What should the solutions architect do to meet these requirements?
Configure a predictive scaling policy with the appropriate scaling metric.
Configure a dynamic target tracking scaling policy with the appropriate scaling metric.
Configure a scheduled scaling policy that launches additional EC2 instances during business hours.
Configure dynamic step or simple policies with Amazon CloudWatch alarms to add and remove EC2 instances based on alarm status.
Explanations:
A predictive scaling policy forecasts future traffic and scales the instances accordingly. While this may help during expected traffic spikes, it does not provide immediate scaling response to real-time latency issues, which could lead to delays during high traffic periods.
A dynamic target tracking scaling policy adjusts the number of EC2 instances in response to current metrics, such as CPU utilization or request count, ensuring that the application can handle increased load promptly while avoiding over-provisioning. This is optimal for addressing user latency concerns during high traffic.
A scheduled scaling policy would add instances at predetermined times, which may not align with actual traffic patterns. This could lead to insufficient capacity if traffic exceeds expectations or unnecessary instances if traffic is lower than anticipated, failing to address real-time latency issues effectively.
Dynamic step or simple policies based on CloudWatch alarms can add or remove instances based on specific thresholds but may not be as responsive as a target tracking policy. They require more manual configuration and may not effectively handle rapid fluctuations in traffic, which can still lead to latency concerns.
It seems to me that the answer is:
Configure a dynamic target tracking scaling policy with the appropriate scaling metric.