Cisco CiscoWork QoS Policy Manager 4.1.2 User Guide

Weighted fair queuing (WFQ) acknowledges and uses a packet’s priority without 
starving low-priority packets for bandwidth. Weighted fair queuing divides 
packets into two classes: interactive traffic is placed at the front of the queue to 
reduce response time; noninteractive traffic shares the remaining bandwidth 
proportionately.

Because interactive traffic is typically low-bandwidth, its higher priority does not 
starve the remaining traffic. A complex algorithm is used to determine the amount 
of bandwidth assigned to each traffic flow. Packet marking is considered when 
making this determination.

Weighted fair queuing is very efficient and requires little configuration. To 
implement weighted fair queuing, you define Weighted Fair Queuing for the 
interface. You do not need to define queuing policies because WFQ automatically 
prioritizes the packets according to their IP precedence or DSCP value. 

When you apply WFQ automatically, consider marking all traffic that enters the 
device (or mark the traffic at the point where it enters your network, to ensure that 
packets receive the service level you intend. Otherwise, the originator of the 
traffic, or another network device along the traffic’s path, determines the service 
level for the traffic.