Implementation and testing of the configuration optimization module for automated testing of multi-component information systems

Abstract

The article presents the results of the first iteration of implementation and experimental testing of the integrated approach to selecting optimal configurations for automated testing of multi-component information systems (IS), previously proposed by the authors. A modular software architecture has been implemented combining combinatorial methods (Pairwise Testing, Orthogonal Arrays) and genetic algorithms with CI/CD integration capabilities. Testing was conducted on synthetic datasets simulating IS with 108–243 configurations. Weighted analysis was used for objective comparison of methods by efficiency criteria. The objective of the study is to obtain experimental verification results of the proposed approach to assess its effectiveness, particularly in reducing the number of tests to 5–15% of the full set while maintaining high coverage of critical scenarios. Methodology includes implementation of greedy combinatorial algorithms, a heuristic genetic algorithm, execution time measurement, coverage validation, and comparative analysis using weighting factors. The scientific novelty lies in the practical confirmation of the approach's effectiveness, confirming test set reduction to 6.2–11.1% (12–15 configurations) with 100% coverage of pairwise interactions and the highest integral score of 0.9075 on a multi-criteria scale