UCD Cluster VM Management Model (ONNX)

This is the first stable release of the DRL VM Management Agent, a Deep Reinforcement Learning model trained using Maskable PPO to optimize Virtual Machine placement and lifecycle management.

The model is exported in ONNX format for platform-independent deployment and comes with a full Python inference pipeline.

📦 What's Included

• Pre-trained Model: vm_management_agent.onnx (Supports up to 32 Hosts).
• Inference Engine: MLSysOpsVMManagementAgent class for easy integration.
• State Serializer: Converts complex JSON infrastructure states into the required 801-dim vector.
• Action Interpreter: Translates integer model outputs into human-readable decisions (e.g., "Create Large VM on Host 5").
• Demo Suite: Includes demo.py and generate_scenarios.py to verify model behavior.

⚠️ Model Constraints (Model Card)

• Scope: Single Datacenter.
• Scale: Maximum 32 Hosts.
• Resources: Optimization is based on CPU Cores only.
• Job Limits: Maximum job size is 8 Cores.

🚀 Quick Start

1. Clone and Install:

   git clone https://github.com/tgasla/MLSysOps-VM-Management-Agent.git
   cd MLSysOps-VM-Management-Agent
   pip install -r requirements.txt
   

2. Run the Demo:

   python demo.py
   

Full Documentation: See the README.md for integration details.