Evaluating Infrastructure-as-Code tools for lifecycle management

Sammanfattning

Infrastructure as Code (IaC) has become a foundational practice in modern DevOps, enabling developers to define and manage infrastructure through version-controlled code rather than manual configuration. However, the growing number of IaC frameworks raises practical questions about which tool to choose for a given context, and how the choice affects both the developer experience and the technical characteristics of the deployment. This thesis evaluates four widely used IaC frameworks, Terraform, Pulumi, Ansible, and Crossplane, within a CI/CD context, with the aim of clarifying the trade-offs each framework presents and the role that paradigm choice plays in the resulting deployment lifecycle. A mixed-methods approach was used. Ten qualitative interviews with software developers at Ericsson explored developer perception of scalability, maintainability, reproducibility, and the broader impact of paradigm choice on the developer experience. In parallel, a technical benchmark was conducted across two deployment scenarios and a set of lifecycle actions, measuring CPU, memory, network usage, and execution time for each framework. The benchmarks ran inside a standardized Kubernetes-based testbed designed to isolate framework characteristics from environmental noise, with statistical validation applied to the resulting data. The findings indicate that no single framework is universally superior. While IaC was unanimously perceived as beneficial for productivity and reproducibility, the differences between frameworks are tied to their architectural choices rather than to the declarative or imperative paradigm alone. The interviews suggested that developer perception of scalability and maintainability is shaped more by framework features such as modularization support and community ecosystem than by paradigm itself. The benchmarks revealed substantial performance variation both between paradigms and within them, with each framework presenting distinct trade-offs between execution speed, predictability, and resource overhead. Together, the results suggest that organizations should evaluate IaC frameworks on a project-by-project basis, weighing specific features and architectural characteristics against their deployment context and team experience, rather than relying on paradigm as the primary criterion.