Abstract

We describe an object-oriented framework for designing and implementing availability management services. We present a library of object-oriented implementation of a suite of dependable distributed protocols, and show how these protocols can be composed together to build an application that provides highly available services to users. Specifically, we introduce our proposed system architecture called JAMS (Java-based Availability Management System) which integrates the availability management service and the online FDIR (Fault diagnosis, isolation and reconfiguration) service on a distributed heterogeneous platform. The JAMS architecture is described in a way that reflects the design and development phases, spanning from requirement analysis to module implementation. A case study of a banking service is presented to illustrate how JAMS can be used to build a dependable distributed application in a modular manner ensuring that the application service remains continuously available to the user despite the presence of failures, maintenance and growth. (Abstract shortened by UMI.)