Dynamic Selection of Services under Temporal Constraints in Cloud Computing

Abstract

To guarantee the successful execution of service based process, one important requirement is QoS-driven dynamic selection of candidate services. The existing methods fail to consider the temporal constraints in build-time, and dynamic re-planning of execution plans in run-time. In this paper, we propose a novel penalty-based genetic algorithm to dynamically select the services for service based process under temporal constraints. Our approach can efficiently select the appropriate services in practical business environment. Furthermore, if some temporal constraints are violated in run-time, we provide solutions to re-plan them so that the violations can be resolved. This approach is particularly applicable and efficient for large-scale and complex processes. Finally, a prototype system has been developed to support the automatically execution of our approach.