Middleware-Based Failure Detection and Recovery Services for Fault-Tolerant E-services

The runtime detection of failure and recovery from failure is a major challenge facing e-business and e-commerce applications. Different types of failure are well understood through the failure model, but the detection and differentiation between these failures still proves difficult at runtime. Even when failures are detected, recovery may be hindered as certain failures may mask the root cause failure making it difficult to elaborate a recovery strategy. Through this paper we describe a pragmatic approach to failure detection and recovery based on the combination of middleware-based instrumentation and control services. In particular, we describe the development of failure detection instruments and failure recovery control services using Jini middleware technology. The failure detection instruments are capable of identifying different failure types and failure recovery control services make use of failure patterns to activate appropriate recovery strategies.