Static analysis learning of annotations in microservices

Ramirez Mendez, Francisco Miguel (2024). Static analysis learning of annotations in microservices. University of Birmingham. Ph.D.

RamirezMendez2024PhD.pdf
Text
Available under License All rights reserved.
Download (5MB)

Abstract

In microservices, an architectural style for building large-scale software applications, annotations play a crucial role in adding essential features for structuring and maintaining system settings. However, incorrect configurations and missing annotations impact the performance, quality and system complexity, posing significant concerns to developers. Moreover, the wrong usage of annotations generates potential bugs, and their detection may take days or even weeks due to the analysis of multiple logs and source code files.

To mitigate this, we advocate an approach to make suggestions for adding and keeping annotations according to similarities between microservice operations. The approach learns semantic relations between annotations and operations based on a database of code fragments with annotated operations. The learning process pursues converting operations into numerical vectors to find similar operations. Additionally, we extend the learning approach for creating clusters and identifying their granularity. Then, our approach predicts to which cluster an annotated operation belongs to identify its range of granularity values.

This thesis contributes to (i) a comprehensive systematic review of annotations in microservice construction, complemented by an empirical study that highlights the relevance of annotations in microservice software development; (ii) a semantics-driven learning approach that captures the relation between code fragments and annotations; and (iii) an extension of our learning approach that mines the granularity limits of new annotated operations.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):