A company is modernizing an application and migrating it to AWS. Currently, user profile data is stored as text in a single table in an on-premises MySQL database. After modernization, users will upload video files up to 4 GB each, and other users must be able to download those videos. The video storage must scale rapidly and must not degrade application performance. Which solution meets these requirements?

The correct solution leverages Amazon S3 for video storage due to its scalability, high availability, and ability to handle large objects (up to 5 TB), which is ideal for 4 GB video files without impacting application performance. Storing only the S3 object key in Amazon DynamoDB, a highly scalable NoSQL database, allows for efficient retrieval of video metadata while keeping the large video data separate. AWS DMS and AWS SCT facilitate the migration of the existing MySQL database to DynamoDB. Storing videos as base64-encoded strings in a database (either Aurora PostgreSQL or DynamoDB) is inefficient and can degrade performance, as databases are not optimized for large binary objects. While Amazon Keyspaces is a scalable NoSQL option, DynamoDB is generally a more common and often simpler choice for key-value and document workloads on AWS, especially when migrating from a relational database and integrating with S3.