A SysOps administrator set up CloudWatch alarms for DiskReadBytes and DiskWriteBytes on an EC2 instance to alert based on read/write activity for attached EBS volumes. A local monitoring tool on the instance shows the volume I/O exceeded thresholds, but the CloudWatch alarms never entered the ALARM state. What change will make the CloudWatch alarms report correctly?

The correct answer is to change the CloudWatch alarms to use VolumeReadBytes and VolumeWriteBytes metrics for the EBS volumes. DiskReadBytes and DiskWriteBytes are instance-level metrics that measure I/O from the instance's perspective, which may not accurately reflect the activity of specific attached EBS volumes, especially if there are multiple volumes or if the instance's internal caching affects the reported disk I/O. VolumeReadBytes and VolumeWriteBytes are EBS-specific metrics that directly measure the I/O operations on the EBS volume itself, providing a more accurate representation of its activity. Installing the CloudWatch agent is incorrect because DiskReadBytes and DiskWriteBytes are standard EC2 metrics that do not require the agent. The CloudWatch agent is used for collecting custom metrics, logs, and detailed monitoring metrics not natively provided by EC2. Installing the AWS Systems Manager Agent is incorrect because SSM Agent is primarily for managing instances (e.g., patching, running commands) and does not directly collect or publish CloudWatch metrics for EBS volumes. Changing the alarms to use VolumeReadBytes and VolumeWriteBytes for the EC2 instance is incorrect because these metrics are associated with the EBS volume resource, not the EC2 instance resource directly.