Introducing TIN: full-text search for Postgres — PlanetScale
TIN is a fast, full-featured, full-text search index for Postgres
PostgreSQL Extension
Summary
Biscuit is a PostgreSQL index access method that accelerates LIKE and ILIKE pattern‑matching queries by storing bitmap-based index structures in memory. It adds utilities for inspecting the in‑memory footprint of indexes and supports multi‑column searches without the recheck cost of trigram indexes.
Highlights
- Provides a specialized IAM for fast LIKE/ILILE pattern matching, including multi‑column support.
- Stores primary index structures in memory with Roaring bitmap compression and offers SQL functions to inspect memory usage versus on‑disk size.
- Removes redundant bitmaps and unifies length‑based filtering for case‑sensitive and case‑insensitive searches, improving performance.
auto-generated
Context
Audience
Database administrators and developers who need high‑performance text search capabilities in PostgreSQL
DomainDatabase Performance
Formatopen source software repository
Accessfree online
Source Overview
Biscuit is a PostgreSQL index access method optimized for high-performance LIKE/ILIKE pattern matching, including multi-column searches, using bitmap position indices and length bitmaps to avoid heap rechecks.
It stores primary index structures in memory with Roaring bitmap compression and provides SQL utilities and a view to inspect in-memory vs on-disk index size. Usage involves creating biscuit indexes on text and other convertible types and benefits from multiple algorithmic optimizations such as negative indexing, wildcard skipping, and early termination on empty bitmap intersections.
Topics
Related
PostgreSQL GIN/GiST indexespg_trgm extensionCRoaring libraryB-Tree and hash indexes
