Research On Self-Healing Techniques Of Composite Services For SOA

The web services are the most popular application technologies based on the Service-Oriented Architecture, which is the latest trend of distributed computing and software development. Usually we compose several web services which are business associated to form a composite service to meet the rapidly changing application requirements. Because of the loose-coupling and dynamics of composite services, the autonomy of web services and the varied changing of running environment, failures appear frequently during the execution of the composite service process. How to recover composite services from runtime failures is an important problem in the area of web services.Manual recovery and system self-healing recovery are two common types of composite service failure recovery methods. Compared with the manual recovery strategy, the self-healing technique can improve the reliability of composite services and alleviate the burdens on the process designers to handle the exceptions. It improves the self-management capabilities of composite services. In this paper, we mainly consider two aspects of self-healing techniques of composite services. They are the constitution of BPEL business processes with the ability of exception recovery and the functional failure monitoring of services. The main research work is as follows:We classify the exception of composite services, analyze the existing exception handling methods and summarize the rule-based exception handling strategies. Then we analyze the disadvantage of WS-BPEL language's exception handling mechanism. If the BPEL business process is deployed the exception handling strategies statically, it may violate the quality of service constraints during the execution. In order to solve this problem, we propose an algorithm in which the exception handling strategies are dynamically deployed. The algorithm can dynamically select exception handling strategies while exceptions arise during the composite services process execution. Consequently, it greatly improves the satisfaction of the quality of service constraints with lower cost of recovery.We summarize the monitoring requirements and the corresponding detection methods of composite services, and describe the monitoring requirements of member service as an optimization problem. Targeting at solving this problem, we propose a service monitoring algorithm which is based on limited resources can discover service failure as soon as possible.Based on the previous results of our research, we design and implement SOA business process prototype system. The exception handling strategies dynamically deploying algorithm and the service monitoring algorithm is used on two major modules: business deployment module and service control module. And we verify the self-healing techniques of composite services on the prototype system.