Time Petri Nets for performance evaluation of composite web services architectures

Nowadays, QoS parameters are more and more integrated in composite web service architectures. The performance evaluation of the latter is becoming a challenging issue, as it demands to satisfy complex synchronization as well as different requirements. We investigate in this paper the use of Time Petri Nets (TPN) for modeling and analyzing the behavior of such architectures in the aim to compute the mean response time of requests and the throughput. We consider here the case of the Fork-Join architectures with a fixed number of invoked elementary services. We show how to model this architecture while assuming queue´s constraints as well as Timeouts. We simulate the obtained TPN models under different loads to compute the considered QoS parameters.