Reducing Costs and Improving Performance With Data Modeling in Postgres
summary
Shows concrete wins: inspects Postgres heap pages and the Free Space Map with hex dumps and an extension, demonstrates reduced table from 73MB to 52MB by reordering columns to eliminate padding and align fields, copies data with SELECT INTO to preserve rows, explains TOAST and per-page headers. Demonstrates query-only optimization that goes from 2000ms to 160ms (contest winner 120ms) without adding indexes, and shows how these changes save disk, memory and CPU. examines page headers and FSM column reordering to remove padding
AI-written summary · about
chapters (7)
- 0:00Start
- 7:10What is this talk about?
- 8:28Memory architecture/hierarchy
- 13:35Disk Data access
- 15:43HDD Performance
- 27:31Heap File in Postgres
- 31:23Page Layout
related talks
50mImproving Postgres' Efficiency
25mPostgreSQL performance tips you have never seen before
34mSimple and Awesome Database Tricks
41mOptimizing your app by understanding your PostgreSQL database
38mShaving off bytes at any scale: Space savings in various subsystems of PostgreSQL
23mBest Practices for Tuning Slow Postgres Queries
26mPostgreSQL performance tips you have never seen before V2.0
34mIn core compression-How to shrink your database