A Proactive Approach to Evaluate Performance of BPEL Workflows by Queuing Theory and Stochastic Petri Net

摘要

Service oriented architecture is a novel paradigm for developing software based on composition of components or services. This innovation in the context of software engineering facilitates reusability of services according to loosely coupled relations. Meanwhile, BPEL standard accelerates composition of services. Although many efforts have been made to prove functional correctness of composed services through formal verification techniques, prediction of non-functional properties such as performance, reliability, and availability received less attention. Since BPEL service is similar to a workflow, it has been modeled with Stochastic Petri Net. We have extended SPN to fit our purpose. In this paper, we have presented an analytical modeling approach based on SPN where it uses a sequence of matrix calculations in a statistical space. The calculations are according to exponential distribution for transitions and Poisson distribution for arcs. Simulation results have validated the model for performance of a complex example. This approach can also be deployed for prediction of reliability and availability.