Chronicles of CI/CD: A Deep Dive into its Usage Over Time

TL;DR

This paper analyzes GitHub repositories to understand the evolution and current usage of CI/CD technologies over the past 12 years, revealing trends, multi-technology usage, and change patterns.

Contribution

It provides a comprehensive analysis of CI/CD technology adoption over time using GitHub data, highlighting multi-technology usage and evolution patterns.

Findings

Developers often use multiple CI/CD technologies simultaneously.

Technology usage in CI/CD pipelines has evolved significantly over 12 years.

Changing CI/CD tools is a common practice among developers.

Abstract

DevOps is a combination of methodologies and tools that improves the software development, build, deployment, and monitoring processes by shortening its lifecycle and improving software quality. Part of this process is CI/CD, which embodies mostly the first parts, right up to the deployment. Despite the many benefits of DevOps and CI/CD, it still presents many challenges promoted by the tremendous proliferation of different tools, languages, and syntaxes, which makes the field quite challenging to learn and keep up to date. Software repositories contain data regarding various software practices, tools, and uses. This data can help gather multiple insights that inform technical and academic decision-making. GitHub is currently the most popular software hosting platform and provides a search API that lets users query its repositories. Our goal with this paper is to gain insights into the technologies developers use for CI/CD by analyzing GitHub repositories. Using a list of the state-of-the-art CI/CD technologies, we use the GitHub search API to find repositories using each of these technologies. We also use the API to extract various insights regarding those repositories. We then organize and analyze the data collected. From our analysis, we provide an overview of the use of CI/CD technologies in our days, but also what happened in the last 12 years. We also show developers use several technologies simultaneously in the same project and that the change between technologies is quite common. From these insights, we find several research paths, from how to support the use of multiple technologies, both in terms of techniques, but also in terms of human-computer interaction, to aiding developers in evolving their CI/CD pipelines, again considering the various dimensions of the problem.