Reliability Prediction and Sensitivity Analysis of Web Services Composition

Web services are emerging as a major technology for deploying automated interactions between distributed and heterogeneous applications. It aims at the transparent integration of Web applications, based on XML-related standards (F.Curbera et al., 2002). Until now, many research efforts have been made in the field of Web services composition. Moreover, many composition languages have recently emerged, including BPEL, BPML or ebXML, these languages focus on tracking and executing collaborative business processes by business applications. An important issue for business process built in this way is how to assess the degree of trustworthiness, especially their performance and dependability characteristics. In this paper we focus on reliability aspects, and propose an approach to predict the reliability of web services composition. Stochastic Petri Nets (SPNs) can be used to specify the problem in a concise fashion and the underlying Markov chain can then be generated automatically. In this paper, we propose the usage of CSPN model, an extension of stochastic Petri nets as a solution to the problems of predicting the reliability of web service composition. The choice of Petri nets was motivated by the following reasons: (a) Petri nets are a graphic notation with formal semantics, (b) the state of a Petri net can be modelled explicitly, (c) the availability of many analysis techniques for Petri nets. The remainder of this paper is organized as follows. Section 2 provides general information about BPEL and stochastic Petri net. In Section 3 we describe our reliability prediction model and propose an approach to transform BPEL process into CSPN model. Section 4 discusses the result of this mapping on an example BPEL process models. Next, Section 5 discusses the sensitivity analysis of the reliability prediction model. Finally, we discuss the related works and conclude this paper.