Research On Dynamic And Adaptive Service Composition In Open Network Environment

With more and more storage resources, computing resources, application resources or even software resources published on the Internet for public access, Internet is becoming the biggest open distributed computing platform. Web service composition is the core technology for the resource aggregation and system integration in this open network environment. Supported by the project of"Distribute Modeling and Simulation Resource Management", this research focused on the coordination model and adaptability of Web service composition. The contributions of this dissertation are as follows:1. Data flow optimization based decentralized Web service composition. Current Web service composition is mostly coordinated in a centralized model. However, centralized Web service composition suffers from bad performance and insufficient support for dynamic process management. Therefore, a data flow optimization based BPEL process decomposition approach is proposed to support decentralized Web service composition. This approach partitioned the BPEL process into sub-processes and deployed these sub-processes distributedly. A data flow optimization algorithm was designed to reduce the data transfer between these sub-processes. Experimental results show that the response time is reduced around 30%-50%, furthermore, the computational complexity is also reduced from O(en) to O(n2).2. A Web service coordination framework to support data flow distribution. Current web services composition models encounter performance or reusability problems. Therefore, a web service coordination framework was presented to support the data flow distribution. The formal definition and meta-model of the coordination protocol were presented. Algorithms to optimize and transfer the BPEL process into the coordination protocol were proposed, so that BPEL based composite services can be deployed in the framework. The experiment results show that the composite service deployed in the coordination framework achieves better performance than BPEL based composite service.3. A reflective framework to improve the adaptability of Web service composition. Current web services composition solutions are rather restricted and inflexible as they are based on pre-defined models of the process environment. But web services may run in a highly dynamic environment. Therefore, a mechanism is required to support web services composition in dynamic and flexible environment. The reflective framework presented here aims to improve the adaptability of BPEL-based web service composition. A meta-model was defined to build the self-representation of the web services composition. The meta-model will be modified to adapt to the changing environment, and then, the reflection mechanism utilized in the framework will adjust the web services composition automatically. To ensure the correctness of dynamic adaptation, a set of constrains and a verification approach is proposed. A prototype adaptive service composition environment has been developing to implement our reflective framework and demonstrate its effectiveness on providing adaptive web services composition.4. Composite service reconfiguration approach under the end-to-end QoS constraints. In SOA, services may become volatile and fail to deliver the QoS as requested by users. We present an approach for repairing failed services by replacing them with new services and ensuring the new service process still meets the user specified end-to-end QoS constraints. An iterative structural inspection algorithm is designed to produce reconfiguration regions that include one or more failed service. By reconfiguring only services in the selected regions, the business process will not be affected significantly. The algorithm may also utilize those available QoS constraints to relax the original constraints of a reconfiguration region and to provide more effective reconfiguration solutions. The middleware components to support the service reconfiguration in the LLAMA framework were also implemented.5. SOA-based modeling and simulation resources management system. TheαLLAMA platform was proposed to support adaptive and dynamic composite service management. Adaptation Manager was implemented to support reflective computation and service reconfiguration. To support different coordination models, the accountable agents were required to optimize and distribute the data flow among all participant services. After studying the requirements of modeling and simulation resource management, we designed a SOA based simulation resource management system. To leverage the accountability and adaptability ofαLLAMA platform, a simulation service bus was built management core simulation services and simulation application services.