EC2 instances in a fleet report CPU and network statistics to CloudWatch, but memory and disk utilization are missing. What is the standard way to publish those two metrics?
Show answer and explanation
Answer: B
The hypervisor can see CPU, network, and disk I/O counters, but it cannot see how much RAM or filesystem space the guest operating system is using. Those values come from inside the OS, so the CloudWatch agent must run on the instance and be configured to collect mem_used_percent and disk_used_percent. The agent publishes them as custom metrics under the CWAgent namespace, where they can be alarmed on like any other metric.
Why the other options are wrong
- A. Detailed monitoring only changes EC2 metrics from five-minute to one-minute resolution. It does not add guest-level counters that the hypervisor never collects.
- C. CloudTrail Insights looks for unusual patterns in API call volume. It does not read memory or disk state from an operating system.
- D. Compute Optimizer consumes existing utilization data to make sizing recommendations. It is not a metrics collector.
Exam tip. Memory and disk are guest-level metrics, so they need the CloudWatch agent.