The MPLS WG Archive

[Fred Baker <fred@cisco.com>]

At 07:14 PM 7/30/2001, Eric Gray wrote:
>Unless the drain rate at each of your egress points exceeds the
>capacity of your core, you don't achieve any particularly good
>guarantees if all you do is police at the edge.

the drain rate of an RPR ring is presumably the same is its internal rate - 
if all the traffic is directed to a particular system, it will drain all 
the traffic from the ring. What it does on the next interface is 
interesting from a larger perspective, but not from the perspective of the 
RPR ring.

The point here is that diffserv may be used to control the rates at which 
various RPR systems put various kinds of traffic onto the RPR ring, and RPR 
will carry it. Like diffserv in any network, if supply exceeds demand 
sometime later, we can only hope that the users' TCPs will notice that and 
slow down.