Simulation Study of Weighted Round-Robin Queueing Policy

In modern networks serving real systems, there are a wide of variety of classes of messages. This results in the need for more sophisticated queueing policies. The desired queueing disciplines are expected to be able to consider all variables which are needed to handle messages appropriately. These variables include: diierent required service times, source traac chacteristics and quality of service requirements. Many investigations of queueing disciplines have been focused on fairness, which is based on the assumption that all users (sources) have equal rights in sharing all the available resources 5, 6, 15]. For the queueing disicpines for the system which allows non-homogeneous service for diierent users, priority has also been studied 4, 7]. In this paper, we face the fact that considering only one of the above components is not appropriate for queueing disciplines of modern networks. We present the Weighted Round-Robin (WRR) queueing discipline as a suitable queueing policy in addressing this problem. We believe that the WRR policy meets the requirements better than the others and we support this claim with simulation results under a variety of traac conditions.