Improving the Robustness, Efficiency, and Transparency of Continuous Integration and Delivery Pipelines

点击翻译按钮获取中文摘要

Context: Modern software is developed at a frantic pace. While the software releases of the past would take several months or even years to prepare, modern software organizations like Google, Uber, and Facebook release several times daily. To achieve this feat, organizations invest in a software release pipeline, i.e., resources, infrastructure, and scripts that specify how change sets (incremental improvements to a codebase produced by its developers to repair defects, enhance the software, or improve non-functional system properties) are processed through three high-level phases. First, the integration phase merges a change set with the central codebase. Second, the build phase orchestrates the execution of development tools that transform the updated codebase into a tested set of deliverables. Finally, the deployment phase automates the procedure that makes the tested set of deliverables available for software users to interact with or acquire. Overarching Goal and Specific Objectives: The frantic pace of modern software development introduces challenges that software organizations must address. To aid in that endeavour, the overarching goal of this research program is to enable and support the development, maintenance, and operation of intelligent software release pipelines. These improved pipelines will be capable of producing high quality releases at a rapid pace (i.e., Continuous Delivery), while providing decision makers within software organizations with additional insights that will help them to make more informed, data-driven decisions (i.e., Continuous Integration). More specifically, the objectives of the proposed research strive to develop concepts, techniques, and prototype tools to improve three desirable characteristics of software release pipelines: (RO1: Robustness) - Improve the reliability of pipeline signals and outputs (RO2: Efficiency) - Reduce waste in pipeline maintenance and execution (RO3: Transparency) - Cultivate awareness using pipeline decisions and data Impact: Investment in release pipelines has allowed the relatively small Mozilla organization (the developers of the Firefox web browser) to compete with software giants like Microsoft (Edge), Google (Chrome), and Apple (Safari). The proposed research will provide 12 students with hands-on training in large-scale empirical analyses and release pipelines. These students will be well-equipped for a research or private sector career in fields like devops, software release engineering, and data science, increasing Canada's innovation potential in these high-impact areas of software engineering.