A company’s stateful application runs in memory on Amazon EC2 instances. The infrastructure was deployed with AWS CloudFormation using the M5 EC2 instance family. As traffic increased, application performance degraded and users experience delays. Which solution will resolve the issues with the MOST operational efficiency?

The correct solution addresses both immediate performance degradation and future capacity planning with operational efficiency. Replacing M5 instances with R5 instances provides more memory, which is crucial for a stateful, in-memory application experiencing performance issues due to increased traffic. Modifying CloudFormation templates ensures infrastructure-as-code best practices and repeatable deployments. Deploying the CloudWatch agent to collect custom application latency metrics is essential because EC2's built-in metrics do not monitor application-level memory usage or latency, which are critical for stateful applications. This allows for proactive monitoring and informed scaling decisions. Incorrect options: T3 instances are burstable and generally not suitable for sustained high-performance, memory-intensive workloads, and manual console changes undermine CloudFormation's benefits. Manually increasing Auto Scaling Group capacity is not operationally efficient; automatic scaling based on metrics is preferred. CloudWatch's built-in EC2 memory metrics are for the instance's overall memory utilization, not the application's specific memory usage or latency, making them insufficient for precise capacity planning for an in-memory application.