Technical Documentation

Excess Bandwidth Distribution on the Trio DPC and MPC/MIC interfaces Overview

Service providers often used tiered services that must provide bandwidth for excess traffic as traffic patterns vary. By default, excess bandwidth between a configured guaranteed rate and shaping rate is shared equally among all queues, which might not be optimal for all subscribers to a service.

You must still apply the profile to the interface

The following statements establish a traffic control profile with a shaping rate of 80 Mbps and an excess rate of 100 percent.

[edit class-of-service traffic-control-profiles]tcp-example-excess {shaping-rate 80m;excess-rate percent 100;}

You can control the distribution of this excess bandwidth for a user with the excess-rate and shaping-rate statements. To configure the excess rate and shaping rate for a traffic control profile, include the excess-rate and shaping-rate statements at the [edit class-of-service traffic-control-profiles tcp-name] hierarchy level and then apply the traffic control profile at the [edit class-of-service interfaces] hierarchy level.

You can also control the distribution of this excess bandwidth for a queue at the scheduler level. To configure the excess rate for a queue, include the excess-rate and (optionally) excess-priority statements at the [edit class-of-service scheduler scheduler-name] hierarchy level.

You must still configure a scheduler map

The following statements establish a scheduler with an excess rate of 5 percent and a low priority for excess traffic.

[edit class-of-service scheduler]example-scheduler {excess-priority low;excess-rate percent 5;}

For queues, you cannot configure the excess rate or excess priority in these two cases:

  • When the transmit-rate exact statement is configured. In this case, the shaping rate is equal to the transmit rate and the queue does not operate in the excess region.
  • When the scheduling priority is configured as strict-high. In this case, the queue gets all available bandwidth and never operates in the excess region.

Also, for interface sets and logical interface units, the excess rate configuration will not be allowed in PIR mode. This is because the excess rate distributes excess bandwidth once the scheduling nodes reach their guaranteed rate. However, in PIR mode, none of the scheduling nodes have guaranteed rates configured (so allowing excess rate configuration makes no sense).

For more information on hierarchical scheduling and operational modes, see Configuring Hierarchical Schedulers for CoS.


Published: 2010-07-16

Help
|
My Account
|
Log Out