whether the DS codepoint could affect
the route.  Do you have a citation to the contrary?  It probably
does not matter much either way, because even if such a mechanism
exists it is not standardized in a routing protocol and would
probably be proprietary if it existed at all.

//cmh