| # OpenVinayaka v3.5: Production Distributed Cluster |
|
|
| This folder contains the **Microservices Architecture** for running OV-Engine at scale. It replaces monolithic inference with a swarm of "Fractal Shards." |
|
|
| ## 🏗️ Architecture |
| 1. **Router (Port 8000):** OpenAI-compatible API Gateway. Broadcasts queries to all shards. |
| 2. **Shards (Ports 8001+):** Independent Vector Memory Units. Each shard holds a specific domain (Science, History, etc.) and calculates `P = S * C * R * W` locally. |
|
|
| ## 🚀 How to Run (Docker) |
| This is the preferred method. It creates a local cluster instantly. |
|
|
| ```bash |
| # 1. Build and Start the Cluster |
| docker-compose up --build |
| |
| # 2. Test the API (OpenAI Compatible) |
| curl http://localhost:8000/v1/chat/completions \ |
| -H "Content-Type: application/json" \ |
| -d '{ |
| "model": "ov-cluster-v1", |
| "messages": [{"role": "user", "content": "What is the speed of light?"}] |
| }' |
| ``` |
|
|
| ## 📦 Manual Run (No Docker) |
| If you want to run it manually (e.g., for dev), install dependencies first: |
| ```bash |
| pip install fastapi uvicorn aiohttp sentence-transformers numpy |
| python3 run_local_cluster.py |
| ``` |
|
|
|
|
