Faster Database Performance With Denormalization (Fixed Audio)
In this episode, Scott discusses database denormalization which involves intentionally introducing redundant data to optimize read performance at the cost of write efficiency, addressing issues like slow queries caused by large, normalized datasets. Techniques include adding redundant columns, splitting tables (horizontally or vertically), using derived columns, or implementing mirrored tables and materialized views. While denormalization can significantly speed up queries, it should be used judiciously, only when performance issues arise that cannot be resolved through hardware improvements or simple optimizations.
Links:
HoneyBadger.io - https://HoneyBadger.io
Our Discord - / discord
00:00 Intro
00:52 What Is Denormalization?
01:47 When to Use Denormalization
02:38 HoneyBadger
03:27 What Are Techniques To Structure Denormalized Data?
03:36 Introducing a Redundant Column/Pre-joining Tables
04:20 Horizontal Table Splitting
04:56 Vertical Table Splitting
05:24 Adding Derived Columns
06:01 Using Mirrored Tables
06:21 Materialized views
06:51 How Do We Build Denormalized Data?
07:25 What You Need to Know
07:37 Outro