sahil-barke01/datacenter_env

⚡ Datacenter Optimization Environment

A realistic OpenEnv simulation environment for datacenter operations. It models real-world infrastructure challenges like latency spikes, GPU overload, and cascading failures — designed for training AI agents to make real-time operational decisions such as scaling servers, rerouting traffic, and stabilizing distributed systems.

🌐 Live Demo: sahil-barke01-datacenter-env.hf.space

🧩 The Challenge

Modern datacenters must constantly balance:

Agents must learn to handle scenarios including latency spikes, GPU overload, and cascading failures — sometimes simultaneously.

⚙️ Action Space

{
  "action_type": "scale_up | restart_service | reroute_traffic | do_nothing",
  "target_servers": 1
}

📊 Observation Space

Each /step call returns:

{
  "cpu_usage": 85.2,
  "latency": 420.5,
  "error_rate": 0.12,
  "active_servers": 4,
  "reward": -3.2,
  "done": false
}

🧠 Reward Function

🔁 Episode Termination

🔌 API Endpoints

Interactive docs available at /docs once running.

🚀 Running Locally

Start server

uvicorn server.app:app --host 0.0.0.0 --port 8000

Then open: http://localhost:8000/docs

🐳 Docker Setup

# Build
docker build -t datacenter_env .
 
# Run
docker run -p 8000:8000 datacenter_env

🧱 Project Structure

datacenter_env/
├── server/
│   ├── app.py
│   └── datacenter_env_environment.py
├── models.py
├── client.py
├── inference.py
├── graders.py
├── demo.py
├── Dockerfile
├── openenv.yaml
└── README.md

🎯 Use Cases

• —Reinforcement Learning research
• —Infrastructure automation & optimization
• —AI-driven DevOps / SRE training
• —Autonomous system control

🔮 Future Improvements

• —Multi-cluster simulation
• —Advanced failure scenario composition
• —Cost-aware reward shaping
• —Pretrained baseline agents

⚙️ Tech Stack

Built with OpenEnv · FastAPI · Docker · Hugging Face Spaces