Method for Detecting Architectural Degradation in Distributed Web Systems Based on the Dynamics of Structural Metrics

The article is intended to discuss the problem of architectural degradation of distributed web applications under conditions of intensive development, CI/CD processes, and increasing technical debt volume. A formalized approach to identification and monitoring of degradation processes based on the analysis of the dynamics of structural metrics of the architecture is proposed. A model that describes the change in the architectural state over time using metrics of connectivity, cyclomatic complexity, density of dependencies, and other quantifiable parameters is developed. The concept of architectural resilience is introduced, which allows to assess the ability of a system to maintain structural integrity in the condition of external and internal changes. A system of rules based on time series metrics is used to assess degradation, and an indicator of the «architectural shift point» is proposed, as the moment emphasizing restoring of the original structure, that requires significant efforts. The model is validated based on simulated and real data obtained from the CI/CD pipeline. The model is validated using simulated and real data obtained from CI/CD pipelines of three web systems of varying complexity. It is shown that timely detection of architectural degradation allows to increase the efficiency of architectural management, reduce the risks of failures and optimize long-term costs of system maintenance. The results obtained can be used to design architectural dashboards, as well as in order to implement intelligent decision support systems in a DevOps environment.