Revenue maximization with quality assurance for composite web services

Service composition is one of the major approaches in service oriented architecture (SOA) based systems. Due to the inherent stochastic nature of services execution environment the issue of composite service quality assurance within SOA is a very challenging one. Such heterogeneous environment requires dynamic, run-time composition of services. We show how to determine a policy that satisfies the quality assurance for the composite service provider with the aim of revenue maximization for this provider. The quality assurance is defined as the probability that end-to-end deadline will be met, while taking into account service availability, (composite) service response-time and costs. The calculated policy is determined using dynamic programming and allows fast decision making for run-time composition. Besides, we determine the end-to-end response-time distributions resulting from determined policies. We illustrate the proposed solution with a number of experiments.