RE: A proposal for moving ahead on BC models

["Choudhury, Sanjaya" <Sanjaya.Choudhury@marconi.com>]

Hi! Few comments:

1. Personally, I think it is not a good idea to define 
   2 "default" models, just because we can't agree on 
   one. If somebody submits a third one tomorrow, are
   we going to specify 3 "default" models!!!  

2. The requirement draft calls for the specification 
   of a default model to ensure that a multi-vendor 
   DSTE LSR network operates without significant 
   interoperability problems. 

   Since the default model is not expected to be the
   "best model" (most efficient in all circumstances), 
   we can consider specifying a trivial model like MAM 
   as the default BC Model.

   When the specification of other BC Models mature, 
   providers can always demand the BC models they need 
   from the prospective vendors.

   [A possibly controversial alternative is to use the
   BC Model that has been out the longest -Not because it
   is optimal in all situations, but more vendors might
   have an implementation for them.]

Thanks,
sanjay
	
	
	

	

	

> -----Original Message-----
> From: Jim Boyle [mailto:jboyle@pdnets.com]
> Sent: Friday, December 20, 2002 9:21 AM
> To: Francois Le Faucheur (flefauch)
> Cc: Ash, Gerald R (Jerry), ALASO; te-wg@ops.ietf.org
> Subject: RE: A proposal for moving ahead on BC models
> 
> 
> 
> I don't think a lack of accurate forsight should weaken protocol 
> requirements. If in two years we realize that one (or both) are in 
> practice never used, we can update the RFC.
> 
> The real question is whether folks think there is a *need* (for 
> consistent/interoperable implementations) to require one (or two)
> BC model(s)?  Or not.
> 
> regards,
> 
> Jim
> 
> On Fri, 20 Dec 2002, Francois Le Faucheur (flefauch) wrote:
> 
> > Jerry,
> > 
> > >> Just to be clear:
> > >> 
> > >> The suggestion is to specify 2 *default* BC models in the 
> > >> proto draft 
> > >> http://ietf.org/internet-drafts/draft-ietf-tewg-diff-te-proto
> > -02.txt.  Recall that the requirements draft
> > http://ietf.org/internet-drafts/draft-ietf-tewg-diff-te-reqts-06.txt
> > calls for at least one BC model and allows for additional 
> BC models to
> > be specified later:
> > 
> > >> So the suggestion is to specify 2 default BC models (RDM 
> & MAM) in
> > the proto draft and allow for additional BC models to be 
> introduced in
> > the future (e.g., MAR, or something else).  I suggest we try to get
> > consensus on that since it will ensure interoperability in 
> the near term
> > and allow for improvements/extensions in the future. 
> > 
> > This is clear.
> > Yes, what we had specified in the Requirements document allows for
> > extension. And as I said, I could live with that. 
> > 
> > The point I have been trying to convey, though, is that if 
> in 2 years
> > time it turns out that we all realise that noone actually 
> uses RDM and
> > MAM because everybody realised that Model_xyz is much 
> better, then we
> > are left with specifications that mandate support for 
> something useless.
> > I am not saying this will happen for sure, I just don't know. 
> > Do you get my point? 
> > Do you feel confident that RDM and MAM will always be 
> required anyway?
> > Or are you saying it is no big deal if we mandate two 
> models that are
> > not the ones eventually used in the long run.
> > 
> > Thanks
> > 
> > Francois
> >  
> > 
> > 
> 
>