Design of Microservices with focus on granularity

Determining the “right sized” boundaries of a microservices architecture remains one of the most persistent design challenges in modern software engineering. Services that are too coarse resemble a monolith, while overly fine grained “nano services” introduce excessive coordination overhead. This thesis proposes a six step, metrics driven methodology that unites qualitative domain analysis with quantitative evaluation to guide architects toward balanced service granularity. The process begins with a thorough gathering of functional and non functional requirements, followed by domain partitioning that leverages domain driven, data centric, event-driven, or hybrid strategies. A lightweight prototype is then produced to surface hidden dependencies before three key metrics—Lack of Cohesion of Methods (LCOM), Structural Coupling (SC), and the Service Granularity Metric (SGM)—are collected to evaluate service boundaries. Decision rules grounded in these measurements indicate when to split, merge, retain or refactor services, and an iterative refinement loop embeds boundary checks into routine development cycles. The methodology is validated through a logistics-marketplace (freight-auction) case study. Metric values confirmed that the initial domain-driven split produced cohesive, loosely coupled services; subsequent monitoring demonstrated how the approach detects boundary drift as the system evolves. The results suggest that integrating domain understanding with focused metrics offers a repeatable, evidence-based path to microservices architectures that remain maintainable, scalable, and aligned with business goals.