The MPLS WG Archive

[David Charlap <david.charlap@marconi.com>]

Lou Berger wrote:
> 
>          Here's my (limited) understanding of the issue.  A while back
> two separate ways of specifying BE/CoS service made it into drafts.
> One of which was the CoS object in the rsvp-lsp draft.  The other,
> while I think came later in time, made it through the IETF process
> first and has been standardized in RFCs2997 and 2996.  This result was
> that the CoS object served the same purpose of an existing standards
> track RFC.   Having two ways to signal the same thing really doesn't
> make sense, and since they got there first, we loose.  While the
> decision is painful now, I think it's better than the alternative.
> Having some more words on this in draft certainly makes sense.  (Eric
> G. made this point privately.)

Thanks.  RFC 2997 allows RSVP to signal best-effor without using
DiffServ extensions.  Replacing support for COS with Null-IntServ will
be a bit annoying in places, but not nearly as difficult as implementing
the diff-ext draft.

Which working group developed 2997?  I don't remember seeing anything
about it here or on the RSVP list.  (Maybe it was on the IntServ list? 
I'm not subscribed to that one.)

With respect to the MPLS drafts (RSVP-TE and diff-ext), these RFCs
should definitely be referenced, since they appear to have become hard
requirements.

-- David