Research on Reliability Prediction of Web Service Composition
|School||National University of Defense Science and Technology|
|Course||Computer Science and Technology|
|Keywords||Web services Service portfolio Software Reliability Software fault-tolerant Reliability Optimization|
With the development of information technology, the Internet has gradually evolved into an open distributed computing environment. Especially with the emergence and promotion of Web services technology, Web services have become a mainstream technology choice recognized implementation services. Hope through the dynamic combination of different services provided by the services in a service-oriented computing, more and more businesses and organizations to build a distributed application system. Which is a very important question is how the combination of dynamic and efficient prediction service quality of service (reliability), and then guide service discovery and composition. At present, many domestic and foreign research institutions around the service portfolio reliability prediction research work and made many valuable research. Shortcomings in the reliability of prediction of the current service portfolio, in-depth research process network model of a combination of services as the analysis and prediction of Web service composition based on the reliability of the forecasting model of the architecture and the technical characteristics of the Web service composition, the reliability of the model. Using this model, the research service reliability prediction based on the combination of BPEL business processes, application software fault-tolerant technology to enhance the reliability of the system, as well as a combination of service reliability optimization problem in the service portfolio. The main work of this paper include: 1. Comprehensive analysis of the current Web services technology features and architecture-based reliability prediction and the typical reliability prediction model and method are summarized. Characteristics for Web services and service portfolio, based on stochastic Petri nets, the proposed service of process description of the network model portfolio of services with a combination of services - use profile. The combination of service processes network model is suitable for such concurrent software system to describe the service portfolio. 3. Based flow network model to a combination of services, service composition based on BPEL business process reliability prediction method. First, convert the BPEL business processes corresponds to a combination of service processes network model, the combination of component service reliability, the forecast combination service reliability information. And, on this basis, to study the combination of service reliability sensitivity analysis method used to determine the component service reliability and the degree of reliability of the profile of the combination of services overall, to help designers select and determine the key components and scene . Redundant software-based fault-tolerant technology is an effective way to improve the reliability of software systems. Redundancy characteristics of the Web services, service-oriented computing, software fault-tolerant technology easier to apply in the combination of services. In this paper, a combination of services in BPEL-based business processes to achieve software fault-tolerant, and quantitative assessment of its reliability and performance. The service portfolio of software fault-tolerant technology to improve the reliability of composite service, it would also bring the cost of the combination of services increase. In this paper meet cost constraints, the combination of reliability optimization problem with fault-tolerant strategy selection and proposed an the reliability optimization algorithm based on Pareto optimal solution. The design and implementation of a combination of service reliability prediction tools, said the combination of BPEL process service automatically converts the combination of the reliability of the service processes network model, forecast combination service. In summary, this work for several key issues that exist in Reliability Prediction of Web service composition depth research, widely used for Web services technology and services portfolio has a certain theoretical and practical values.