Which solution will provide EC2 instances to meet these requirements MOST cost-effectively?
Purchase a 1-year Savings Plan for Amazon EC2 that covers the instance family of the Auto Scaling group that the batch job uses.
Purchase a 1-year Reserved Instance for the specific instance type and operating system of the instances in the Auto Scaling group that the batch job uses.
Create a new launch template for the Auto Scaling group. Set the instances to Spot Instances. Set a policy to scale out based on CPU usage.
Create a new launch template for the Auto Scaling group. Increase the instance size. Set a policy to scale out based on CPU usage.
Explanations:
A Savings Plan provides a discount on EC2 usage, but it does not automatically optimize cost for batch jobs that run only at specific times. It may not be the most cost-effective for this use case.
Reserved Instances require a long-term commitment, which may not be ideal for batch jobs with variable usage. They are more suited for consistent workloads, not for instances that only run at specific times.
Spot Instances are the most cost-effective for workloads that can tolerate interruptions. Since the batch jobs run only at specific times, Spot Instances would provide significant savings, and scaling out based on CPU usage is appropriate for auto-scaling.
Increasing the instance size and using CPU-based scaling will not address cost optimization and does not take advantage of cheaper instance types like Spot Instances. Scaling out based on CPU usage might increase costs unnecessarily.
To the best of my knowledge, the answer is:
Create a new launch template for the Auto Scaling group. Set the instances to Spot Instances. Set a policy to scale out based on CPU usage.
I draft that the answer is:
Create a new launch template for the Auto Scaling group. Set the instances to Spot Instances. Set a policy to scale out based on CPU usage.