Abstract

Traditional admission control algorithms for on-demand multimedia servers concern the acceptance decisions for new clients' requests so as to guarantee that continuous services to all clients are executed. These algorithms determine whether a new client can be accepted, based on the consideration whether the underlying hardware can satisfy the quality of service (QoS) requirements of admitted client requests. In this paper, we consider a richer class of admission control algorithms that make acceptance/rejection decisions not only to satisfy the hardware requirements of client requests but also to optimize the reward of the system based on a performance criterion as it services clients of different priority classes. We divide the server capacity into a number of ‘priority threshold values’ based on which the system decides whether to accept clients of different priority classes dynamically in order to maximize the system value. The resulting threshold-based admission control algorithm is developed based on the idea that admission control can be driven not only by hardware requirements, but also by knowledge regarding the workload characteristics of client requests, thus allowing the system to adjust dynamically the threshold values in response to changes in client workload characteristics. We derive a close-form expression for the value which the system can obtain when operating under the threshold-based algorithm as a function of model parameters, and discuss how the server can utilize the analytical solution at run time so as to maximize the system value dynamically without violating clients' continuity requirements.

You do not currently have access to this article.