Performance Modeling with Deterministic and Stochastic Petri Nets
Science of Computer Programming
New York, 1998, 405 pages, ISBN 0-471-97646-6 . The book provides a thorough discussion of the deterministic stochastic Petri net formalism that is becoming increasingly important for the numerical analysis of computer=communication systems. Deterministic stochastic Petri nets (DSPNs) are presented as extensions of generalized stochastic Petri nets (GSPNs) that have received in the last 15 years considerable attention from performance evaluation professionals. DSPNs are characterized by the
... ence of immediate and timed transitions whose ÿring delays can be either constant (deterministic) or exponentially distributed random variables. Obviously DSPNs without deterministic transitions are GSPNs and this is the type of extension that is considered in the book. While GSPN models of dynamic systems are known to yield Markov chains that can be used to evaluate their behaviours, DSPNs represent a formalism for the construction of much more complex stochastic processes. Algorithms and numerical techniques for the stationary analysis of DSPNs are presented in this book when the underlying probabilistic models are either semi-Markov or Markov regenerative stochastic processes. The book is divided in four parts that address the basic formalisms used for the performance analysis of computer systems, the algorithms and numerical techniques for the solution of DSPN models, the application of the DSPN formalism for the analysis of some important case studies, and ÿnally the description of the software package DSPNexpress that the author developed during the last few years and that implements most of the algorithms discussed in the book. The book concludes with two appendices containing basic probability theory deÿnitions and properties as well as the characteristics of some common probability distributions. Part I is mainly represented by Chapter 2 that contains a fair introduction of the different modelling formalisms that are most common in the analysis of computer=communication systems. Keeping this discussion quite concise, the book avoids the overlapping with other books on the same topics and emphasizes the fact that its main objective is the discussion of the solution methods for a very speciÿc class of stochastic Petri nets.