L2VPN QoS

I'm just playing and trying to understand ethernet-to-ethernet QoS in a service provider environment, be it from a dot1q frame using CoS or a dot1q-tunnel again using CoS to using CoS in the MPLS header ideally I just want what ever extended ping from the customer devices to copy the IPP over to the dot1q CoS then over again to CoS in the MPLS header

I have tried to lab this on GNS using 7200's running 15.2.4 and it seems I cannot map dynamic CoS to EXP bits on label imposition from the interface facing the CE routers, I can only map the topmost label on imposition to a specific CoS value in the MPLS header therefore this isn't dynamic in nature

I am able to set a policy from CE's to PE's to copy the IPP value over to the CoS in the dot1q frames however

CE1 dot1q ----- (targetted ldp to PE2) PE1 ------ P ------ PE2 (targetted ldp to PE1) ------ dot1q CE2

My second question is if the customer wanted a 2mbps VPN then I can shape 2000kbps outbound and police inbound to 2000kbps facing the CE devices or ask the customer to use shaping outbound in place of the PE policing inbound, is their an RSVP or VFI QoS applied in provider environments, essentially how do I go about this and what is common practice

--
BR
Tony
Sent from my iPad
Blogs and organic groups at http://www.ccie.net