Extracting the Architecture of Microservices: An Approach for Explainability and Traceability

The polyglot nature of microservice architectures and the need for high reliability in security analyses pose unique challenges that existing approaches to automatic architecture recovery often fail to address. This article proposes an approach for extracting detailed architecture models from polyglot microservice source code focusing on explainability and traceability. The approach involves abstracting code navigation as a tree structure, using an exploratory algorithm to detect architectural aspects, and providing a set of generic detectors as input. The architecture models are automatically annotated with detailed information that makes them useful for architecture conformance checking and violation fixing. Our case studies of microservice software systems validate the usefulness of our approach, providing insights into its completeness, accuracy, and effectiveness for software architecture tasks.