Q22 — AWS DOP-C02 Ch.1

Question 22 of 100 | ← Chapter 1

A DevOps engineer is researching the most cost-effective way to implement an image processing cluster on AWS. The application cannot run in Docker containers and must run on Amazon EC2. The image processing workload checkpoints data to NFS and can tolerate interruptions. Configuring the cluster software on an EC2 Linux AMI takes 30 minutes.

• A. Use Amazon EFS for checkpoint data. Provision EC2 instances temporarily using an EC2 Auto Scaling group and the On-Demand pricing model.
• B. Use GlusterFS on EC2 instances for checkpoint data. Manually provision EC2 instances to run batch jobs and manually terminate them after job completion.
• C. Use Amazon EFS for checkpoint data. Launch EC2 Spot Instances using EC2 Fleet and use user data to configure the EC2 Linux instances at launch.
• D. Use Amazon EFS for checkpoint data. Launch EC2 Spot Instances using EC2 Fleet. Create a custom AMI with pre-installed cluster software and use the latest AMI when launching instances. ✓

Correct Answer: D. Use Amazon EFS for checkpoint data. Launch EC2 Spot Instances using EC2 Fleet. Create a custom AMI with pre-installed cluster software and use the latest AMI when launching instances.

Explanation

Option D is the most cost-effective solution. It combines low-cost EC2 Spot Instances with a custom AMI that eliminates the 30-minute configuration delay at launch — significantly reducing instance runtime and cost per job. Amazon EFS provides scalable, managed, shared storage ideal for checkpointing. Option A uses expensive On-Demand pricing. Option B incurs high labor cost and lacks automation/scalability. Option C still suffers the 30-minute setup delay on every Spot Instance launch, increasing total compute time and cost. Pre-baking software into an AMI is a well-established optimization for stateless, interruptible workloads like this.