Determing Performance of Choreography-based Composite Services

Web Service Choreography Description Language is a main-stream standard for the description of peer-to-peer collaborations for the participants for service composition. To predict the performance of composite service processes gives the way to tell whether the process meet the non-functional requirements, and to choose the process with better performance from those with identical function. Unfortunately, very limited research attention is paid to performance of service choreography. In this paper, we propose a translation-based approach for performance prediction of composite service. To translate a composite service into a state-transition model for quantitative analysis, we first give a set of translation rules to map service choreography elements into general-stochastic-petri-nets (GSPN). Based on the GSPN representation, we introduce the prediction algorithm to calculate the expectedprocess-normal-completion-time. In the case study, we also obtain experimental results using the WSCDL+ execution engine and validate the accuracy of theoretical results by showing 95% confidence intervals derived from experimental results perfectly cover theoretical prediction values.