Optimizing Performance of Composite Services in Multiple Networks Enterprise Environment

Many approaches have been proposed for performance optimization of centralized services composition. However, with the prevalence of web services, services orchestration evolves from intra-enterprise integration to cross-enterprise integration and involves multiple networks. In such environment, the centralized orchestration suffers from several performance and reliability issues due to the limitation of centralized execution engine and the network conditions. In this paper, we try to optimize the performance of composite services in enterprise environment by adjusting the deployment location of component services. First, we propose a decentralized orchestration model for multiple networks environment. Based on this model, we propose the concept called service interaction strength, which is denoted by the data interaction between component services in a composite service as well as the importance of composite service itself. Then we design an objective function for all composition services with overall minimum cross domains service interaction costs and thus reduce the response latency and increase the performance of composite services. At last we introduce a mixed repair genetic algorithm to get the optimized deployment plans for all component services. Experimental results on simulation prototype show the optimization effectiveness of our approach and the performance of algorithm itself is also acceptable. © 2011 Asian Network for Scientific Information.