A mobile game reads most metadata from an Amazon RDS DB instance. As popularity increased, metadata load times slowed and metrics indicate scaling the database alone won't help. The solutions architect must consider options that support snapshots, replication, and sub-millisecond response times. What should the solutions architect recommend?

The correct answer is to add an Amazon ElastiCache for Redis layer in front of the database. This solution directly addresses the slow metadata load times by caching frequently accessed data, providing sub-millisecond response times. Redis supports snapshots for data persistence and replication for high availability and read scaling. Migrating to Amazon Aurora with Aurora Replicas would improve database read performance but might not achieve sub-millisecond response times for all metadata queries, as it still involves database lookups. Migrating to Amazon DynamoDB with global tables is a NoSQL solution that offers low latency, but it's a significant database migration and might not be necessary if the core issue is read-heavy access to existing relational data. Adding an Amazon ElastiCache for Memcached layer is an option for caching, but Redis is generally preferred for its richer data structures, persistence (snapshots), and replication capabilities, which align better with the requirements.