Master thesis : Scalable Automated Deployment of Student Laboratory Environments in an OpenShift Cluster

Abstract

Computer-based laboratories play a major role for many courses, as they are often factored in student evaluation. However, configuration mismatch risks often make virtual machine backed laboratories the most convenient solution, but they are still cumbersome and require student setup. The goal of this thesis is to create a scalable lab-deployment system on the university's OpenShift cluster, requiring no prior setup from the students, but offering them pre-configured laboratory environments. The system presented relies on a two-operator architecture and uses declarative configuration for laboratories, relying directly on OpenShift for user authentication. It indirectly leverages ULiège's SSO and allows the use of role-based access control for managing laboratory environment access, for groups or individual students, with supervision if needed. The system achieves very good performance on the cluster at ULiège, creating 50 student environments in less than 10 seconds, tearing them down in less than a second, thanks to utilising native OpenShift and Kubernetes resources and design philosophy. The performance of the system allows for significant real-life workloads with several labs occurring at the same time, by leveraging an existing OpenShift cluster. This eliminates the problem of configuration mismatch and is accessible to all students without requiring them to set up a VM, simply through their browser or terminal emulator.