Re: WG Consensus Call: draft-swallow-gmpls-overlay-00.txt

[George Swallow <swallow@cisco.com>]

> Let's see, a carrier wants interface A.
>
> A vendor proposes to standardize another interface B which is similar, but
> doesn't quite solve all the same problems as A, and tells the carrier "If
> you don't like B you don't have to use it".
> 
> It seems to me that the carrier concern is that if interface A and interface
> B are both standardized, and if (the/some) carrier(s) want interface A and
> (many/most) vendors choose to build only interface B, then the carriers
> don't get what they want.

If the OIF UNI meets the carrier's requirements, then all the carriers
will ask for it and those who want to sell into that market will build
it.

> This fear is what makes people reluctant to progress work on an interface
> they don't feel is useful for them.

I'm getting requests from other markets (mostly research) that want
something that interfaces with GMPLS is a more direct way.  Are you
saying they can't ahve what they want?  They have to use the thing
that the OIF developed?

Note that the OIF membership represents only a small segment of the
IETF participants.  There are other people in the world!  They do have
other needs.  I want to satisfy all comers.

On the other hand they do represent a very large part (I would guess
the majority) of the market.  I do not think that the
overlay adequately addresses all of the requirements addressed by UNI
1.0.  I do not think they should or would be forced into using some
other interface.  Anyone who wants to address this segment of the
market would be foolish not to build UNI 1.0.  

I really think this is all FUD!

> The idea of standards is surely to promote the deployment of interoperable
> implementations, but part of accomplishing this is to try to limit the
> number of "standardized" solutions to the same problem.
> I think the next debate would be what we really mean by "the same" problem.
> Some arguments have appeared that this interface is directed at a
> different problem, so I have to ask whether it is different enough
> to justify standardizing a different solution.

The OIF UNI took GMPLS signalling, added some stuff, and made some
other stuff illeagal.   All of the changes were to meet carrier
requirements on how they wanted to manage and interface which was at
"arms distance".  

The overlay draft is not new protocol!  It is GMPLS signalling with a
few relatively minor changes.  I will have one protocol engine and a
few protocol knobs.  By the way we use the same protocol engine for
UNI 1.0 as well.

...George

==================================================================
George Swallow       Cisco Systems                  (978) 497-8143
                     250 Apollo Drive
                     Chelmsford, Ma 01824