Auto Scaling
Auto scaling automatically adds or removes instances based on load metrics, in horizontal (more instances) or vertical (larger sizes) forms.
Purpose
Automatically adapt to load fluctuations, optimizing cost and performance.
Mechanism
Scale out: add instances when a metric exceeds a threshold (e.g., CPU>70%). Scale in: remove instances when it drops (e.g., CPU<30%), with cooldown periods to prevent flapping. Predictive scaling: pre-scales by time patterns. Implemented by K8s HPA/VPA and AWS ASG.
Use cases
- Elastic scaling
- Cost optimization
- Load adaptation