Thanx Andrew & Gary. I think I understood how it works now..
One aside question: Is there a reason why the COS (specifically the EXP
bits) to Q mapping is not available as a knob and is hardware-coded ?? Or is
there some "hidden way" of doing that ??!
/bala
> -----Original Message-----
> From: Andrew Ramsey [mailto:akramsey@juniper.net]
> Sent: Thursday, August 09, 2001 11:36 AM
> To: Bala Subrahmanyam Venkata; juniper-nsp@puck.nether.net
> Subject: RE: Need help with MPLS configuration example
>
>
> Hi Bala,
>
> > -----Original Message-----
> > From: Bala Subrahmanyam Venkata [mailto:bsubrahm@doradosoftware.com]
> > Sent: Thursday, August 09, 2001 2:17 PM
> > To: juniper-nsp@puck.nether.net
> > Subject: RE: Need help with MPLS configuration example
> >
> >
> > Andrew-
> >
> > To interrupt your thread, if you do this (map 3 flows each
> > with different
> > TOS marking to the same LSP) then you'll end up with ONE LSP
> > that carries
> > the traffic that you have differentiated via TOS values...do
> > you want that
> > (some sort of "TOS merge") ?
>
> Why not? Remember that the LSP is an aggregate of multiple flows. The
> only thing that you should be concerned with is putting packets from the
> *same* flow into different queues.
>
> >The reason is anyway traffic in
> > one LSP can be
> > mapped to only ONE output queue (via "mpls-cos-map" ).
>
> No, it may be mapped into differnent queues (via "mpls-cos-map")
>
> > If
> > this happens then
> > the differentiation you made by setting the TOS values is
> > lost right ?
>
> No, above
>
> Andy
>
> > Since
> > all of them will be put in the same queue.....or am I missing
> > something ??
> >
> >
> > /bala
> >
> >
>
This archive was generated by hypermail 2b29 : Mon Aug 05 2002 - 10:42:36 EDT