Transforming Monolithic into Scalable Hybrid Microservices for Enterprise Resource Planning
Taufiq Maulana Firdaus, Firda Nur Rizkiani · 2025
The shift from monolithic to microservices architecture has become a key strategy in modern software development, enabling scalability, maintainability, and flexibility. However, transitioning to a hybrid microservices architecture requires a careful balance between cohesion and coupling to ensure system stability. This research integrates the Quality Attribute Workshop (QAW), Lack of Cohesion of Methods (LCOM4), afferent and efferent, and graph neural network (GNNs) to evaluate the architectural transformation. QAW provides qualitative insights through identifying essential quality attributes and guiding architectural decisions for visualization. LCOM4 and afferent and efferent offer a quantitative evaluation, measuring internal cohesion and service dependencies. GNN is used as the final evaluation to identify hidden anomalies linked to coupling and cohesion that hidden from the traditional evaluation. The results indicate a high cohesion and low coupling evaluation with overall scores of LCOM4 reach 1.22 with afferent and efferent of 0.52, and no anomalies are found within the hybrid microservice architecture that might causes by over-centralized services and high potential of resource bottlenecks. This study demonstrates the importance of combining qualitative and quantitative assessments to ensure architectural robustness. By integrating QAW, LCOM4, afferent and efferent, and GNN based anomaly detection, this framework provides a comprehensive evaluation approach that ensures scalability, maintainability, and resilience in hybrid microservices systems.