A web application on an EC2 Linux instance shows short spikes in CPU that last about 5 minutes several times each night. You need to capture which process ID is using the CPU during those spikes with as little effort as possible. What should you implement?

The CloudWatch agent's procstat plugin is designed to collect detailed metrics about individual processes, including CPU utilization per process ID (PID). By configuring this plugin, you can automatically capture which specific process is causing the CPU spikes without manual intervention. Setting up a Lambda function is overly complex for this task and requires custom scripting to identify PIDs and send notifications. SSHing into the instance nightly to run top is a manual, reactive approach that doesn't scale and requires constant monitoring. The default CloudWatch CPU utilization metric provides aggregate CPU usage for the instance but does not include per-PID information.