R3303-HP HSR6800 Routers ACL and QoS Configuration Guide

WFQ queuing

Figure 27 WFQ queuing

WFQ is similar to WRR. They both support scheduling weights in queue length, and can work with SP

scheduling together. The difference is that WRR enables you to set the maximum time a packet waits in

queue, but WFQ enables you to set guaranteed bandwidth a WFQ queue can get during congestion.

CBQ

CBQ provides one FIFO queue for each user-defined class to buffer traffic of the class. When the network

is congested, CBQ classifies packets into user-defined classes, and assigns different classes of packets to

different queues after performing congestion avoidance and bandwidth restriction check. When

dequeuing packets, CBQ schedules packets from queues in proportion to their weights.

CBQ provides the following queuing types:

• Low latency queuing (LLQ)—LLQ queues are EF queues, and ensure strict priority service for

real-time traffic. CBQ always schedules traffic in LLQ queues preferentially. To guarantee that other

queues can get served when congestion occurs, you can set the maximum bandwidth for each LLQ

queue. In normal cases, an LLQ queue can use more bandwidth than allocated. When congestion

occurs, the exceeding traffic is dropped. You can also configure a burst size for LLQ queues.

• Bandwidth queuing (BQ)—BQ queues are AF queues. BQ provides strict, exact, guaranteed

bandwidth for AF traffic, and schedules the AF classes proportionally.

• WFQ—One WFQ queue is available for BE traffic, and uses the remaining bandwidth to send the

BE traffic.

BQ and WFQ use tail drop by default. You can configure a WRED drop policy to limit traffic.

Hardware congestion management configuration

approaches

To manage hardware congestion, you can do the following:

• Configure queue scheduling for each queue in interface view or port group view, as described

in Configuring per-queue hardwar

e congestion management.