A method of automating the process of calculating reliability indicators of software systems and their components

Abstract

The existing means of calculating reliability indicators of software systems are analyzed. It has been established that to determine the reliability indicators of software systems, it is advisable to use the structural-logical analysis of reliability block diagrams, since it clearly and most adequately reflects the process of calculating the reliability indicators of the software system as a whole and its components in particular. Despite the  external simplicity of such an analysis, conducting it is not a trivial task, because even building the condition of technical system operability is a difficult task, especially in the case of the presence of many elements with various connections between them, when solving which manually there is a very high probability human error. Also, the construction and visualization of the graph of states / transitions is a nontrivial task, since the  number of possible states of the software system depending on the number of elements grows exponentially, and, in turn, increases the complexity of the system of differential equations, the solution of which makes it possible to calculate the necessary reliability indicators. It was determined that the process of reliability design of complex software systems in general, and their components in particular, requires automation of all its  stages, starting from the compilation of the reliability block diagram (RBD), and ending with the visualization of the obtained results. A method of automating the process of calculating the reliability indicators of software systems and their components has been developed, which consists of eight steps and, unlike the existing ones, allows the designer to intuitively enter not only input data about the structure, but also the software  architecture itself from the point of view of its reliability, and also automates all stages of calculating reliability indicators, from the stage of constructing a reliability block diagram to the stage of finding the distribution of probabilities of the software system being in all possible states. The proposed method makes it possible to use in various combinations the methods,  lgorithms and software tools used for the reliability design of software systems and to choose from them the most adequate to the needs of the user in a specific situation. The use of the developed method makes it possible to reduce the influence of the human factor and the probability of making an error in the process of calculating reliability indicators of software systems at all stages of reliability design and to reduce its time by at  least 21 %.