No abstract

No abstract

PurposeThe paper seeks to provide new stochastic measures for evaluating software reliability and performance characteristics comprehensively.Design/methodology/approachThe time‐dependent behavior of the software system itself alternating between operable and inoperable states is described by a Markov process. Then several software performance measures when the arrival process of tasks follows a homogeneous Poisson process can be derived with the infinite‐server queuing theory.FindingsProvides a new evaluation method for understanding software reliability‐related performance quantitatively. Presents a way to decide the testing time duration to satisfy the software performance requirement based on the model.Research limitations/implicationsProduction of the reasonable method for deciding the model parameters characterizing the processing time and the arrival process of the task remains as future study.Practical implicationsMost of the past models have paid attention only to the software failure‐occurrence phenomenon. On the other hand, this model includes the software restoration and the task‐processing characteristics as well and can evaluate customer‐oriented quality. In software project management, this model can provide more useful and comprehensive information than before for software development managers.Originality/valueThis paper introduces the concept of performability into the stochastic software reliability modeling and provides new quantitative evaluation measures considering reliability and performance simultaneously.

This paper develops a plausible software availability model for operational use. In particular, we consider two kinds of restoration scenarios; one involves the debugging activity and the other does not involve the debugging activity. The time‐dependent behavior of the software system during the operation phase is described by a Markov process. Several stochastic quantities for software availability measurement are derived from this model. Finally, numerical examples are shown to illustrate software availability analyses.

No abstract