Network Computingwww.networkcomputing.com

Priority Queuing (PQ)

Packets are funneled into queues depending on their priority. Higher-priority queues transmit packets before lower-priority queues. Low-priority traffic can become starved.

Class-Based Queuing (CBQ)

Designed to prevent the starvation present in priority queuing, CBQ funnels packets into queues dependent upon configured parameters. All queues are guaranteed a set minimum amount of bandwidth. Classes can be configured to "borrow" bandwidth if it is available and traffic demands would put the total amount of the classes traffic over the configured limit. This is often referred to as the "burst" amount for a class of traffic.

Weighted Fair Queuing (WFQ)

WFQ assigns weights to higher priorities by increasing the queue sizes. Real-time bandwidth utilization is not taken into consideration.

Hierarchical Weighted Fair Queuing (HWFQ)

A variation of WFQ, this technique uses as its evaluation measure the worst-case packet delay under various traffic scenarios. It evaluates packet delay based on dynamic, real-time traffic.

TCP Rate Shaping (TRS)

This technique regulates traffic by intercepting acknowledges and changing TCP window sizes.

TRS does not address the needs of other protocols such as UDP. To achieve traffic management for traffic other than TCP, vendors implement a separate queuing facility often utilizing PQ or WFQ.