The MPLS WG Archive

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

Title: RE: MPLS Support of E-LSPs

 

         Hi! I agree that this draft brings a much needed clarity to the E-LSP support (

        specifically from a CAC point of view). However, I differ from your suggestion

        to replace the PSC field with the Class-Type field. Here are the reasons:

 

            1. Platforms that do not support the class-type extensions proposed

                in the diff-te-reqts-00.txt draft, can still take advantage of this

                draft.

 

                Platforms that do implement the class-type extensions, can still

                determine the class-type from the signaled PSC,

 

            2. In my opinion Class-Type concept proposed in the diff-te-reqts-00.txt,

                is a good optimization of the IGPs, but not very useful for signaled

                LSPs. Since PSC-->Class-Type information is locally configured anyway,

                one can easily determine the class-type from the associated PSC.

 

                Infact, in my opinion the idea proposed in this draft, is a cleaner

                approach to signal bandwidth for OAs, than the one specified in the

                draft-ietf-mpls-diff-te-ext-01.txt.

 

         Thanks,

         sanjay

-----Original Message-----
From: Dimitry Haskin [mailto:dhaskin@axiowave.com]
Sent: Monday, April 30, 2001 3:51 PM
To: 'Sudhakar Ganti'; mpls@UU.NET
Cc: Nabil Seddigh; Biswajit Nandy
Subject: RE: MPLS Support of E-LSPs

Sudhakar,

Thanks for the draft. The proposed extensions will provide a needed signaling support for the Diff-Serv's Uniform Tunneling model described in draft-ietf-mpls-diff-ext-09.txt.

To make your proposal consistent with the Class-Type concept of draft-ietf-mpls-diff-te-reqts-00.txt, I would suggest to replace the PSC field of the proposed Traffic Profile with a 4-bit Class-Type (CT) field. In addition if your allow the ELSP Object to carry resource requirements for L-LSP also (you'd need to rename the object) there would be no need for the CLASSTYPE Object defined in draft-ietf-mpls-diff-te-ext-01.txt.

Dimitry

-----Original Message-----
From: Sudhakar Ganti [mailto:sganti@tropicnetworks.com]
Sent: Monday, April 30, 2001 12:39 PM
To: mpls@UU.NET
Cc: Nabil Seddigh; Biswajit Nandy
Subject: MPLS Support of E-LSPs

Hello,

The following draft is available in the IETF draft
repository that extends E-LSP signaling:

Title: MPLS Support of Differentiated Services using E-LSP
Authors: S.Ganti, N.Seddigh, B.Nandy
URL:
http://www.ietf.org/internet-drafts/draft-ganti-mpls-diffserv-elsp-00.txt

The current E-LSP approach allows multiple Ordered
Aggregates  (OAs)  to  be carried  by  a  single
LSP. The  existing  E-LSP specification only supports
collective bandwidth signalling for all the OAs in
an E-LSP. However, there exist uses of the E-LSP
approach where it is desirable to support bandwidth
signalling on a per-OA or per-PSC  basis.
This draft extends the original E-LSP approach by
defining extensions to support per-OA signalling.
The extensions are proposed  for  both RSVP and CR-LDP.

Please comment on the same. Thanks

-Sudhakar
------------
Dr. Sudhakar Ganti
Technical Specialist
Tropic Networks Inc,
Kanata, Ontario, Canada