Question

A SysOps administrator manages a web application behind an Application Load Balancer with an EC2 Auto Scaling group. Traffic patterns are unpredictable, and CPU utilization stays low even during peak periods because the workload is I/O bound. During traffic surges, users experience slow responses even though the CPU-based target tracking policy never triggers scale-out. The administrator needs the group to scale based on how much traffic each instance actually handles. Which scaling configuration BEST addresses this?

Accepted Answer

Create a target tracking scaling policy using the ALBRequestCountPerTarget predefined metric with a target value for requests per instance.

Answer

Create a step scaling policy triggered by an alarm on the ALB's total RequestCount metric.

Answer

Lower the target value on the existing CPU utilization target tracking policy so it scales out sooner.

Answer

Enable detailed monitoring on the instances so CPU metrics are published every minute.