Autoscaling

Problem

Your application needs:
02:00 → 2 servers
12:00 → 20 servers
Black Friday → 100 servers

Keeping 100 running permanently wastes money. Keeping two guarantees overload.

Automatically adjust capacity based on demand or other signals.

Load ↑ → instances ↑

Load ↓ → instances ↓

Core idea: autoscaling adjusts capacity as workload requirements change.