Використання рівняння автомодельного типу для моделювання надійності програмного забезпечення, реалізованого за допомогою хмарних технологій

A description of the testing and debugging process of a software for implementing finite element analysis in a cloud environment is provided. The types and causes of errors during configuration and execution of the program code are analyzed, including: errors that occur during infrastructure deployment and prevent the creation of the necessary cloud environment resources; errors in the infrastructure configuration that do not interfere with the operation of the environment, but make its use ineffective or unreliable; errors related to access to resources in the cloud environment; ineffective solutions that do not interfere with operation, but reduce its efficiency. Tools for ensuring operation in a cloud environment are described, including Azure Resource Manager, Terraform Cloud, GitHub Actions and others. The "main - worker" architecture is used. The selected system configuration is based on the research results is described. The issue of cloud resource management is considered. Based on the analysis of the experimentally obtained dependence of the number of errors and failures on time, a mathematical model is proposed for predicting the reliability of software operations. The model is built on the basis of time integration of a differential equation of the automodel type for the failure parameter, the proximity of which to zero in a certain time interval provides a forecast of the required reliability of the software tool. The procedure for identifying model parameters with a power dependence on the number of failures is discussed. The satisfactory ability of the automodel type equation to predict the reliability of software tools of the considered configuration is shown when compared with real testing data. The use of the developed mathematical model of reliability prediction for its application to other types of software systems using cloud technologies is proposed.