Improve Aurora PostgreSQL throughput by up to 165% and price-performance ratio by up to 120% using Optimized Reads on AWS Graviton4-based R8gd instances

This article demonstrates significant performance improvements for Aurora PostgreSQL using Graviton4-based R8gd instances with Optimized Reads tiered cache.

• Upgrading from Graviton2 R6g to Graviton4 R8gd achieves 165% higher throughput
• Price-performance ratio improves by up to 120% compared to previous generation
• Application response time (p99) improves by up to 80% with greater consistency
• Aurora PostgreSQL 17.5 supports up to 256 TiB storage in single cluster
• Optimized Reads extends cache capacity up to 5x instance memory using local NVMe SSDs
• R8gd scales up to 48xl with 1.5 TiB memory and 10.4 TiB local NVMe capacity
• Ideal for internet-scale apps, real-time analytics, generative AI with pgvector workloads
• Existing clusters can upgrade via instance modification or blue/green deployments

Aurora PostgreSQL 17.5 on Graviton4 R8gd with Optimized Reads delivers substantial performance and cost improvements for read-heavy and write-intensive database workloads.