AI-Powered Load Balancer

This project implements an intelligent load balancer that uses AI agents to dynamically distribute incoming traffic across a set of backend nodes. The agents, built with crewai and xai-sdk, analyze the current traffic and node health to make smart routing decisions.

Features

• Intelligent Traffic Distribution: Uses AI agents to decide the best node for each request.
• Node Health Monitoring: Actively checks the status of each backend node.
• Scalable Agent-based Architecture: Built with crewai for a modular and extensible agent design.
• Load Testing: Includes a k6 script to simulate traffic and test the load balancer's performance.

Getting Started

Prerequisites

• Python 3.10+
• k6

Installation

1. Clone the repository:

   git clone <repository-url>
   cd load_ai2

2. Install Python dependencies:

   pip install -r requirements.txt

3. Set up your XAI API key:

   • Create a .env file in the root directory.
   • Add your XAI API key to the .env file:

     XAI_API_KEY="your-api-key"
     

Usage

1. Start the backend nodes and the load balancer:

   ./run.sh

   This will:

   • Start the backend nodes on ports 8001-8005.
   • Start the main load balancer on port 8000.
   • Run a k6 load test.

2. Send requests to the load balancer: You can use the provided k6_script.js or send your own requests to http://localhost:8000/.

Load Testing

The k6_script.js file is configured to simulate a load test with the following stages:

• Ramp-up: 100 virtual users over 1 minute.
• Sustain: 100 virtual users for 3 minutes.
• Ramp-down: 0 virtual users over 1 minute.

To run the load test, simply execute the run.sh script.