.png){kind=icon}
Return
BenchPDM: Benchmarking pattern detection methods in microservice-based systems using automatically generated pattern-assisted testbeds
.png){kind=icon}
PRE
AI
DOI：10.1007/s10664-026-10949-6.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Microservices architecture (MSA) is an increasingly popular trend in software architecture design. While the use of microservices offers numerous benefits, such as enhanced flexibility and scalability, it also introduces specific challenges and complexities. To address these challenges, various design patterns have been applied in system implementations, each aimed at resolving specific issues within microservices-based systems (MBS). Understanding the design patterns employed in a large software system can significantly enhance comprehension of the system’s functionality. However, organizations often lack strong documentation regarding the types of patterns used, the number of patterns, and their locations within the system. To date, various pattern detection methods (PDMs) have been developed with the goal of identifying design patterns within software systems. The effectiveness of these PDMs may vary depending on factors such as the scale of the software system, the types of patterns used, and etc. Therefore, the selection of an appropriate PDM is highly dependent on the specific software system being targeted. Consequently, there is a critical need for a tool that can evaluate and compare PDMs from various perspectives. Although several benchmarks have been proposed for MBSs, none have focused on assessing PDM. In this article, we present a benchmark named BenchPDM aimed at benchmarking PDMs in MBS using automatically generated pattern-assisted test beds. To achieve this, BenchPDM includes two fundamental modules. In the PBMC module, pattern-assisted microservices-based systems (PAMBS) are generated with varying degrees of complexity to simulate real-world MBSs of small, medium, and large scales. The PDM-ES module then evaluates and compares the performance of PDMs using the generated PAMBS from different factors. To assess the performance of BenchPDM, we conducted experiments on five PDM methods, two of which are among the most popular and commonly adopted approaches, SVM-based and graph-based methods. The experimental results demonstrated that BenchPDM can effectively evaluate and compare PDMs across various factors.
Keywords:
Microservices architecture
Microservice design patterns
Design pattern detection
Benchmarking
Synthetic microservice systems
Pattern detection methods
Journal
IF:
3.6
Papers:
2.0K
Citations:
5.3K