The application of FSP models in automatic optimization of software deployment

Abstract

The correct deployment of software objects over the computational resources has a significant impact on the software performance. Achieving the optimal deployment manually is a tedious work as there are many different alternative solutions. In this paper a heuristic algorithm for optimizing the deployment of software objects is proposed which evaluates each deployment in the search space, considering its communicational and computational delays. In order to estimate these delays for an object deployment, our algorithm takes into account both the resource capacities and the execution load of the software for a given input-workload. The execution load of the software is measured by simulating the software use-case scenarios using the Finite State Process (FSP) models. From the simulation, the values of some metrics such as utilization, population and mean response times corresponding to the objects and threads participating in software use-case scenarios are recorded as the execution load indicators. These recorded simulation results are subsequently applied to estimate the goodness of a deployment in the search space. © 2011 Springer-Verlag.