qos cbwfq on subinterfaces

[marcusbrutus]

Hi Experts,

Just wish to ask:

If a nested CBWFQs are applied to several subinterfaces under the same main interface, and the same service policy is applied to the subinterfaces referencing that actual physical bandwidth with the shape average percent 100 command, when there is actual congestion, how will the bandwidth gonna be shared by these service policies?  I mean if the main interface has 1 Mbps BW and there are two subinterfaces each is configured with the same service policy, each service policy referencing 100 percent of this 1Mbps, will each service policy be AWARE of the other service policy on the other subinterface and work concurrently?

Thanks in advance.

[CSCO11256353]

depends on the configuration.

if both are using the 100% as reference then it will be wrong as both are not aware of each other.

HTH

[John Blakley]

Marcus,

I believe it's going to end up being a first come/first served basis. The physical interface can only go so fast, and if your subinterfaces are configured to use 100%, then if subif A is using 50%, subif B only has 50% to work with. If subif A isn't using anything, then subif B will get 100%. It will be divided between the subinterfaces.

And by default, you can only use 75% of the physical interface anyway. You can change that with "max-reserved-bandwidth 100".

HTH,

John

[marcusbrutus]

Hi John,

Am curious, if subifA is using 50% of the BW and subifB is using also 50% of the BW, then both service policies on both subinterfaces do not know the link is actually already congested?  By definition, at this state, queueing will occur and it should trigger QoS right?  And if it does trigger QoS, would the service policy see the other service policy using the LLQ bucket or will it only see its own sets of logical bucket and is unaware of the other service policy's buckets?

Confusing.