Leveraging Network Methods for Hub-like Microservice Detection

Context: Microservice Architecture is a popular architectural paradigm that facilitates flexibility by decomposing applications into small, independently deployable services. Catalogs of architectural anti-patterns have been proposed to highlight the negative aspects of flawed microservice design. In particular, the Hub-like anti-pattern lacks an unambiguous definition and detection method. Aim: In this work, we aim to find a robust detection approach for the Hub-like microservice anti-pattern that outputs a reasonable number of Hub-like candidates with high precision. Method: We leveraged a dataset of 25 microservice networks and several network hub detection techniques to identify the Hub-like anti-pattern, namely scale-free property, centrality metrics and clustering coefficient, minimum description length principle, and the approach behind the Arcan tool. Results and Conclusion: Our findings revealed that the studied architectural networks are not scale-free, that most considered hub detection approaches do not agree on the detected hubs, and that the method by Kirkley leveraging the Erdos-Renyi encoding is the most accurate one in terms of the number of detected hubs and the detection precision. Investigating further the applicability of these methods to detecting Hub-like components in microservice-based and other systems opens up new research directions. Moreover, our results provide an evaluation of the approach utilized by the widely used Arcan tool and highlight the potential to update the tool to use the normalized degree centrality of a component in the network, or for the approach based on ER encoding to be adopted instead.