From Lab Chaos to Automated Discovery
A leading university's research department was hampered by slow, inconsistent, and manually-intensive processes for setting up computational environments. Researchers lost valuable time on IT tasks instead of focusing on discovery. prAIor implemented a fully automated, self-service platform that provisions containerized research environments in minutes, ensuring 100% reproducibility and accelerating the pace of innovation.
Project Goals
Reduce environment setup time from weeks to minutes
Guarantee 100% reproducible research environments
Implement a self-service model for researchers with IT oversight
Our Approach
**Step 1: Infrastructure as Code (IaC) Blueprint.** We analyzed the existing on-premise hardware and designed a robust, scalable cluster using Terraform and Ansible to automate the deployment of Proxmox VMs and a Kubernetes foundation.
**Step 2: Containerization & CI/CD.** We developed a standardized process using Docker and GitLab CI/CD, allowing researchers to define their required software stacks in simple configuration files, which were then automatically built into secure, reusable container images.
**Step 3: Self-Service Portal Integration.** We created a simple web interface connected to the Kubernetes API, enabling one-click deployment of pre-approved, isolated research environments (JupyterHub, R-Studio, etc.).
Solution
- **Kubernetes on Proxmox:** We built a cost-effective and powerful container orchestration platform on the university's existing hardware, avoiding expensive cloud provider costs.
- **Terraform & Ansible Automation:** Provided a complete Infrastructure as Code solution for provisioning and managing the entire stack, from bare metal to running applications, ensuring consistency and rapid recovery.
- **GitLab CI/CD Pipelines:** Automated the building, testing, and storing of custom research environment images, creating a 'single source of truth' for every experiment's software stack.
Results
- Researchers now launch complex computational environments in under 5 minutes.
- Eliminated 'it works on my machine' problems, leading to higher quality, verifiable research.
- The IT department gained full visibility and control over resources, improving security and optimizing hardware usage.
Contact
Do you have a question or want to discuss a project? Write to us.