THE POWER OF INTERNING: MAKING A TIME SERIES DATABASE 2,000X SMALLER IN RUST

Summary from the TLDR Web Dev newsletter

The interning design pattern involves replacing repeated data with integer IDs to drastically reduce memory usage. Through various optimizations including interning strings and arbitrary types, schema tuning like sorting sets and using enums, and custom serialization with Serde, this developer achieved a 2,000x reduction in data size of a time series database. The final result was a lightweight, append-only in-memory database.