Re: MPLS Inter-area TE requirement draft

[Jim Boyle <jboyle@pdnets.com>]

On Sun, 4 Jan 2004, Jean Philippe Vasseur wrote:
<snip>
> 
> On the other hand, the goal is to be able to compute an optimal path where 
> optimal means "shortest path", which would be as optimal as the path 
> computed by means of CSPF with a single area.
> 
> JP.

Well, this is my concern, that goal is very tangible, however it seems
in my mind at least to potentially conflict with something which I
feel should not be compromised, scalability.  However, scalability is
subjective - thus more easily bartered away.

Revisiting your draft, it seems we may actually be in strong
agreement.

In your draft, you stated that scalability is a hard requirements
(e.g. "MUST" in section 5.1 - as you pointed out in email) but path
optimality as defined above is a softer requirement (e.g. "SHOULD" in
5.4).  You can only deliver a SHOULD if it doesn't negate a MUST,
right?

I think it would be ok to allow for what is requested in section 5.4
but only if it can be done without impact on scalability, and that
should be clear in the requirements.  Scalability is of utmost
importance.

Are we in agreement on this point then?

regards,

Jim

(btw, in section 5.1, I think you are missing a "not" after the first
MUST in the 3rd sentence :)