What should a solutions architect recommend to meet these requirements?
Configure Amazon EC2 Auto Scaling to use a scheduled scaling plan and launch an additional 8 EC2 instances during business hours.
Configure AWS Auto Scaling to use a scaling plan that enables predictive scaling. Configure predictive scaling with a scaling mode of forecast and scale, and to enforce the maximum capacity setting during scaling.
Configure a step scaling policy to add 4 EC2 instances at 50% CPU utilization and add another 4 EC2 instances at 90% CPU utilization. Configure scale-in policies to perform the reverse and remove EC2 instances based on the two values.
Configure AWS Auto Scaling to have a desired capacity of 5 EC2 instances, and disable any existing scaling policies. Monitor the CPU utilization metric for 1 week. Then create dynamic scaling policies that are based on the observed values.
Explanations:
This option suggests using a scheduled scaling plan to launch 8 additional instances during business hours. This approach can lead to over-provisioning since it does not account for actual demand and can result in unnecessary costs, especially if traffic fluctuates during those hours.
This option recommends using predictive scaling, which leverages historical data to forecast traffic patterns and adjusts capacity accordingly. By enforcing maximum capacity settings during scaling, this method helps avoid over-provisioning while effectively responding to demand increases.
While step scaling allows for more granular scaling based on CPU utilization, adding 4 instances at 50% CPU utilization may lead to over-provisioning, especially if the demand does not require that many instances. The scale-in policy would need to be carefully monitored to prevent unnecessary costs.
This option proposes a static approach by setting a desired capacity of 5 instances and monitoring CPU utilization for a week. Although this could provide insights, it delays the scaling response to demand and does not adapt to real-time traffic patterns, potentially leading to under-provisioning during peak times.
I classify that the answer is:
Configure AWS Auto Scaling to use a scaling plan that enables predictive scaling. Configure predictive scaling with a scaling mode of forecast and scale, and to enforce the maximum capacity setting during scaling.