Software Engineering Practice of Microservice Architecture in Full Stack Development: From Architecture Design to Performance Optimization
Machine Learning Theory and Practice · 2025
With the rapid development of the Internet and the growth of business requirements, the traditional single system is faced with dual challenges of maintenance and expansion due to its huge code base and highly coupled modules.The existing microservice migration methods rely on subjective judgment or single dimensional analysis, which makes it difficult to fully capture system characteristics.This article proposes a microservice reconstruction scheme based on multi feature fusion, integrating source code and runtime data to construct an inter class dependency undirected weighted graph model, extracting microservice modules through clustering analysis, and supporting database splitting strategies and distributed transactions to ensure data consistency.The core innovation includes a multi feature fusion mechanism that integrates semantic similarity, code structure dependencies, and runtime interactions; Develop the microservice extraction tool MicroRefactor, which significantly improves efficiency and accuracy through a three-stage process; Provide database splitting and consistency assurance solutions.Experimental verification shows that this scheme achieves efficient and accurate module extraction in JPetStore system and medium-sized e-commerce system.After reconstruction, the system functions are complete and performance optimization is significant.The current research is limited by the need to improve code automation processing capabilities and relies on manual design of test cases.In the future, we will explore automated processing technology and test case generation methods to further improve the refactoring plan.