An analysis of time-sharing computer systems using Markov models

Abstract

The development of RQA (Recursive Queue Analyzer), a program for the numerical solution of the stationary distribution of large scale Markov processes, has made possible the accurate analysis of large stochastic systems with modest computational costs. In particular, time-shared computer systems with their random program and user characteristics are examples of systems which can be modeled as multidimensional Markov queueing processes and analyzed by the method. Having obtained a solution for the limiting state probabilities of the model using RQA, one can readily derive many time average performance and usage characteristics. Thus a useful tool is available to provide guides in the design and modification of such systems and to forecast user response and system capacity in terms of the number of users and the operating statistics.