It is becoming common to see software applications taking advantage of web services available publicly to meet their needs, rather than developing an in-house solution. This introduces the problem where failures can occur on the network, or on the service provider, outside the influence of the developer. This thesis proposes a reliability-aware framework with a focus on availability, which applies a recovery block scheme to services provided by different developers. The proposed framework allows developers to specify alternative services which meet the core specifications of their primary service. When a failure is determined to have occurred, the request to the primary service is mapped to an alternative service. A prototype has been developed as a proof of concept, which has been evaluated on metrics based on potential use cases. The experimental results show that the system is successful at providing availability when failure occurs, at a cost to overall performance.

No abstract