Estudo de equivalência de modelos markovianos construídos com os paradigmas redes de filas e redes de petri.

Abstract

The queueing networks performance evaluation system (SAVAD) is a software that allows to solve queueing network models that represents discrete systems, particularly computer network systems. SAVAD presents facilities to solve markovian models, i.e., models that only presents exponential probability distributions. This facilities are available in a tool denominated ALLOS, that permits to solve models by its Markov chains. This masters dissertation presents equivalence studies between models that represent markovian systems constructed as queueing network systems, by SAVAD, and models constructed as generalized stochastic Petri nets (GSPN). These studies aim the integration of GSPN to SAVAD, allowing, the addiction of qualitative analysis to the solution of this models. The qualitative analysis realized in this dissertation is directed to the observation of two important properties in the models represented in GSPN: liveness and boundedness. The equivalence studies are based in the isomorphism existent between the Markov chains obtained by models represented as queueing network systems, solved by the tool ALLOS of SAVAD, and by models represented in GSPN.