Reliability Maximization of Component-Based Software Systems

Most of the previously reported literature focused primarily on the reliability evaluation and prediction of component–based software systems, rather than exploring cost/reliability tradeoffs among competing architectural alternatives. A few techniques which treat the subject of reliability optimization assume the system to be series–parallel, an assumption that does not hold even in the case of moderately sized software systems. In this paper we present an optimization framework for a Markovian system using a genetic algorithm, and demonstrate its utility to maximize the reliability of a component–based software system for a given cost. The problem of minimizing the software cost for a given reliability target is important in the context of systems where the reliability target is non negotiable, and has been the topic of our previous research [3]. In the context of many commercial software systems, however, cost is the most prominent driving factor, and the problem then is to achieve maximum reliability while staying within the pre–specified cost limits. The use of genetic algorithm as an optimization technique is motivated by three facts, namely, a potentially large and discontinuous search space, usually a nonlinear but monotonic relation between the cost and reliability of individual components comprising the software, and complex software architectures giving rise to nonlinear dependencies between individual component reliabilities and the overall system reliability [2]. 2 Architecture–based reliability analysis