A company is moving to the cloud and wants to size new Amazon EC2 instances accurately. It needs to gather metrics such as CPU, memory, and disk utilization, obtain an inventory of running processes on each VM, and monitor network connections to map inter-server communications. Which approach will collect this information MOST cost-effectively?

Deploying the AWS Application Discovery Service agent to each virtual machine provides the most comprehensive and cost-effective data for migration planning. The agent collects detailed system performance metrics (CPU, memory, disk), process lists, and network connection data, which is crucial for accurate EC2 instance sizing and dependency mapping. Configuring the CloudWatch agent would collect metrics but would not provide the process inventory or detailed network connection mapping needed for inter-server communication analysis. Enabling agentless discovery with Application Discovery Service is less granular and might miss specific process-level details or network connections that an agent can capture. Enabling Application Discovery Service in the console and configuring firewall rules without deploying agents won't collect the necessary detailed operational data from within the VMs.