Modeling and survivability analysis of service composition using Stochastic Petri Nets

In this paper, we propose a service composition model method that supports quantitative computation based on Stochastic Petri Nets (SPN). It can capture the semantics of complex service combinations and their respective specifications. In this method, services are divided into interior services and exterior services. The exterior services will be published to the users, while the interior ones do not need to be published. Six equivalent simplified theorems which can be used to simplify the complex models of interior services to simple models of exterior services are presented. They enable the minimization of the state space of the model and make quantitative computation feasible. In addition, since Grid services are always affected by all kinds of churns in actual applications, we also research survivability and its main attributes for Grid service composition. The definition and computational methods based on the model are put forward. In the end, we use the method presented above to describe and analyze an example of travel Grid services successfully.