SafeIaC: Análise e Reparação Automática para Infraestrutura como Código

This project's primary goal is to enhance Infrastructure as Code (IaC) systems' reliability. More specifically, we propose the first polyglot solution for reliable analysis and automated repair for Infrastructure as Code that provides formal correctness guarantees. Given the multitude of IaC technologies, the scattered ecosystem and the fact that it is very common for software projects to use more than one IaC technology, polyglot solutions that support multiple languages and environments are extremely valuable. Also, given the catastrophic impact that configuration errors can cause (see Description for examples), the development of tools that support automated repair and that provide formal guarantees is necessary. However, existing polyglot solutions do not support automated repair and do not provide any formal guarantees (Saavedra et al. 2022, 2023; Santolucito et al. 2017). On the other hand, solutions that provide formal guarantees, or that are based on formal models, are focused on a single IaC technology (Lepiller et al. 2021). SafeIaC will encompass both the versatility of polyglot support across various languages and environments and the reliability of formal correctness guarantees, addressing the needs of the industry by providing a singular solution to the previously unmet challenge of polyglot automated repair and reliable IaC management. This project faces significant challenges due to the diversity of Infrastructure as Code (IaC) technologies and the complexity of creating general yet impactful intermediate models. To tackle this, the project will leverage recent advancements in polyglot smell detection: in particular, we will use our own tool GLITCH and its intermediate model as the starting point (Saavedra et al. 2022, 2023). Addressing the issue of assigning formal semantics to these models is critical, given the varied semantics across IaC technologies, which influences error detection capabilities. The expertise in formal verification will be pivotal here: in particular, the team’s expertise on intermediate representations for formal verification (Alam et al, 2021). Additionally, automating program repair poses difficulties due to differing semantics; this will be approached by integrating insights from language parametric transformations (Koppel et al, 2018) and leveraging polyglot detection techniques. Given these challenges, our team is uniquely positioned to push the boundaries of the current state of the art and make a significant impact. In summary, our objectives are to: - Develop a polyglot analysis and repair framework for IaC capable of automatically detecting and repairing challenging problems that can have a high impact in practice, such as incorrect ordering constraints (Sotiropoulos et al., 2020) and configuration drifts (Weiss et al., 2017) . - Provide formal correctness guarantees about the results and repairs produced. - Ensure that the framework can be seamlessly incorporated into the current development and deployment pipelines. - Conduct empirical research on IaC practices with the goal of uncovering insights into the needs and challenges faced by the community.