Bazaka Y. Models, methods and information technology to increase the efficiency of testing distributed systems

The dissertation is devoted to the decision of actual problem of software testing models and methods development of components by the creation of information technology that allows increasing the efficiency of distributed systems functioning. The analysis of software testing systems is carried out. The considered frameworks for testing distributed systems do not fully satisfy the work of a software tester who has to learn new technologies or write code in closed systems. A model of software testing of distributed systems based on software components autonomous deployment is proposed. Modified the Mike Kohn testing pyramid for adaptation when testing in distributed information processing systems, which allowed to expand the testing capabilities and apply the features of distributed systems. The comparison of schemes with different numbers of services on the indicator of testing efficiency, as which the a posteriori probability of good condition are accepted on condition of receiving a positive test result, is made. The method of distributed system software testing nodes has been modified. The mechanism of application of contract tests with the analysis of the forecasted result of the service’s behavior is developed. The software testing method for the distributed system user interface of nodes has been modified. It is shown that in comparison with end-to-end testing of user interfaces, the advantages of using simulators for testing user interfaces allow reducing the time spent on testing any service of user interface in comparison with end-to-end testing. The time is reduced by reducing the number of simultaneous user interface services. An experimental study of the application of the distributed system software testing method of nodes in determining the average testing time. For three or fewer nodes, the contract testing system is ineffective compared to the end-to-end testing mechanism, otherwise, the average testing time is reduced. For 5 nodes of the distributed system, the API testing time decreased by 6.5%. An experimental study of the application of the user interface software testing method of the distributed system nodes, which showed that with the increasing number of nodes, the time required to test distributed system services by simulation tests becomes less than the time required to test the same system through. For 5 interfaces of the distributed system, the test time decreased compared to Selenium WebDriver by 7.3%.