A performability approach for evaluating the impact of risks in software development

Abstract

Successful software projects are still an important challenge for the industry. In this context, risks deserve particular attention, since they may lead to project failure. Several techniques have been proposed to evaluate the effects of such undesirable issues, but probability estimates are usually neglected, affecting a proper evaluation of risks. The primary objective of this study is the adoption of dependability and performability techniques for probabilistic evaluation of development risks in software projects. This work utilises dependability and performability models, based on stochastic Petri nets and reliability block diagrams, for quantitative evaluation of risks regarding the developer turnover and requirement implementation in software development projects. Besides, the proposed approach also considers fault-tolerance techniques to mitigate the impact of risks. The case studies contemplate two real-world software projects with developer turnover and requirement implementation as remarkable risks. The proposed models are utilised for risk assessment, and techniques, such as dynamic redundancy and periodic meetings, are adopted to improve project availability. The proposed experiments demonstrate project managers have a prominent tool to evaluate risks and different configurations to avoid or mitigate these undesirable issues.