Enhancing Automated Microservice Decomposition via Multi-Objective Optimization

The microservices architecture (MSA) has become widespread across various industries to enhance the maintainability of applications. However, manual migration of monolithic applications to MSAs via microservice decomposition (MSD) can be intricate and may adversely impact the overall maintainability of a system if not executed correctly. To address these challenges, a multi-objective optimization approach can be used to generate optimal solutions, known as Pareto-optimal solutions. However, selecting the optimal MSD solution from the set of Pareto-optimal solutions can be challenging. To mitigate this challenge, we propose a multi-objective MSD method of using reference lines, a mathematical concept used in multi-objective optimization approaches, to efficiently select the best MSD solutions. We also define a set of violations and fix operations on the basis of MSD policies to prevent generating a vast amount of semantically meaningless MSD solutions. The definition of violations and fix operations and the use of reference lines accelerates the generation of MSD solutions. Our proposed method aids information technology architects by streamlining hyperparameter determination, a task deemed intricate for such architects.