A company plans a global image service where up to 10,000 users worldwide may upload images at peak. Each uploaded image must have text overlaid and then be published to the website. Which architecture should the solutions architect implement to handle uploads and processing at scale?

The correct architecture leverages S3 for scalable storage and SQS for reliable, decoupled processing. S3 event notifications trigger messages to SQS when new images are uploaded. A fleet of EC2 instances, scaled automatically based on SQS queue depth, processes these messages, ensuring efficient handling of varying loads. Processed images are stored in a separate S3 bucket, and CloudFront delivers them globally from this S3 origin, providing low latency and high availability. Option 1 is incorrect because EFS is not ideal for global, high-scale uploads, and using CloudWatch Logs for processing triggers is inefficient and not designed for this purpose. Option 2 is incorrect because SNS is a publish/subscribe service, not a message queue for worker pools, and using EFS for processed images is less scalable and cost-effective than S3. Option 4 is incorrect because EBS volumes are not shared across multiple EC2 instances in this manner, and Spot Instances are not suitable for critical, continuous processing without robust interruption handling.