Reliable software systems using reusable software components

Summary form only given, as follows. J. Von Neumann (1956) and E.F. Moore and C.F. Shannon (1956) discussed building reliable automata and relay circuits (hardware) using less reliable components. They show that carefully designed replication of components in a hardware system can increase the probability of failure free operation of that system. There is a powerful trend in the industry now to build software systems using as many software components as possible. These components might be commercial off the shelf (COTS) or in-house software libraries and modules; we call all such components reusable software components. We argue that the reliability of such a software system can be improved not only by replicating the software components, but also by active monitoring, checkpointing and rejuvenation, and providing facilities for cold, warm and hot fail over/restart of those components. These capabilities themselves can be built as reusable software modules that can be linked to the actual system components. We present the architecture of such a software system and a preliminary analysis to show the feasibility of this approach for building reliable software systems using reusable software components. Research into analyzing the reliability of such systems is gaining attention. These facilities provide diversity in the execution environment of a software component leading to a higher level of reliability of the software system, much as replication provides diversity in the physical environment of a hardware component giving rise to improved reliability of the hardware system that Von Neumann and others have pioneered