Auto-Scaling, Explained

Why fixed capacity always loses to real traffic, how auto-scaling decides when to add or remove servers, and the gotchas that catch teams who turn it on and walk away without a load balancer.

1. Why You'd Want This at All The peak-vs-average traffic problem — why provisioning for your busiest moment is expensive and provisioning for average traffic is risky.
2. How It Actually Decides to Scale Metrics, thresholds, cooldown periods, and scaling policies — the mechanism that turns a number like CPU usage into an actual scale-up or scale-down event.
3. The Gotchas Cold-start lag, the thundering herd of a spike arriving before new capacity is ready, and why auto-scaling only works paired with a load balancer.