On the maintenance support for microservice-based systems through the specification and the detection of microservice antipatterns

• Published in: The Journal of systems and software Vol. 204; p. 111755
• Main Authors: Tighilt, Rafik, Abdellatif, Manel, Trabelsi, Imen, Madern, Loïc, Moha, Naouel, Guéhéneuc, Yann-Gaël
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.10.2023
• Subjects: Antipatterns; Detection; Maintenance; Microservices; Antipatterns; Maintenance; Detection; Microservices
• ISSN: 0164-1212; 1873-1228
• DOI: 10.1016/j.jss.2023.111755

The software industry is currently moving from monolithic to microservice architectures, which are made up of independent, reusable, and fine-grained services. A lack of understanding of the core concepts of microservice architectures can lead to poorly designed systems that include microservice antipatterns. These microservice antipatterns may affect the quality of services and hinder the maintenance and evolution of software systems. The specification and detection of microservice antipatterns could help in evaluating and assessing the design quality of systems. Several research works have studied patterns and antipatterns in microservice-based systems, but the automatic detection of these antipatterns is still in its infancy. We propose MARS (Microservice Antipatterns Research Software), a fully automated approach supported by a framework for specifying and identifying microservice antipatterns. Using MARS, we specify and identify 16 microservice antipatterns in 24 microservice-based systems. The results show that MARS can effectively detect microservice antipatterns with an average precision of 82% and a recall of 89%. Thus, our approach can help developers assert and improve the quality of their microservices and development practices. •Microservice antipatterns may hinder the maintenance of microservice-based systems.•We propose MARS, a highly automated tool to detect 16 microservice antipatterns.•We validate the detection results of MARS on 24 microservice-based systems.•We accurately detect 16 antipatterns with a precision of 82% and a recall of 89%.