Question

A CloudOps engineer manages an e-commerce application whose request rate follows a strong daily and weekly pattern, with predictable spikes during business hours and much lower traffic overnight and on weekends. The team currently uses a static CloudWatch alarm on the RequestCount metric with a fixed threshold, but it generates false alarms during normal daily peaks and fails to catch abnormal drops during off-hours. The engineer needs an approach that adapts the alarm threshold to the metric's expected behavior throughout the day and week with minimal ongoing maintenance. Which solution meets these requirements?

Accepted Answer

Enable CloudWatch anomaly detection on the RequestCount metric and create an alarm based on the anomaly detection band.

Answer

Create multiple static alarms, one for each hour of the day, and use EventBridge scheduled rules to enable or disable them.

Answer

Configure a metric math expression that averages RequestCount over 24 hours and alarm when the current value exceeds that average.

Answer

Increase the static threshold to accommodate peak traffic and set the alarm to trigger only after 15 consecutive breaching datapoints.