A modular approach for model-based dependability evaluation of a class of systems

Abstract

Modeling for dependability and performance evaluation has proven to be a useful and versatile approach in all the phases of the system life cycle. Indeed, a widely used approach in performability modeling is to describe the system by state space models (like Markov models). However, for large systems, the state space of the system model may result extremely large, making it very hard to solve. Taking advantage of the characteristics of a particular class of systems, this paper develops a methodology to construct an efficient, scalable and easily maintainable architectural model for such class, especially tailored to dependability analysis. Although limited to the class considered, the proposed methodology shows very attractive because of its ability to master complexity, both in the model design phase and, then, in its solution. A representative case study is also included. © Springer-Verlag Berlin Heidelberg 2005.