Question

A CloudOps engineer is troubleshooting intermittent CPU spikes on a fleet of EC2 instances that last around 90 seconds. When reviewing the default CloudWatch metrics, the spikes are hard to correlate because data points are too coarse. The team needs finer-grained visibility into these short-lived spikes without installing additional software. What should the engineer do?

Accepted Answer

Enable detailed monitoring on the instances so CloudWatch delivers metrics at 1-minute granularity

Answer

Install the CloudWatch agent and configure a 10-second collection interval for CPU metrics

Answer

Create a high-resolution custom metric alarm with a 10-second period on the default EC2 CPUUtilization metric

Answer

Enable CloudWatch anomaly detection on CPUUtilization to automatically surface the spikes