Availability modeling of modular software

Dependability evaluation is a basic component in assessing the quality of r epairable systems. A general model (Op) is presented and is specifically de signed for software systems; it allows the evaluation of various dependabil ity metrics, in particular, of availability measures. Op is of the structur al type, based on Markov process theory. In particular, Op is an attempt to overcome some limitations of the well-known Littlewood reliability model f or modular software. This paper gives the mathematical results necessary to the transient analysis of this general mo del, algorithms that can efficiently evaluate it More specifically, from the parameters describing the: evolution of the execution process when there is no failure, failure processes together with the way they affect the execution, recovery process, the results are obtained for the: distribution function of the number of failures in a fixed mission, dependa bility metrics which are much more informative than the usual ones in a whi te-box approach. The estimation procedures of the Op parameters are briefly discussed. Some simple examples illustrate the interest in such a structur al view and explain how to consider reliability growth of part of the softw are with the transformation approach developed by Laprie et al. The complet e transient analysis of Op allows discussion of the Poisson approximation b y Littlewood for his model.