D is the correct answer. "in-memory tasks" => need the "R" EC2 instance type to archive memory optimization. So we are concerned about C & D. Because EC2 instances don't have built-in memory metrics to CW by default. As a result, we have to install the CW agent to archive the purpose.

It's D, EC2 do not provide by default memory metrics to CloudWatch and require the CloudWatch Agent to be installed on the monitored instances : https://aws.amazon.com/premiumsupport/knowledge-center/cloudwatch-memory-metrics-ec2/

B will reduce operational overhead and better solution than keep changing the family. Also, I don't think the exam will require you to remember instance families like M5 and R5

D .....

R5 instances are better optimized for the in-memory workload than M5. Auto Scaling alone doesn't handle stateful applications well, manual capacity adjustments would still be needed. Custom latency metrics give better visibility than built-in metrics for capacity planning.

By replacing the M5 instances with R5 instances, which are optimized for memory-intensive workloads, the application can benefit from increased memory capacity and performance. In addition, deploying the CloudWatch agent on the EC2 instances allows for the generation of custom application latency metrics, which can provide valuable insights into the application's performance. This solution addresses the performance issues efficiently by leveraging the appropriate instance types and collecting custom application metrics for better monitoring and future capacity planning. A. Replacing with T3 instances may not provide enough memory capacity for in-memory tasks. B. Manually increasing the capacity of the ASG does not directly address the performance issues. C. Relying solely on built-in EC2 memory metrics may not provide enough granularity for optimizing in-memory tasks. The most efficient solution is to modify the CloudFormation templates, replace with R5 instances, and deploy the CloudWatch agent for custom metrics.

Option D is the correct answer.

will go for C

Would go with D

ı think D