A Petri net-based modeling technique to address the gap in software requirements modeling methodologies for distributed systems is addressed in this paper. Its use is for modeling systems composed of concurrently active components that cooperate and share work responsibilities. It extends the authors’ previous work on stochastic process algebra (performance evaluation process algebra (PEPA)), and integrates it within a Petri net-based technique for performance modeling. It is true that traditional modeling and analysis tools for software design and development do not adequately address all of the issues of time, synchronism, and concurrency for distributed systems. This paper illustrates the application of the authors’ methodology by using it to model mobile agents and a secure Web service.

The paper describes the implementation of a translational process from PEPA nets to PEPA algebra for analysis of the PEPA Petri net model. The tool is available for download from http://www.dcs.ed.ac.uk/pepa/, and is implemented in ML and Java. It contains a complete modeling kit for PEPA, including solvers, time analyzers, model-checkers, and so on. The Web site indicates that the ML edition of the PEPA workbench transforms PEPA descriptions into a form suitable for solution by another solution tool, such as Maple, Matlab, or Mathematica. The Java edition of the PEPA workbench can solve models without the need for a separate solution tool. The PEPA workbench has Linux, Solaris, and Windows distributions and examples, to illustrate using PEPA for the generation of systems with very large state spaces; for example, a PEPA Web service models 26,651,520 states.

The work presented here will be very useful for graduate students working in the areas of software modeling, performance analysis, distributed systems, and mobile computing.