WFQ

Weighted Fair Queuing (WFQ) is a packet scheduling algorithm used in packet-switched networks to allocate outgoing link bandwidth among active flows according to assigned weights. It provides a practical approximation of Generalized Processor Sharing (GPS), aiming to give each active flow a proportional share of the bandwidth while preserving some level of isolation between flows.

In WFQ, each active flow is assigned a virtual time and a finish time for each packet, based on the packet’s size and the flow’s weight. A packet’s virtual finish time is computed from the flow’s last finish time and the packet length divided by the flow’s weight. Packets are transmitted in the order of increasing virtual finish times across all flows. This scheduling yields a fair distribution of bandwidth proportional to weights and tends to maintain per-flow service guarantees, even when traffic mixes are diverse.

Implementation often involves per-flow queues or a mechanism that tracks per-flow state to determine the next packet to send. Weights reflect the relative importance or service level agreements (SLAs) for different flows or classes. In practice, network devices may implement WFQ directly or as part of class-based or differentiated services QoS schemes (for example, class-based weighted fair queuing) to support QoS requirements in heterogeneous traffic environments.

Limitations include the need to maintain state for each active flow, which can impact scalability on very high-speed links. Fairness depends on accurate weight configuration and traffic behavior; highly bursty or adversarial traffic can complicate scheduling. WFQ remains a foundational concept for QoS-aware scheduling in many routers and switches.