Case Study: Amazon.com achieves scalable fraud detection and halves database costs with Amazon Web Services (AWS)

Amazon.com Buyer Fraud Service Gains Scalability, Cuts Costs in Half Using AWS

Amazon.com’s Transaction Risk Management Services (TRMS) team runs the Buyer Fraud Service (BFS), a real-time, machine‑learning system that analyzes over 2,000 data points per order to screen hundreds of purchases per second. Faced with more than 100 on‑premises Oracle databases holding 40 TB of data, the team struggled with high administrative overhead (three full‑time DB engineers), peak‑load latency, lengthy provisioning, and the need to maintain 99.995% availability as transaction volume grew.

TRMS re‑platformed BFS to Amazon Aurora (PostgreSQL‑compatible) and used AWS Database Migration Service to replicate data during cutover. The migration moved 40 TB in six months with only one hour of downtime, scaled to 900 transactions per second per shard with minimal CPU, eliminated database outages, cut database administration ~70%, and delivered Oracle‑level performance at less than half the cost.

Amazon.com

Josh Gage

Senior Software Development Engineer