The microservice-based architecture – a SOA-inspired principle of dividing systems into components that communicate with each other using language-agnostic APIs – has gained increased popularity in industry. Yet, migrating a monolithic application into microservices is a challenging task. A number of automated microservice decomposition techniques have been proposed in industry and academia to help developers with the migration complexity. Each of the techniques is usually evaluated on its own set of case study applications and evaluation criteria, making it difficult to compare the techniques to each other and assess the real progress in this field. To fill this gap, this paper performs an independent study comparing eight microservice decomposition tools that implement a wide range of different decomposition principles with each other on a set of four carefully selected benchmark applications. We evaluate the tools both quantitatively and qualitatively, and further interview developers behind two of the selected benchmark applications. Our analysis highlights strengths and weaknesses of existing approaches, and provides suggestions for future research, e.g., to provide differential treatment of application elements of different types, to customize the decomposition strategy and granularity per specific application, and more.

@inproceedings{WangBR24,
  title = {Microservice Decomposition Techniques: An Independent Tool Comparison},
  author = {Yingying Wang and Sarah Bornais and Julia Rubin},
  booktitle = {ASE},
  year = {2024},
  doi = {10.1145/3691620.3695504}
}