A Microservices Identification Approach based on Problem Frames

As software grows in size continuously, traditional monolithic architecture is difficult to upgrade and update. As a more flexible architecture, microservices consist of smaller pieces or services that are loosely coupled and can be deployed independently. More and more IT companies are adopting microservices architecture. However, migrating to microservices is still a big challenge due to the lack of a more systematic methodology on microservices identification. Most existing studies on microservices identification are mainly based on refactoring the system module rather than considering the actual functional requirements. Therefore, we propose a microservices identification approach based on the problem frames to address this challenge, which considers real-world requirements. It relies on the merging process with the problem diagram, where the problem domain is processed, and the causal chain is obtained by searching the problem diagram. Subsequently, we calculate the correlation degree of the domains in the problem diagram. Then, based on the calculated correlation, we implement a clustering algorithm for the problem domains and get the categorization result of the problem domains which is the initial microservice candidate. The developers can consider each category of the problem domain as a microservice by converting the problem domains into the microservices. Finally, a case study verifies the validity of our proposed approach.