Question #782
A gaming company deployed 8 m5.12xlarge (general purpose) EC2 instances behind an Application Load Balancer. CloudWatch metrics indicated CPU and memory usage were 30% of anticipated levels. Now, demand fluctuates, and the company wants to implement Auto Scaling using these metrics. Which solution is most cost-effective?
Use m5.6xlarge instances. Min 2, desired 2, max 8.
Use m5.4xlarge instances. Min 3, desired 3, max 12.
Use m5.3xlarge instances. Min 4, desired 4, max 16.
Use m5.2xlarge instances. Min 6, desired 6, max 24.
Explanation
The correct answer is B because m5.4xlarge instances (16 vCPU) provide a cost-effective balance between instance size and scaling granularity. The current setup uses 8 m5.12xlarge instances at 30% utilization, indicating over-provisioning. By switching to smaller instances, Auto Scaling can adjust capacity more precisely. Option B's configuration (min 3, max 12) allows scaling in 16 vCPU increments, matching demand fluctuations efficiently. Larger instances (e.g., m5.6xlarge in A) scale coarsely, risking over-provisioning, while smaller instances (C, D) require more instances, increasing management overhead without cost savings. Since AWS pricing scales linearly per vCPU, B's mid-sized instances optimize cost by aligning scaling steps with demand, avoiding under/over-provisioning. Other options either lack granularity (A) or introduce unnecessary complexity (C, D).
Answer
The correct answer is: B