A Soft Constraint-Based Approach to QoS-Aware Service Selection

Service-based systems should be able to dynamically seek replacements for faulty or underperforming services, thus performing self-healing. It may however be the case that available services do not match all requirements, leading the system to grind to a halt. In similar situations it would be better to choose alternative candidates which, while not fulfilling all the constraints, allow the system to proceed. Soft constraints, instead of the traditional crisp constraints, can help naturally model and solve replacement problems of this sort. In this work we apply soft constraints to model SLAs and to decide how to rebuild compositions which may not satisfy all the requirements, in order not to completely stop running systems.