Modeling Performance of Distributed Programs by Stochastic Decision Free Petri Nets
Mansi Ghodsi · 2001
In this paper, we are concerned with performace modeling of synchronization delays in a distributed program that consists of a number of processes that interact via message passing only. A class of Timed Petri Nets called Stochastic Decision Free Petri Nets (SDFPN) is used to model such distributed programs with deterministic control flow. We propose an exact solution technique for this model which does not follow the usual approach of reachability analysis for Petri nets and solving global balance equations for a Markovian system. Therefore, it does not require exponential distributions and does not suffer from state space explosion. The complexity of exact solution is still exponential in terms of the number of transitions. Based on this solution, an iterative approximate algorithm is also proposed which is applicable to reasonably large models with much less complexity. Experimental results verify this claim. Keywords: performance evaluation, stochastic Petri nets, Markov model, no...