RE: [PWE3] Inter-op is not necessary

[Sasha Vainshtein <Sasha at AXERRA.com>]

Danny (and all),
Following your suggestion, I would like to sound my "preferences and
opinions" 
regarding protocols for setup of MS-PW on the list.

1. I strongly support the LDP-based approach to setup and maintenance of
MS-PWs
   for the following reasons:
   - it meets the requirements of specific MS-PW applications I am primarily

     dealing with (single- or, at most, dual-homed access devices acting as
T-PEs)
     without adding excessive complexity 
   - it simplifies backward compatibility with the already deployed SS-PWs
     (including interworking from FEC 128 to FEC 129). I agree with the 
     many people on this list who see backward compatibility with SS-PWs as
     a MUST requirement and suggest explicitly adding it to the MS-PW
     requirements draft (draft-ietf-pwe3-ms-pw-requirements-01.txt)
   - it also simplifies backward compatibility with statically stitched
MS-PWs
     (as per draft-ietf-pwe3-segmented-pw-01.txt).

   In other words, I see a well-defined and substantial application space
where
   MS-PWs would benefit from the LDP-based approach. Based on that, I
support
   adopting draft-balus-bocci-martini-dyn-ms-pwe3-00.txt as a WG document
NOW.

2. I consider existence of a valid application space for the RSVP-TE-based
approach 
   that is not covered by the LDP-based approach as neither proved nor
disproved at 
   this stage. Hence this approach should be, IMO, given "the benefit 
   of doubt". This said, the "burden of proof" (of existence of a distinct
meaningful 
   application state for this approach lies with its proponents and, IMHO,
includes: 
   - an expanded definition of the relevant application space with explicit
specification
     of problems that do not admit a solution with the LDP-based approach
   - a clear demonstration of these problems being addressed by the
RSVP-TE-based approach 
    (a list of references to multiple MPLS and CCAMP WG drafts would not be
enough)
   - specification of a reasonable "interworking" process between the two
protocols
     (because, IMO, the LDP-based approach will stay with us anyway),

   I see resolution of the issues listed above as A GATE for advancing the
work on 
   the RSVP-TE-based approach in the PWE3 WG.

3. I do not support attempts to resuscitate CR-LDP for setup 
   and maintenance of MS-PWs as an alternative to the RSVP-TE-based approach
   for the following reasons:
   - After being "suspended", CR-LDP has missed quite a few features that
     have been or developed or are being developed and deployed for RSVP-TE
   - AFAIK, there is no deployment experience with CR-LDP
   - CR-LDP has been only designed to operate in the DoD mode (see RFC 3212,
     Section 3 "Solution Overview") while all the LDP-based proposals for
     setup and maintenance of PWs require LDP operation in the DU mode.
     IMO this means that the backward compatibility problem between 
     the deployed SS-PW setup and maintenance mechanism and a CR-LDP-based
approach
     would not differ that much from that for an RSVP-TE-based approach.
   This does not mean that we should not use some applicable TLVs defined
   for CR-LDP, e.g., ER etc. But these TLVs are not the key protocol
element.

Hopefully these notes clarify my position.

Regards,
				Sasha

> -----Original Message-----
> From: Danny McPherson [mailto:danny at tcb.net]
> Sent: Saturday, November 19, 2005 12:34 AM
> To: pwe3
> Subject: Re: [PWE3] Inter-op is not necessary
> 
> 
> 
> Eric et al.,
> This is one of the primary reasons why true consensus should
> only be obtained via the mailing list - things are much easier
> qualified, quantified, and .. [re]verified.
> 
> Folks, I suggest we let what we each recollect pass and if you've
> not voiced your preference/opinions on the mailing list please
> consider this the place which you need to be heard.
> 
> Thanks!
> 
> -danny
> 
> 
> On Nov 18, 2005, at 10:28 AM, Gray, Eric wrote:
> 
> > Stewart,
> >
> > 	Your version below is pretty much aligned with mine except
> > for two points:
> >
> > 1) "significant difference" is not quantified (or quantifiable);
> >    a good many people raised hands in both of the two latter cases
> >    (more than 20 - possibly more than 30 - in both cases), so a
> >    "significant difference" might be as little as 25% in actual
> >    difference in numbers.  If we're going to treat this as a vote,
> >    then we need to count hands.
> > 2) you also asked if any one objected in both of the latter cases
> >    and no one objected in one case (MS-LDP) and a few did in the
> >    other case (MS-RSVP).
> >
> > I am certain that this is what Dimitri is referring to when he
> > said that "a few" was associated with a different question than
> > as is reported in the minutes.
> >
> > 	Also, I do not see how your recollection is different from
> > (or inconsistent with) John's interpretation.  The first question
> > (in your list of three below) was answered in a way that implies
> > a clear majority wanting to advance both.
> 
> _______________________________________________
> pwe3 mailing list
> pwe3 at ietf.org
> https://www1.ietf.org/mailman/listinfo/pwe3
> 

_______________________________________________
pwe3 mailing list
pwe3 at ietf.org
https://www1.ietf.org/mailman/listinfo/pwe3