Pgvector

As explained in Just Use Postgres!

pgvector is a PostgreSQL extension that adds vector data types and similarity search capabilities.

pgvector is a PostgreSQL extension that adds vector storage, similarity-search operators, and specialized indexes. It turns PostgreSQL into a vector database for recommendation services, semantic search, clustering, and retrieval-augmented generation. section 8.0 section 8.2

A practical pgvector workflow includes:

• Start PostgreSQL with a pgvector-enabled image, such as pgvector/pgvector:0.8.0-pg17. section 8.2
• Enable the extension with CREATE EXTENSION vector; after connecting to PostgreSQL. section 8.2
• Store embeddings and compare them with cosine distance (<=>), Euclidean distance (<->), or inner product (<#>). section 8.4
• Optimize searches with IVFFlat or HNSW indexes, balancing search performance, recall, and index size. section 8.7

Enable pgvector in Docker and PostgreSQL

Create a persistent volume and start the database: `docker volume create postgres-pgvector-volume` followed by `docker run --name postgres-pgvector -e POSTGRES_USER=postgres -e POSTGRES_PASSWORD=password -p 5432:5432 -v postgres-pgvector-volume:/var/lib/postgresql/data -d pgvector/pgvector:0.8.0-pg17`. Connect with `docker exec -it postgres-pgvector psql -U postgres`, then run `CREATE EXTENSION vector;`. Querying available extensions afterward shows vector version 0.8.0 installed. section 8.2

Figure 8.6 Performing vector similarity search with Postgres

Use the workflow diagram to follow embedding generation, PostgreSQL storage, vector comparison, and recommendation results. section 8.4

How Just Use Postgres! covers it

How deep it goes

Explains setup, distance operators, and index choices, but stops short of detailed index configuration or deployment administration.

Examples

Uses a movie recommendation service, Docker commands, extension SQL, cosine distance, and Star Wars and Jurassic Park descriptions.

What it assumes

Builds on PostgreSQL and embedding models while introducing vector storage, distance operators, and similarity search.

What you can do after

Operate/configure PostgreSQL with pgvector, store embeddings, run similarity searches, and choose HNSW or IVFFlat indexes.

How it connects

Connects PostgreSQL with vector embeddings, embedding models, semantic search, RAG, HNSW, and IVFFlat.

liveBookmanning.com

Read Just Use Postgres! on liveBook Get the book at manning.com

Related concepts

• PostgreSQLuses · PostgreSQL is an open-source object-relational database management system that implements SQL and supports transactional data storage and extensibility.
• database extensionkind of · A database extension is an installable software package that adds data types, functions, operators, indexes, or other capabilities to a database system.
• container imageimplemented by · A container image is a packaged filesystem and configuration artifact from which a container can be created and run.
• ivfflatuses · IVFFlat is an approximate nearest-neighbor vector index that partitions vectors into centroid-based inverted lists and compares vectors within selected lists.
• generative modelinguses · Generative modeling is a machine-learning approach that learns patterns or a data distribution to generate new samples resembling the data.
• HNSWuses · HNSW is a graph-based approximate nearest-neighbor indexing and search method that organizes data points in a hierarchy of navigable proximity graphs.

This explainer was generated by Manning's knowledge graph from the text of Just Use Postgres!. The section links point to the parts of the book it draws on.