The MPLS WG Archive

[Shahram Davari <Shahram_Davari@pmc-sierra.com>]

Hi,

> -----Original Message-----
> From: Satinder101@aol.com [mailto:Satinder101@aol.com]
> Sent: Tuesday, July 31, 2001 2:58 PM
> To: flefauch@cisco.com; davari@ieee.org
> Cc: mpls@UU.NET
> Subject: question on draft-ietf-mpls-diff-09.txt
> 
> 
> Francois et-all,
> 
> I have a couple of questions on 
> draft-ietf-mpls-diff-09.txt and would appreciate
> if you can clarify for me.
> 
> 1. The DIFFSERV Object (say in RSVP) has EXP-PHB
>    mappings. So I can see that populating the
>    'Encaps-->PHB mapping' for E-LSP in Sec 3.2
>    is straightforward. But I am not sure how  to
>    create the 'Set of PHB-->Encaps mapping' in
>    the Diffserv Context (sec 3.4.1) from the signalled
>    EXP-PHB map. Is this just a simple reverse mapping ?

Yes.

> 
> 2. According to Sec 5.3, each intermediate LSR just
>    records the DIFFSERV Object. So for a given E-LSP,
>    the incomming and outgoung mappingd are the same ?

Yes.

>    If that is so, how can the outgoing EXP on a given 
>    LSR be different from the incomming EXP on the 
>    same LSR.  Is this because we may have changed
>    the drop precedence,  so the PHB used
>    to compute the outgoing EXP is is different from
>    the PHB we got when we mapped the incomming EXP
>    to PHB.

Yes. This is the case for example when you do policing.

> 
> 3. It looks to me that in case of RSVP, the EXP-PHB
>    mapping is selected by the box initiating the 
>    PATH message (which will be the upstream node
>    w.r.t data flow). Whereas in case of LDP 
>    (downstream unsolicited) this mapping is selected
>    by the downstream node. Is my understanding corect ?

Yes.

-Shahram

> 
> 
> thank you very much
> 
> - satinder
>