Centrality Change Proneness: an Early Indicator of Microservice Architectural Degradation

2025-06-09

AI Summary

This paper addresses the challenge of early detection of microservice architecture degradation by proposing a novel diagnostic paradigm driven by temporal centrality change. Methodologically, it introduces the Centrality Change Proneness (CCP) metric, which models the evolutionary trend of node centrality in temporal service call graphs as a quantitative indicator of degradation propensity. CCP integrates static code analysis, multi-version architecture reconstruction, and Spearman/Pearson correlation testing for empirical validation. Key contributions include: (1) the first empirical demonstration that centrality stability exhibits significant correlations with seven size- and five complexity-related software metrics; and (2) CCP’s strong predictive capability—achieved without relying on conventional software metrics—enabling timely, lag-free degradation detection. Evaluated on seven open-source microservice systems comprising 42 services, CCP demonstrates both effectiveness and independent diagnostic value.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Web Mining and Content Analysis: Models for Web evolution
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization

Abstract

Over the past decade, the wide adoption of Microservice Architecture has required the identification of various patterns and anti-patterns to prevent Microservice Architectural Degradation. Frequently, the systems are modelled as a network of connected services. Recently, the study of temporal networks has emerged as a way to describe and analyze evolving networks. Previous research has explored how software metrics such as size, complexity, and quality are related to microservice centrality in the architectural network. This study investigates whether temporal centrality metrics can provide insight into the early detection of architectural degradation by correlating or affecting software metrics. We reconstructed the architecture of 7 releases of an OSS microservice project with 42 services. For every service in every release, we computed the software and centrality metrics. From one of the latter, we derived a new metric, Centrality Change Proneness. We then explored the correlation between the metrics. We identified 7 size and 5 complexity metrics that have a consistent correlation with centrality, while Centrality Change Proneness did not affect the software metrics, thus providing yet another perspective and an early indicator of microservice architectural degradation.

Problem

• Investigates temporal centrality metrics for early degradation detection
• Explores correlation between centrality and software metrics in microservices
• Introduces Centrality Change Proneness as an early degradation indicator

Innovation

• Temporal centrality metrics for degradation detection
• Centrality Change Proneness as new metric
• Correlation analysis of size and complexity metrics