Building an Optimal Service Environment for ASBS Based on Pareto Dominance

Self-adaptation is a salient feature to make an adaptive service-based software system (ASBS) able to guarantee high dependability as well as deliver high quality of service (QoS) in its dynamic operating environment. It is usually realized by reselecting and rebinding services at runtime, and will increase the overhead of an ASBS if it happens frequently. Being aware of the expensive computing cost of optimal service selection involved in self-adaptation, this paper propose a Pareto dominance based approach to build a sharable optimal service environment for all executing system instances, which could provide optimal services by precomputation triggered by underlying candidate service changes continuously. Experimental results demonstrate that this approach can make effective overhead reduction in providing dependable and high quality services.