Auto Scaling
Auto scaling policies and strategies for dynamic workloads.
Auto scaling is the process of automatically adding or removing servers (instances) based on current traffic or resource usage to maintain performance and optimize costs.
Auto Scaling
Define Metrics to Monitor Decide what triggers scaling. Common metrics: CPU utilization (e.g., > 70%) Number of requests per second Memory usage Custom metrics (like queue length)
Set Thresholds and Rules Create rules that say when to add or remove servers: Scale out (add servers) when metric goes above upper threshold Scale in (remove servers) when metric drops below lower threshold
Choose Minimum and Maximum Instances Set limits on how many servers can run: Minimum: baseline capacity (for low traffic) Maximum: upper limit (to avoid runaway costs)
Configure Health Checks Ensure only healthy servers receive traffic: Load balancer checks server status Auto scaling replaces unhealthy instances automatically