In a Video-on-Demand server, requests from different clients are independent of each other and may arrive at random time. Commercial systems may contain hundreds to thousands of clients and thus providing an individual stream for each client may require very high disk bandwidth in the server. Therefore the disk bandwidth may become a bottleneck resource, restricting the number of concurrent displays in the system. In this paper, we propose a scheme that trades memory for disk bandwidth and strikes a balance in order to prevent either memory or disk bandwidth from becoming a bottleneck resource. Moreover, we obtain the balance point of trading memory for disk bandwidth which leads to the most cost-effective system configuration.