A company hosts an online order-processing application on Amazon EC2 instances.

A scheduled scaling policy is appropriate only when workload changes occur at known and predictable times. The random slow responses indicate that application-tier demand does not consistently match the predefined schedule. A target tracking policy continuously monitors a metric such as average CPU utilization or Application Load Balancer request count and automatically adjusts the Auto Scaling group’s desired capacity to maintain the configured target. This allows the EC2 tier to respond to unexpected traffic increases without redesigning the application. Dedicated Hosts and placement groups do not provide elastic capacity for unpredictable demand. Migrating to DynamoDB would substantially change the application architecture. ElastiCache would introduce new infrastructure and is primarily effective for cacheable read traffic, whereas the reported operation creates new orders and is write-oriented. Target tracking is therefore the least disruptive solution.

================