From Monolith to Microservices: A Comparative Evaluation of Decomposition Frameworks

TL;DR

This paper tackles the problem of inconsistent evaluation in microservice decomposition for monolith-to-microservice migration by proposing a unified, reproducible evaluation framework. It applies a common metric pipeline across static, dynamic, and hybrid decomposition methods on four benchmark systems ($\text{JPetStore}$, $\text{AcmeAir}$, $\text{DayTrader}$, $\text{Plants}$), using metrics $SM$, $IFN$, $ICP$, and $NED$, with results harmonized via z-score normalization and a weighted aggregate score incorporating $W=\{3,-1,-1,-1\}$ and the Score formula $\mathrm{Score}(T)=\frac{\sum_m w_m\left(\frac{x_{m,t}-\mu_m}{\sigma_m}\right)}{\sum_m|w_m|}$. Key contributions include curating benchmark datasets, combining published results with replication outputs, and identifying $HDBScan$ as the most consistently effective decomposition tool, with $a$-BMSC and Mono2Micro offering strong, though trade-off–laden, alternatives; embedding-based methods show dataset sensitivity. The work provides a fair, reproducible basis for selecting decomposition approaches in real-world migrations and highlights practical implications for achieving scalable, modular architectures. Overall, the study advances objective comparison in microservice boundary identification and suggests directions for improving embedding- and graph-based techniques.

Abstract

Software modernisation through the migration from monolithic architectures to microservices has become increasingly critical, yet identifying effective service boundaries remains a complex and unresolved challenge. Although numerous automated microservice decomposition frameworks have been proposed, their evaluation is often fragmented due to inconsistent benchmark systems, incompatible metrics, and limited reproducibility, thus hindering objective comparison. This work presents a unified comparative evaluation of state-of-the-art microservice decomposition approaches spanning static, dynamic, and hybrid techniques. Using a consistent metric computation pipeline, we assess the decomposition quality across widely used benchmark systems (JPetStore, AcmeAir, DayTrader, and Plants) using Structural Modularity (SM), Interface Number(IFN), Inter-partition Communication (ICP), Non-Extreme Distribution (NED), and related indicators. Our analysis combines results reported in prior studies with experimentally reproduced outputs from available replication packages. Findings indicate that the hierarchical clustering-based methods, particularly HDBScan, produce the most consistently balanced decompositions across benchmarks, achieving strong modularity while minimizing communication and interface overhead.