Identification of microservices from monolithic applications through topic modelling

Microservices emerged as one of the most popular architectural patterns in the recent years given the increased need to scale, grow and flexibilize software projects accompanied by the growth in cloud computing and DevOps. Many software applications are being submitted to a process of migration from its monolithic architecture to a more modular, scalable and flexible architecture of microservices. This process is slow and, depending on the project's complexity, it may take months or even years to complete. This paper proposes a new approach on microservice identification by resorting to topic modelling in order to identify services according to domain terms. This approach in combination with clustering techniques produces a set of services based on the original software. The proposed methodology is implemented as an open-source tool for exploration of monolithic architectures and identification of microservices. A quantitative analysis using the state of the art metrics on independence of functionality and modularity of services was conducted on 200 open-source projects collected from GitHub. Cohesion at message and domain level metrics' showed medians of roughly 0.6. Interfaces per service exhibited a median of 1.5 with a compact interquartile range. Structural and conceptual modularity revealed medians of 0.2 and 0.4 respectively. Our first results are positive demonstrating beneficial identification of services due to overall metrics' results.