www.youtube.com
AWS re:Invent 2025 - PostgreSQL performance: Real-world workload tuning (DAT410)
In this code talk, we dive deep into practical tuning techniques to avoid common pitfalls that silently degrade performance by improving an underperforming PostgreSQL database workload. Learn how excessive indexes hurt write throughput, why HOT updates fail, and how vacuum behavior can stall your system. We’ll demonstrate how to use Query Plan Management (QPM) and pg_hint_plan for plan stability and decode wait events to uncover hidden bottlenecks. ... … Otherwise, you'll see high storage and IOPs utilization. {ts:201} And if you have more indexes which are unused or duplicate, every modification to the database will lead to update those indexes unnecessarily, which is where you see storage and I obstacleization. And Postress SQL uses work memory to control the query operations such as
Related Pain Points2件
Query plan instability causes unpredictable performance degradation
7PostgreSQL query execution plans can become unstable, causing previously performing queries to suddenly degrade. Developers must use advanced tools like Query Plan Management (QPM) and pg_hint_plan to ensure consistent query performance.
Excessive or duplicate indexes degrade write performance and storage
6Unused or duplicate indexes cause every database modification to unnecessarily update those indexes, resulting in high storage utilization and IOPs consumption. This creates a silent performance drag that isn't immediately obvious.