earticle

영어

A growing number of enterprises have been moving their works to encapsulate system functions, business logics and processing modules into Web service because of its flexibility and low-cost. However, service-oriented software calls for constantly adjusting its architecture in order to respond to varying user requirements and instable runtime environments. One of the most challenging issues is how to effectively implement a reconfiguration to ensure the business-critical application is trustworthy. In this paper, it proposes a method to dynamic composition for service replacement, which focuses on the probabilistic optimization to service planning of candidate compositions when the service failure is occurred. First, the input and output data specification is defined to describe interface behaviors, and then the probabilistic solution graph is introduced to formalize replacement strategies. Second, corresponding algorithms are discussed for optimization selection purpose, which includes reliability calculation process and model modification process. The former computes the probability value of each service planning generated from probabilistic solution graph. The latter modifies probabilistic solution graph model to recommend Top-k solutions, pruning the service planning which does not satisfy the specified probability value. Third, the architecture of prototype is presented to demonstrate the feasibility of the proposed method. Our method provides a reference to guarantee the reliability of service process in E-commerce.