Abstract :
Slow access to files of information (directories, tables, executable code, etc.) held on secondary memory devices such as drums and disks is often the prime limitation on the response time and throughput of a computer system. Therefore, the algorithm used for scheduling data transfer requests involving these devices should be carefully considered in the design of large-scale systems. In this paper an attempt is made (by closed form analysis of mathematical models) to compare the effect of several different scheduling algorithms for drums and disks. The scheduling algorithms treated are primarily designed to improve the utilization (or throughput) of the device over that obtained by serving all requests in order of arrival (FCFS). The results are of interest but in the case of the disk analysis the comparisons are invalid because the author, unwittingly, has not derived the same parameter for each case.
