Amazon, the company behind the cloud, underwent a massive migration from relational databases to NoSQL databases as part of its "Rolling Stone" project. The company found that the cost of relational database management systems was spiraling out of control and that most access patterns involved simple inserts, updates, or queries on a single row or table. By choosing to use its own NoSQL database and reworking the engagement model for training, Amazon was able to migrate all Tier 1 RDBMS workloads to NoSQL and reduce infrastructure costs by more than 50% while maintaining high business growth rates. The key to this success was understanding access patterns and modeling data in a way that mirrors those patterns, using techniques such as sharding, indexing, and multikey arrays to eliminate the need for complex joins and reduce time complexity. By adopting NoSQL, Amazon was able to create systems that are fast at high-velocity access patterns, reducing overhead and improving overall performance.