HTTP Load Balancer

A multi-threaded TCP-based HTTP load balancer implemented in C++ for distributing client requests across backend servers with automatic health monitoring and real-time statistics.

Core Implementations

1. Weighted Least Connections Algorithm — Intelligent traffic distribution based on current server load
2. Automatic Health Monitoring — Continuous health checks with failover capabilities
3. Real-time Statistics — Live monitoring of connections, requests, and server status
4. Multi-threaded Architecture — High-performance concurrent connection handling

Workflow Diagram

Diagram created using StarUML 7.0.0

How to Run

Prerequisites

• C++14 compatible compiler (g++)
• Python 3.x (for test backend servers)
• Linux/Unix environment

Build

# Clone the repository
git clone https://github.com/psychomita/load-balancer.git
cd load-balancer

# Build the project
make

Usage Examples

# Equal Weight Distribution (as shown in the diagram) 
./bin/load_balancer 8080 127.0.0.1:9001 127.0.0.1:9002 127.0.0.1:9003

# Weighted Distribution
./bin/load_balancer 8080 127.0.0.1:9001:2 127.0.0.1:9002:1 127.0.0.1:9003:1

# With Connection Limits
./bin/load_balancer 8080 127.0.0.1:9001:2:500 127.0.0.1:9002:1:200

Start Test Servers in the Backend

# Terminal 1
python3 test_server.py 9001 9002 9003

Start Load Balancer

# Terminal 2
./bin/load_balancer 8080 127.0.0.1:9001 127.0.0.1:9002 127.0.0.1:9003

Test the Load Balancer

# Terminal 3

# Single request
curl http://localhost:8080

# Multiple requests to see load balancing
for i in {1..1000}; do curl -s http://localhost:8080 > /dev/null & done
wait

Demo

Test Servers (written in Python for testing purpose)

Test servers run on ports 9001, 9002 and 9003, serving content about Bjarne Stroustrup

Health Check and Statistics Display

• When all backend servers are running, they are marked as HEALTHY.
• When all servers are stopped, the load balancer reports no healthy servers available.
• Metrics tracked are Total requests processed, Number of healthy servers and Server-wise statistics.

Load Balancing

The load balancer uses a Weighted Least Connections algorithm:

1. Health Check: Only healthy servers are considered
2. Weight Calculation: effective_load = active_connections / weight

Requests from port 8080 are routed to backend servers 9001, 9002 and 9003

Core Concepts Implemented

• Operating Systems: Process Management, Memory Management, Inter-Process Communication
• Computer Networks: Socket Programming, HTTP Protocol, Network Architecture
• Data Structures & Algorithms: Vector Containers, Load Balancing Algorithm, Concurrency Control

For visual reference, watch the Demo Video