Case Study: Amazon.com achieves 50% lower latency and 90% reduced operations cost while scaling Wallet to 10+ billion records with Amazon Web Services (Amazon DynamoDB)

Amazon Wallet Service Scales Using Amazon DynamoDB

Amazon’s Wallet service, which securely stores customers’ payment methods and handles as many as five billion transactions daily, was running on a complex, horizontally partitioned Oracle setup that required frequent, time-consuming scaling and heavy operational overhead. Rapid growth and replication challenges forced regular vertical/horizontal scaling and consumed months of engineering and DBA effort each year, driving the team to seek a simpler, more scalable solution.

The team migrated Wallet to Amazon DynamoDB (with AWS CloudFormation and Amazon Redshift for analytics), adding a lightweight application interface for the NoSQL model and using DynamoDB features like auto scaling, TTL, and point-in-time recovery. The zero-downtime, three-step migration moved more than 10 billion records (consolidating 8 Oracle tables into 6 DynamoDB tables) and delivered a 50% latency reduction, 40% higher transactions per second per host, and a 90% cut in operations effort, freeing the team to focus on features.

Amazon.com