Traditionally, SonicOS bandwidth management distributes traffic to 8 queues based on the priority of the traffic class of the packets. These 8 queues operate with strict priority queuing. Packets with the highest priority are always transmitted first.
Strict priority queuing can cause high priority traffic to monopolize all of the available bandwidth on an interface, and low priority traffic will consequently be stuck in its queue indefinitely. Under strict priority queuing, the scheduler always gives precedence to higher priority queues. This can result in bandwidth starvation to lower priority queues.
Weighted Fair queuing (WFQ) alleviates the problem of bandwidth starvation by servicing packets from each queue in a round robin manner, so that all queues are serviced fairly within a given time interval. High priority queues get more service and lower priority queues get less service. No queue gets all the service because of its high priority, and no queue is left unserviced because of its low priority.
For example, Traffic Class A is configured as Priority 1 with a maximum bandwidth of 400 kbps. Traffic Class B is configured as Priority 3 with a maximum bandwidth of 600 kbps. Both traffic classes are queued to an interface that has a maximum bandwidth of only 500kbps. Both queues will be serviced based on their priority in a round robin manner. So, both queues will be serviced, but Traffic Class A will be transmitted faster than Traffic Class B.
The following table shows the shaped bandwidth for each consecutive sampling interval: