SHF: Small: Towards a Unified Approach for DevOps Configuration Drift Detection

Modern software systems rely on Development and Operations (DevOps), a widely used approach that automates application build, test, and deployment. DevOps processes are managed through a set of configuration files that must remain consistent with the software code, computing environments, and external services on which they depend. Over time, inconsistencies - known as configuration drift - arise due to software evolution, environment changes, or misconfigurations, leading to system failures, delays, and increased maintenance effort. This project investigates how such inconsistencies emerge and how they can be identified and managed effectively. The project’s novelties are a systematic understanding of configuration drift across diverse software environments and DevOps tools, and the development of general, workflow-agnostic solutions that operate across different tools and workflows. The project's broader significance lies in improving the reliability and efficiency of software systems that support critical sectors such as healthcare, transportation, and technology, while also reducing the burden on developers who maintain them. The project establishes a foundation for analyzing and mitigating configuration drift through three integrated research efforts. It characterizes the contexts, categories, and impacts of drift using large-scale empirical analyses of real-world software repositories, providing a comprehensive view of this problem. It introduces a unified representation that captures heterogeneous configuration artifacts across different DevOps tools and domain-specific languages, enabling cross-system reasoning. Building on this representation, it develops automated, AI-based techniques to detect inconsistencies among configuration files, source code, and execution environments in a scalable and technology-agnostic manner. These advances contribute new knowledge, models, and tools for improving software reliability. The outcomes enhance developer productivity, support education and workforce development, and promote more dependable software systems for everyday use. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria. NSF Award ID: 2537321 | Program: 01002627DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Foyzul Hassan | Institution: Regents of the University of Michigan - Dearborn, Dearborn, MI | Award Amount: $600,000 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2537321 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2537321.html