The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2000-Oct> msg00362



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

[IP-Optical] RE: Optical link bundling. Was Re: DraftMinutes From Pittsburgh

  • From: Anil Rao <ARao@oni.com>
  • Date: Mon, 23 Oct 2000 14:06:49 -0700
  • Cc: ip-optical@lists.bell-labs.com, mpls@UU.NET

Title: RE: [IP-Optical] RE: Optical link bundling. Was Re: DraftMinutes >From Pittsburgh

Neil/Kireeti,

> Note - to fully satisfy the example
> you gave on a
> truly global case across all customers requires that the
> world only has 1 SP
> (assuming SPs would not want to share all their duct/server
> layer routing
> with everyone else, irrespective of the scalability issues
> this implies).  I
> don't think this scenrario is likely however.
>
>       The only consistent/generic model I see working here is that a
> client makes a request of the server and the server says
> whether this can be
> honoured.

I think there is value to both the models.  There are various carriers and service providers who want to run both the client and server with a well distributed information model(in which the client understands the shortcomings of the server) and also in a client/server model where the client requests the server for services. 

The gmpls model should support both.  In one model, as Neil pointed out information will not be available for an explicit route object while in the other case information will be available for explicit route object.

-anil.