Load Balancing in System Design: Algorithms and Architecture Patterns

What is a Load Balancer?

A load balancer distributes incoming traffic across multiple servers — solving scalability and availability simultaneously.

---

Load Balancing Algorithms

Round Robin

Request 1 → Server 1
Request 2 → Server 2
Request 3 → Server 3
Request 4 → Server 1 (repeats)

Best for: identical servers, equal-cost requests

Least Connections

Server 1: 100 active connections
Server 2: 20 active connections  ← next request goes here
Server 3: 80 active connections

Best for: long-lived connections, variable request cost

---

Layer 4 vs Layer 7

---

Health Checks

---

Session Persistence

---

Key Takeaway

• Round robin for simple cases, least connections for variable workloads
• Layer 7 for most web apps — content-based routing, SSL termination
• Always implement health checks — automatic failure detection is critical
• Use shared session store (Redis), never sticky sessions