In: Queueing and Related Models, Bhat, U.N.; Basawa, I.V. (eds.), pages 262-284. Oxford University Press, 1992.
Abstract: Analysis of queueing networks for transient measures is a difficult task. Closed-form solutions for transient measures can be derived only for very simple queues. Numerical computation of these measures, on the other hand, is comparatively easier. However, this entails generation and solution of large Markov models. In this paper we discuss an automated method for the generation and transient solution of large Markov chains. We present a new uniformization-based method that shows significant improvement in its execution time performance over uniformization, for stiff problems. We utilize these techniques for analyzing a closed queueing network model of a computer system. We illustrate the use of transient solution of Markov chains in two different contexts. The first is in computing time-dependent measures like the average queue length and system throughput. The second is in computing the steady-state response time distribution of a job