Auto-Scaling
Domain: MLOps
🎯 Motivation
Automatically adding or removing model instances based on demand. Scale up when busy, scale down when quiet.
📋 Concrete Example
Auto-scaling: peak hours (9am-5pm) -> 20 instances. Off-peak (2am) -> 2 instances. Saves money while maintaining performance. Automatic.
🔗 Analogy
A store opening more registers when busy - lunch rush: 5 registers open. Quiet afternoon: 1 register. Auto-scaling opens and closes model instances based on demand.