Probabilistic Approach to Service Commitment in Service-Oriented Systems

In this paper, we consider the problem of committing service instances to the applications in a service-oriented environment. We propose a heuristic algorithm which is able to analytically compute the over-commitment and application failure probabilities, and consequently, commit required service instances to each application instance while controlling these probabilities. Moreover, this proposed algorithm is suitable for general distribution functions for the services execution times and applications interarrival times. We also formulate a linear programming problem for maximizing the overall system reward, and through simulations and performance comparisons we show that the proposed mechanisms significantly improve the system's overall performance while achieving the target level for the over-commitment and application failure probabilities.