Question

A CloudOps engineer manages an Application Load Balancer fronting a fleet of EC2 instances. Users report that some requests are extremely slow even though the CloudWatch dashboard widget showing the Average of the TargetResponseTime metric appears healthy at around 120 ms. The team wants the dashboard to accurately surface the slow-request experience affecting a subset of users. What should the engineer do to the dashboard widget?

Accepted Answer

Change the widget statistic to the p99 percentile of TargetResponseTime to reveal tail latency affecting the slowest requests.

Answer

Change the widget statistic to Sum of TargetResponseTime so the total time reflects the slow requests.

Answer

Add the RequestCount metric to the same widget so spikes in traffic correlate with the average latency.

Answer

Enable detailed monitoring on the ALB to increase the metric resolution to one minute for the average statistic.