A company is migrating an Oracle database to AWS. The database contains a single table with millions of high-resolution GIS images, each identified by a geographic code. During natural disasters, tens of thousands of images are updated every few minutes. Each geographic code maps to a single image. The company requires a highly available, scalable, and cost-effective solution for these events. Which solution best meets these requirements?

Storing images in Amazon S3 buckets is cost-effective for large objects and offers high availability and scalability. Amazon DynamoDB provides a highly scalable, high-performance NoSQL database, ideal for storing key-value pairs like geographic codes and S3 URLs. Its ability to handle tens of thousands of updates per minute makes it suitable for the described workload during natural disasters. Storing images directly in a relational database like Oracle on RDS (option 1 and 4) is generally inefficient and expensive for large binary objects, and can lead to performance bottlenecks. Storing images directly in DynamoDB (option 3) is not recommended as DynamoDB is optimized for smaller items, not large binary objects like high-resolution images. While DAX improves read performance, it doesn't address the fundamental issue of storing large images within DynamoDB.