Load Balancer — Traffic Distribution

01 · Core Component

Load Balancer — Traffic Distribution

Spreads incoming requests across multiple servers so no single machine gets overwhelmed. When one server dies, traffic is re-routed automatically.

REQUESTS

LOAD BALANCER nginx / AWS ALB / HAProxy

Server 1 us-east-1a

Server 2 us-east-1b

Server 3 us-east-1c

Load bars show CPU utilization in real-time

● LIVE

① Round Robin

Requests go to S1 → S2 → S3 → S1 in order. Simple, equal distribution.

② Least Connections

Route to server with fewest active connections. Better for variable request durations.

③ IP Hash

Hash user's IP to always route to the same server. Enables session stickiness.

④ Weighted Round Robin

Powerful servers get more requests. S1(weight 3) gets 3× more than S2(weight 1).

Netflix uses multiple LB layers: AWS ELB (Layer 4) + Zuul (Layer 7).
Cloudflare routes 100M+ req/s across Anycast network.
Google invented Maglev — handles 1M+ packets/sec per server.