The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2000-May> msg00376



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

MPLS Performance analysis.....

  • From: Martin Cooper <mjc@cooper.org.uk>
  • Date: Wed, 31 May 2000 17:03:47 +0100
  • Cc: mpls@UU.NET

Sean Doran <smd@ebone.net> writes:

> Several people write:
> 
> > [lots of commentary about how wonderful MPLS is at making forwarding fast]
> 
> You are analysing the performance of MPLS using the wrong metric.
> I think MPLS performs exactly as intended.
> Ipsilon basically died, and not even Nokia could revive it.
> UUNET found a new layer-2 religion, essentially killing backbone ATM.
> Many companies are distracted from actually learning how to do IP routing.
> 
> Score: MPLS 3, Rest Of World 0.

As a relative newcomer to the church of MPLS (I have read
the prolific works-in-progress of AWD at any rate...), it
seems to me that the alleged performance benefits are not
so much supposed to be found in the label-switching over
packet switching paradigm-shift per se, but in the de-
coupling of end-to-end path-selection from the extremely
dynamic hop-by-hop routing information, and the more
stable re-routing of end-to-end constraint-based paths
thus enabled (but without ATMs slow-moving standardisation
process and bandwidth-overhead related baggage).

On the other hand, destination-prefix based IP routing
is quintessentially a hop-by-hop algorithm that doesn't
concern itself with end-to-end constraint-based path-
selection bar a few TOS bits, so maybe you care not
about CBPS?

M.