Types of Dynamic Auto Scaling Policies

Types of Dynamic Auto Scaling Policies Target Tracking Scaling Policy (most common, easiest) Works like a thermostat. You set a target metric (e.g., keep average CPU at 50%). Auto Scaling adds/removes instances to maintain that target. Example: If CPU goes above 50%, new instances launch. Step Scaling Policy Responds in steps based on thresholds. Example: If CPU > 70% for 5 min → add 1 instance. If CPU > 90% for 5 min → add 2 instances. If CPU < 30% → remove 1 instance. Simple Scaling Policy (legacy, rarely used now) Adds or removes a fixed number of instances when a CloudWatch alarm is triggered. Example: If CPU > 75%, add 2 instances. Difference from Other Scaling Types Dynamic Auto Scaling → Responds to real-time demand using metrics & policies. Predictive Auto Scaling → Uses machine learning to forecast future demand (e.g., traffic spikes at 9AM every weekday). Scheduled Scaling → You define fixed times for scaling (e.g., add 2 instances at 8AM, remove at 6PM).