Load Balancing

Introduction

Load balancing distributes traffic across multiple servers for reliability and scalability.

Key Concepts

Upstream: Group of backend servers in Nginx.

Load Balancing Methods: Round-robin, least connections, IP hash.

Health Checks: Automatically remove unhealthy servers.

Deep Dive

Nginx Load Balancing

nginx
upstream django {
    # Round-robin (default)
    server 127.0.0.1:8001;
    server 127.0.0.1:8002;
    server 127.0.0.1:8003;
    
    # Or least connections
    least_conn;
    
    # Or IP hash (session persistence)
    ip_hash;
}

server {
    location / {
        proxy_pass http://django;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
    }
}

Weighted Servers

nginx
upstream django {
    server 127.0.0.1:8001 weight=3;  # Gets 3x traffic
    server 127.0.0.1:8002;
    server 127.0.0.1:8003 backup;     # Only when others fail
}

Best Practices

1. Use least_conn for varying request times: Better distribution.
2. Add health checks: Remove failed servers automatically.
3. Consider session persistence: Use ip_hash if needed.

Summary

Nginx load balancing distributes traffic across multiple Gunicorn instances. Use least_conn for most applications and add backup servers for reliability.