A company runs a proprietary stateless Linux binary (source code not available) on an EC2 instance. The app is single-threaded, requires 2 GB of RAM, is CPU intensive, and executes roughly every 4 hours for up to 20 minutes. You need to redesign the architecture to meet these requirements. Which approach should you choose?

The correct answer is to run the binary as a container on AWS Fargate and schedule the Fargate task using Amazon EventBridge. Fargate is suitable because it allows running containers without managing servers, and the application's stateless nature and infrequent execution align well with a containerized, on-demand approach. EventBridge provides a reliable way to schedule the task every four hours. Using AWS Lambda is incorrect because Lambda functions have a maximum execution duration of 15 minutes, and this binary can run for up to 20 minutes. AWS Batch is designed for large-scale, high-throughput workloads, which is overkill for a single-threaded, infrequent task. EC2 Spot Instances with CodeDeploy are less ideal as they require server management and CodeDeploy is primarily for application deployment, not scheduled task execution.