Research On Key Techniques For Self-Recovery Of Service Composition

In the service-oriented environment, a single Web service can hardly satisfy the given request, so the composition of multiple Web services is required to fulfill the goal. Because the reliability and effectiveness of remote Web services are unclear, and the performance of Internet is also unpredictable, it is difficult to guarantee the performance of these service-oriented applications, which are developed on Web services. As web services in a business process are loosely coupled and autonomous, failures occur frequently during the execution of the business process. Therefore how to reduce failure occurrence and recover service compositions from failures automatically becomes a significant problem in the area of SOC.To address the above problems, we propose several self-recovery techniques for composite services in this thesis. Our techniques take into account the autonomy and heterogeneity of web services, and thus can improve the reliability of composite services in the service environment. The main research works and contributions of this thesis are described as follows:1. Redundancy-based fault tolerance strategies are proposed for building reliable service composition applications, which are usually developed on the unpredictable remote web services. We propose a self-recovery service composition framework. Based on this framework composite services can automatically generate their optimal fault tolerance strategy to achieve good service reliability as well as good overall performance.2. We propose a utility-driven recovery strategy selection algorithm which explicitly balances the cost of provisioning with the expected value of the service composition. We propose a business model for composite services, and analyze the relationship between the utility of a business process and the requirement of customers. According to the SLA analysis, we give a selection algorithm which chooses optimal fault tolerance strategy. We conduct various experiments and the results show our selection algorithm performs well.3. In this research, we study the model of fault diagnosis to detect and defuse the real cause of a problem when service errors (such as incorrect result or SLA violation) occur in a service composition. Our approach leverages Bayesian networks to identify the most likely problematic services in a process and selectively inspect those services. The performance study shows that our proposed mechanism is effective on identifying the root cause of problems and can achieve significant cost savings.4. A prototype system that support self-recovery service composition is implemented based on above research works. The design and implementation of the core components are exprained in details. Self-recovery strategies are automatically transformed into standard WS-BPEL language and integrated into normal business logic.