Biscuit Documentation

Welcome to the official documentation for Biscuit.

Contents:

Biscuit Index for PostgreSQL

High-Performance Pattern Matching Index for LIKE Queries

What is Biscuit?

Biscuit is a specialized PostgreSQL index access method designed to dramatically accelerate LIKE pattern matching queries. Unlike traditional B-tree or GIN indexes, Biscuit uses bitmap-based character indexing to provide near-instantaneous pattern matching across single or multiple columns.

Key Features

  • Blazing Fast: Optimized for complex LIKE patterns including %, _, prefixes, suffixes, and substrings

  • Multi-Column Support: Query across multiple columns simultaneously with intelligent predicate reordering

  • Aggregate Optimization: Special optimizations for COUNT(*) and EXISTS queries

  • Full CRUD Support: INSERT, UPDATE, DELETE operations with automatic index maintenance

  • Crash-Safe and Replicated: WAL-logged storage, point-in-time recovery, and physical streaming replication

  • Smart Query Planning: Automatic reordering of predicates based on selectivity analysis

  • Memory Efficient: Uses Roaring Bitmaps for compact in-memory representation

Biscuit suits read-mostly, analytical workloads: load data, build the index, then query. Writes against a live index generate considerably more WAL than the heap writes alone, and each backend holds its own cached copy of the index, so plan for connection-pool size. See Performance Tuning.

What’s New — 3.0.0

Durability

  • WAL-logged, crash-safe storage. All index state lives in the index relation’s own pages and is WAL-logged, replacing the earlier external-file snapshot mechanism. The index takes part in crash recovery, point-in-time recovery and physical streaming replication, including index scans served from a hot standby.

  • Cross-backend cache coherency. Cached copies are validated against the metapage generation and reloaded when stale, so a backend reliably observes other backends’ committed writes through the index.

Query Execution

  • Candidate-mask threading across scan keys. Conjunctive queries evaluate the most selective key first and restrict later keys to the surviving rows, rather than evaluating each key independently.

  • Length-predicate support. Patterns made up only of _ wildcards are recognised as length predicates and answered directly from the length bitmaps.

  • Rewritten cost model. Costs derive from pattern shape, column statistics and relation size, so the planner can weigh Biscuit against pg_trgm and a sequential scan.

Index Definition

  • biscuit_like_ops / biscuit_ilike_ops operator classes, for columns that need only one case mode, avoiding the build and maintenance cost of the structure set they will never be queried with.

Upgrading: this is a breaking on-disk format change. Indexes built under 2.x must be REINDEXed; there is no automatic migration.

Quick Start

-- Install the extension
CREATE EXTENSION biscuit;

-- Create a Biscuit index
CREATE INDEX idx_products_name ON products 
USING biscuit (name);

-- Query using LIKE patterns
SELECT * FROM products 
WHERE name LIKE '%laptop%';

Documentation Structure

When to Use Biscuit

Perfect For:

  • Text search with wildcards (%product%, item_%)

  • Email/domain filtering (%@company.com)

  • Prefix/suffix matching (admin%, %_test)

  • Multi-column pattern queries

  • High-cardinality string columns

  • Frequent LIKE/ILIKE queries in analytics

Not Ideal For:

  • Full-text search (use tsvector instead)

  • Exact equality matches (B-tree is sufficient)

  • Low-selectivity patterns (single %)

  • Tables under continuous write load, or deployments with large connection pools

System Requirements

  • PostgreSQL 16 or higher

  • Linux, macOS, or Windows

  • Recommended: CRoaring library for enhanced performance

License

Biscuit is open-source software released under the PostgreSQL License.

Support

Ready to accelerate your pattern matching queries?

👉 Start with the Installation Guide