A Fault Detection Mechanism for Service-Oriented Architecture Based on Queueing Theory

SOA is an ideal solution to application building, since it reuses the existing services as many as possible. The fault tolerance is one important capability to ensure the SOA- based applications are high reliable and available. However, fault tolerance is such a complex issue for most SOA providers that they hardly provide this capability in their products. This paper provides a queuing-theory-based algorithm to fault detection, which can be used to detect the services whose performance becomes unsatisfactory at runtime according to the QoS descriptor. Based on this algorithm, this paper also gives the reference models of the extended service and the architecture of fault-tolerance control center of enterprise services bus for SOA-based applications.