Question

A media company is designing a new photo-sharing web application. Authenticated users upload high-resolution images (5 MB to 4 GB each) directly from browsers. Traffic is highly unpredictable, spiking during marketing campaigns from a few hundred to over 100,000 concurrent uploads. The architecture team wants uploads to bypass application servers entirely to avoid bottlenecks, must enforce that only authenticated users can upload, and needs the solution to scale automatically with zero capacity planning. Which approach best meets these requirements?

Accepted Answer

Use an API backend to generate S3 presigned URLs (with multipart upload for large files) after validating the user's authentication, and have browsers upload directly to Amazon S3 using those URLs.

Answer

Have the browser POST files to an Auto Scaling group of EC2 instances behind an Application Load Balancer, which then writes objects to Amazon S3.

Answer

Mount an Amazon EFS file system on a fleet of EC2 web servers and stream uploads through the servers to EFS, then copy objects to S3 asynchronously.

Answer

Configure browsers to write directly to a public S3 bucket with a bucket policy allowing anonymous PutObject, then use S3 Event Notifications to tag uploads by user.