A Semi-Automated Approach for Incremental Migration from Monolithic to Microservices Architecture

Abstract

As software applications grow in size and complexity, maintaining and scaling them becomes increasingly difficult. Traditional monolithic architectures, where all components are combined into a single unit, often face issues such as limited scalability, cumbersome maintenance, and problematic deployment. The microservices architecture has emerged as a solution, breaking down applications into smaller, loosely coupled services, each responsible for a specific business function. This transition process, known as migration, is challenging due to the difficulty in determining the optimal decomposition of the monolithic system.

This thesis presents a novel framework designed to facilitate the migration from monolithic to microservices architecture using the Strangler Fig Pattern. Unlike existing approaches that typically attempt to decompose the monolith in a single iteration, our framework supports a gradual, iterative migration process. This allows for smoother transitions, reduced risk, and better management of complexity.

Key contributions of this work include the development of a tool that leverages both static and dynamic analysis to identify microservice candidates. The tool integrates these heterogeneous data sources using the Single Source of Truth (SST) paradigm, ensuring consistency and reliability. The performance of the tool is evaluated on two well-known Java Spring projects, demonstrating its effectiveness in creating well-modularized, cohesive, and loosely coupled microservices.

The results show that our approach not only meets the desired principles of microservice architectures but also compares favorably with other state-of-the-art methods. By providing a practical and systematic solution for gradual migration, this thesis addresses a significant gap in the existing literature and offers valuable insights for practitioners seeking to modernize large-scale software systems.