Dependable Service Engineering: A Fault-tolerance based Approach

This paper is concerned with the engineering of dependable web services. We have developed an approach based on deploying existing web services within a middleware framework so that they are fault tolerant. Our approach is independent of the services themselves and may be configured to support a range of different fault tolerance mechanisms. Central to the approach are what we call fault tolerant service containers. These ‘contain’ externally provided services and incorporate support for fault detection and recovery. Containers are configurable with an XML-based recovery policy model that specifies what kind of fault tolerance mechanisms may be applied to the services it contains. The container serves as a proxy so that the existence of fault tolerance can be transparent to a service client. A tool and SDK simplify the creation and deployment of the container and its policy. We discuss the strengths and weaknesses of our approach and include some measurements of the overheads involved. We conclude that our container-based mechanism provides a simple, low cost approach to enhancing web service dependability. Categories and subject descriptors: D.2.2 Design Tools and Techniques, D.2.13 Reusable Software General terms: Design, Reliability Additional key words and phrases: web services, service engineering, fault-tolerance