How to install ParadeDB in PostgreSQL

How do I install ParadeDB in PostgreSQL?

install ParadeDB adds the ParadeDB extension so you can run high-performance hybrid search and vector functions directly inside PostgreSQL.

Welcome to Galaxy!
You'll be receiving a confirmation email.

In the meantime, follow us on Twitter

Oops! Something went wrong while submitting the form.

Description

Example H2

Example H3

Why install ParadeDB in PostgreSQL?

ParadeDB turns PostgreSQL into a fast vector and hybrid search engine, letting you store product embeddings next to traditional data. A single database simplifies architecture, cuts latency, and keeps ACID guarantees intact.

What are the prerequisites?

Ensure PostgreSQL ≥ 14, the pgvector extension (bundled with ParadeDB), and superuser access. On macOS, use Homebrew; on Linux, fetch the pre-built .deb or .rpm. Docker users can pull the official image.

How do I install ParadeDB with Homebrew?

Execute:

brew tap paradedb/paradedb brew install paradedbHomebrew places the shared library in $HOMEBREW_PREFIX/lib/postgresql. Restart PostgreSQL so it picks up the new extension.

How do I install ParadeDB on Ubuntu?

Run:

wget https://packages.paradedb.com/paradedb_latest_amd64.deb sudo dpkg -i paradedb_latest_amd64.deb sudo systemctl restart postgresqlThe package copies binaries to /usr/lib/postgresql/<version>/lib/.

How do I install ParadeDB via Docker?

Pull the image:

docker run --name pg-parade -e POSTGRES_PASSWORD=secret -p 5432:5432 paradedb/paradedb:latestThe container boots PostgreSQL with ParadeDB pre-loaded.

How do I create the extension in my database?

Connect as a superuser and run:

CREATE EXTENSION IF NOT EXISTS paradedb;This statement registers the SQL functions, types, and index access methods.

How can ParadeDB vector-search Products?

Store embeddings and index them:

ALTER TABLE products ADD COLUMN embedding vector(384); CREATE INDEX products_embedding_idx ON products USING ivfflat (embedding vector_l2_ops); SELECT id, name FROM products ORDER BY embedding <-> '[0.12,0.98,...]'::vector LIMIT 5;The operator <-> returns the 5 nearest products to the supplied vector.

Best practices for ParadeDB installation

Compile with the same compiler flags as PostgreSQL, keep shared_preload_libraries blank (ParadeDB doesn’t need it), and version-lock packages to avoid unexpected upgrades.

Common mistakes when installing ParadeDB

See below.

Why How to install ParadeDB in PostgreSQL is important

How to install ParadeDB in PostgreSQL Example Usage


-- 1. Install extension (once per database)
CREATE EXTENSION paradedb;

-- 2. Add embeddings column to Products
ALTER TABLE products ADD COLUMN embedding vector(384);

-- 3. Populate with OpenAI embeddings (pseudo-code)
UPDATE products SET embedding = openai_embed(name);

-- 4. Hybrid search: full-text + vector on Orders & Products
SELECT o.id, p.name, o.total_amount
FROM orders o
JOIN orderitems oi ON oi.order_id = o.id
JOIN products p ON p.id = oi.product_id
WHERE to_tsvector('english', p.name) @@ plainto_tsquery('english', 'wireless')
ORDER BY p.embedding <-> '[0.09,0.11,...]'::vector
LIMIT 10;

How to install ParadeDB in PostgreSQL Syntax


CREATE EXTENSION [IF NOT EXISTS] paradedb [VERSION 'x.y.z'] [CASCADE];
-- Example:
CREATE EXTENSION IF NOT EXISTS paradedb VERSION '0.6.0';

Common Mistakes

Skipping the server restart after copying shared libraries. PostgreSQL must be restarted so it can discover the new <code>paradedb.so</code> file. Fix: run <code>sudo systemctl restart postgresql</code> or restart the Docker container.
Creating the extension in the wrong database. <code>CREATE EXTENSION</code> is database-scoped; running it in <code>postgres</code> doesn’t make it available in <code>ecommerce</code>. Fix: connect to each target DB and rerun the command.

Frequently Asked Questions (FAQs)

Is ParadeDB compatible with pgvector?

Yes. ParadeDB bundles a patched version of pgvector and exposes identical operators (<->, vector_l2_ops, etc.). Indexes created with pgvector work unchanged.