[c-nsp] ASR1K - Aggregate QoS Across subinterfaces?

[Brian Turnbow]

Hi Mitch,

If I understand what you are asking aggregate class may but what you are looking for.
The keyword  is fragment in the qos docs

Create aggregate shaper

policy-map aggregate-member-link    
  class BestEffort-class service-fragment BestEffort-fragment
  shape average 20000000

and associate your parent class like this with your child class for 

policy-map parent
  class class-default fragment BestEffort-fragment
    shape average 5000000  
service-policy child

It's got some limitations but may be what you need.

Brian


Brian Turnbow
Network Manager 
TWT S.p.A.

> -----Original Message-----
> From: cisco-nsp [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of
> Mitch Dyer
> Sent: giovedì 9 luglio 2015 02:51
> To: cisco-nsp at puck.nether.net
> Subject: [c-nsp] ASR1K - Aggregate QoS Across subinterfaces?
> 
> Hello Everyone,
> 
> I apologize if this post is either off-topic or inappropriate, first time posting.
> 
> We have a 1Gb circuit from a local carrier that aggregates circuits from
> several of our customers. Each customer gets handed to us with a separate
> dot1q tag. I'm looking to queue/police that traffic but am having trouble
> coming up with the appropriate strategy to do so across the two services we
> are looking to provide. The first service can be described as transit at a CIR
> while preserving voice markings to an upstream SIP provider that we peer
> with. The second service leverages MPLS L3VPNs to provide access to an IaaS
> service we provide, along with some other centralized services we host
> and/or peer with.
> 
> We're terminating this aggregated circuit on an ASR1002X. The transit
> customers seem pretty straight forward, at each sub interface have a parent
> policy that polices/shapes to the agreed upon CIR and then a child policy
> which handles individual classes of traffic. The L3VPN service seems to be a
> bit trickier as it's been sold as an aggregated service and doesn't have a
> specific CIR associated with each customer.
> 
> Is there a way to shape/queue the L3VPN customers as an aggregate of
> whatever resources are available?
> 
> A colleague suggested H-QoS but I've had a hard time distinguishing
> between it and the MQC documentation for the ASR1000 series.
> 
> Any help would be greatly appreciated.
> 
> Thanks,
> Mitch
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/