AWS Load Balancer

• A fully managed load balancing system
• Not a single server, it's a distributed system
  • Multiple lb nodes at different availability zones
  • Auto managed these nodes
• Automatically:
  • Scales capacity
  • Handles failures
  • Distributes traffic

- Multiple LB nodes
- Anycast IP (very hard to implement)
- Dynamic scaling scripts
- Service discovery

User (Malaysia) → Singapore LB
User (US)       → US LB
User (Europe)   → EU LB
(connect to nearest lb node)

LB → App servers

Auto Scaling Works

Load Balancer scaling

• AWS auto:
  • Adds more LB nodes when traffic increases
  • Removes them when traffic drops

Backend (instances) scaling

• Done using AWS Auto Scaling
• Scales your:
  • EC2 instances (app servers)

VPS “Closer to Reality” Setup

Basic setup:
User → VPS (NGINX) → App servers

------------------------------------

Advanced:
           → LB1 (NGINX)
User → DNS
           → LB2 (NGINX)

LB → App servers

Add:

• Health checks
• DNS failover (via Cloudflare)