A solutions architect is designing a stateless application to run on AWS. They created an AMI and launch template for the application. Jobs must be processed in parallel, and EC2 instances should be added and removed automatically based on workload. The application must be loosely coupled and job items must be durably stored. Which solution meets these requirements?

The correct solution uses Amazon SQS for durable storage of job items and loose coupling, as SQS is a message queuing service designed for this purpose. An Auto Scaling group dynamically adjusts EC2 instance capacity based on the SQS queue depth, ensuring parallel processing and automatic scaling according to workload. This scaling policy directly addresses the requirement for adding and removing instances based on the actual number of jobs awaiting processing. Incorrect options: Using Amazon SNS for job processing is unsuitable because SNS is a publish/subscribe service, not a queue for durable storage and processing by a pool of workers. Scaling based on CPU usage is a generic metric and might not accurately reflect the number of jobs needing processing. Scaling based on network usage is also a generic metric and doesn't directly correlate with the number of jobs in a queue. Scaling based on the number of messages published to an SNS topic is incorrect because SNS is for notifications, not for holding jobs for processing by a worker pool.