Reliability Evaluation of Software Architectural Styles Based on Correlated Component Failure

Abstract

The aim of this study is to provide an efficient and scalable way to evaluate the reliability of different ‎software ar-chitectural styles with regard to correlated components failures. In this way, a method ‎based on the discrete time Markov chain (DTMC) model is proposed. In the proposed method, software architecture styles are used for reliability evaluation. The four main styles are transformed into Markov chain models and the transfer matrix is created for them, then using the Bernoulli distribution, the correlation between component is shown in the matrix and used in the evaluation process. The proposed method is scalable ‎such that it can be used for large software architectures with heterogeneous and homogeneous ‎styles. The results of the evaluation on the case study show that this method is more accurate than ‎the other methods for reliability prediction of the software architectures. As a result, it is ‎concluded that the proposed method is suitable for the preliminary estimation of the software ‎architecture reliability and can make a better comparison between various architectural styles to ‎choose the best and most suitable one from the available options.‎