DevOps
What is Autoscaling?
Automatically adding or removing compute capacity based on load metrics, keeping performance steady while minimizing idle cost.
Definition
Autoscaling adjusts capacity to match demand without human intervention. Horizontal scaling adds or removes instances — AWS Auto Scaling Groups launch EC2 nodes when CPU crosses a threshold, and Kubernetes' Horizontal Pod Autoscaler adds pods based on CPU, memory, or custom metrics like queue depth. Vertical scaling resizes individual instances instead. Good policies include cooldown periods to prevent flapping, scheduled scaling for predictable peaks, and sensible minimums so a traffic spike never starts from zero. Serverless platforms like Lambda take the idea to its limit, scaling per request.