A multi-class probabilistic priority scheduling discipline for differentiated services networks

Abstract

Differentiated services (DiffServ) is a promising architecture for the next generation Internet due to its highly flexible, scalable, and interoperable design. In DiffServ, scheduling disciplines play an important role in achieving service differentiation. In this paper, we extend the average delay analysis of the probabilistic priority (PP) scheduling discipline proposed to the multi-class case [Proceedings of 2001 IEEE Workshop on High Performance Switching and Routing (HPSR 2001) (2001)]. The PP discipline is based on the strict priority discipline with the difference that each priority queue is assigned a parameter pi ∈ [0,1], which determines the probability that the queue is served, when the queue is polled by the server. We derive the relationship between the average queuing delay for each class and these parameters, as well as the upper and lower bounds of the average queuing delay for each class. This relationship shows that PP can provide different quality of service to different priority classes in a controllable way and is also able to provide relative and proportional DiffServ [ACM SIGMETRICS (1999); Proceedings of ACM SIGCOMM'99 (1999)]. Simulation results of multi-class PP are presented here. In addition, we implemented multi-class PP on a DiffServ testbed and experimental results from this will also be discussed. © 2002 Elsevier Science B.V. All rights reserved.