Facing Uncertainty in Web Service Compositions

Web service compositions run in complex computing infrastructures where arising events may affect the quality of the system. However, crucial Web service compositions cannot be stopped to apply changes to deal with problematic events. Therefore, the trend is moving towards context-aware Web service compositions, which use context information as a basis for autonomic changes. Under the closed-world assumption, the context and possible adaptations are fully known at design time. Nevertheless, it is difficult to foresee all the possible situations arising in uncertain contexts. In this paper, we leverage models at runtime to guide the dynamic evolution of context-aware Web service compositions to deal with unexpected events in the open world. In order to manage uncertainty, a model that abstracts the Web service composition, self-evolves to preserve requirements. The evolved model guides changes in the underlying WS-BPEL composition schema. A prototype and an evaluation demonstrate the feasibility of our approach.