When using nested policies, the cbQoS charts are misleading because they show double the bandwidth being used. This is because the charts show the root policy as well as all the nested policies.and the bandwidth used by the root policy is a total of all the classes in the nested policy. A facility to filter out classes from the charts would be ideal. It also means that the percentage values shown for the rates are actually wrong because the root class should not be included in this calculation.
This ought be a common problem because traffic shaping is required any time the port size is less than the access circuit size. E.g. a 30Mbps port on a DS-3 access, or a 20Mbps port on 100M Ethernet access. All our WAN circuits are delivered on Ethernet therefore we use traffic shaping everywhere on the WAN. Ethernet use is also becoming much more popular in delivering WAN circuits as it is much cheaper.
Take this as an example (I've cut out all the irrelevant bandwidth/set statements etc. out of the policy-maps to make it easier to read) which we have on a 35Mbps circuit, delivered on Ethernet.
policy-map WAN-OUT
description Outbound shaping policy
class class-default
shape average 34300000 343000
service-policy WAN-OUT-Q
policy-map WAN-OUT-Q
description Outbound queuing policy.
class VOICE-Q
class VIDEO-Q
class MISSION-CRITICAL-Q
class TRANSACTIONAL-DATA-Q
class BULK-DATA-Q
class SCAVENGER-Q
class class-default
All I want to see on the charts are the contents of the nested policy, WAN-OUT-Q. I want to be able to filter out the WAN-OUT policy class-default class. I have attached a CBQoS chart that at first glance shows throughput hitting 45Mbps on a 35Mbps circuit, which is obviously misleading and also skews the percentages.
Many thanks for listening.
