Netflix consolidates relational database infrastructure on Amazon Aurora, achieving up to 75% improved performance

This article details Netflix's migration from self-managed distributed PostgreSQL to Amazon Aurora PostgreSQL, achieving significant performance and cost improvements.

• Netflix consolidated fragmented relational database infrastructure onto Aurora PostgreSQL
• Achieved up to 75% performance improvement and 28% cost savings across applications
• Spinnaker (Front50) reduced average latency by 50% (67.57ms to 41.70ms)
• Policy Engine reduced latency by 75% on key endpoints (26.72ms to 6.51ms)
• Aurora's log-based writes and shared storage architecture eliminated operational overhead
• Fully managed service removed manual deployment, patching, scaling, and maintenance burden
• PostgreSQL compatibility enabled minimal code changes and preserved developer productivity
• Aurora's higher memory allocation (75% vs 25-40%) improved query performance from cache
• Automated failover within 100ms eliminated complex manual recovery procedures

Netflix's migration demonstrates how Aurora consolidates database infrastructure while delivering measurable performance gains, cost savings, and reduced operational complexity for enterprise workloads.