Publication

Abstract : Future advances in networking coupled with the rapid advances in storage technologies will make it feasible to build a multimedia on-demand server on a metropolitan-area network. However, the economic viability of such a multimedia on-demand server is dependent on the ability to amortize its operational costs over a large number of clients. In this paper, we present a taxonomy of policies for servicing multiple clients, and propose round robin and quality proportional admission control algorithms for determining whether a retrieval request from a new client can be admitted without violating the real-time requirements of any of the clients already being serviced. We evaluate the performance of various servicing policies, and show that they are an order of magnitude scalable compared to straightforward multiplexing techniques such as servicing one subscriber per disk head.