Jitter controlled WFQ algorithm on network processors and latency constrained hardware

摘要

Systems and methods of reducing service jitter in WFQ scheduling schemes used in packet traffic management are described. Service jitter is the variance in time between when a queue should have been selected for servicing and when it was actually serviced. The service jitter is generally not a problem in lower speed applications but in a high speed implementation such as a OC192 device latency can lead to downstream service contract violations. According to the invention jitter is controlled by applying a dampening factor to a difference amount that is used by the WFQ process to adjust its timing of queue selection. The difference amount is queue-specific and is a running difference between calculated and actual queue servicing times.