Storage

PlanetScale's managed Postgres and MySQL services leverage B-trees and B+trees for efficient data indexing and storage. This post details the internal mechanics of these data structures, their application in database indexing, and the implications of primary key choices on performance. It covers how B+trees are used in InnoDB for storing table data and secondary indexes, and analyzes the performance trade-offs between sequential integer primary keys and random UUIDs, particularly in the context of insertions and data locality. The post also touches on the impact of node size on disk block efficiency and the benefits of shallower trees for faster lookups.