Hi All,

I would like to discuss QoS configuration scenario over an MPLS backbone (Frame Mode MPLS over ATM Internetworking backbone using BPXs) for a provider to provide the MPLS VPN customer an appropriate QoS for their Voice traffic, my initial design is as follows:

CE-1 <-> PE-1 <-> P-1 <-> P-2 <-> PE-2 <-> CE-2

CE-1/CE-2:

----------

Classification and Marking.(Egress)

LFI (MLPPP over FR) (Egress)

WRED (For class-default) (Egress)

FRTS (Egress)

LLQ (Egress)

PE-1/PE-1:

----------

Remarking with MPLS Exp bits (Ingress from CE-1 to be used Egress to P-1, and then further on P-routers)

LLQ (on the egress facing CE-1 and P-1)

LFI (on the egress facing CE)

WRED (for class-default)

Policing (not CB-Policing, we shall rely on the ATM-PVC paramters)

P-1/P-2:

-------

LLQ (Egress facing both PE and P)

WRED (class-default)

First of all, i need comments on this design. Second i've a couple of questions:

1. since the P router won't be aware about either the DSCP or the IPP how will WRED work on it ?

2. How will i configure the LLQ between the PE(s) and the P(s), will i have to calculate the aggregate estimate required bw and apply it in the LLQ ?

thanks in advance,

Mohammed.