The problem of service composition is the process of assigning resources to services from a pool of available ones in the shortest possible time so that the overall quality of service is maximized. This article provides solutions for the composition problem that takes into account its scalability, services’ locations, and users’ restrictions, which are key for the management of applications using state-of-the-art technologies. The provided solutions use different techniques, including genetic algorithms and heuristics. We provide an extensive experimental evaluation, which shows the pros and cons of each of them, and allows us to characterize the preferred option for each specific problem. Since no solution dominates the others, we propose a decision tree, based on our results, to select the best composition algorithm in each situation.
