Constraint Satisfaction Approaches in Cloud Resource Selection for Component Based Applications

Cloud resource provisioning for applications con-sisting of interacting components requires solving a constrained optimization problem. In this paper two exact methods (Constraint Programming and Satisfiability Modulo Theory) and a newly proposed population-based metaheuristic are investigated with respect to their potential in finding low-cost assignment of components to virtual machines such that all constraints are satisfied. The results obtained for three case studies show that the exact methods are appropriate as long as the cloud provider's list of offers is rather small (a few dozens). On the other hand, the metaheuristic provides acceptable solutions, but not necessarily optimal, even in the case of hundreds of offers.