IRLFirst physical meetup — Bengaluru, Sat May 23, 4PM · RSVP on Luma
HomeToolsMCPHow It WorksStoriesPhilosophyCommunityArchitectureStar on GitHub
All Tools
P
Vector DBFreeOpen Source

PGVECTORSCALE

Faster vector search for PostgreSQL at any scale

PostgreSQL

ABOUT

Standard pgvector indexes become slow and memory-intensive as the vector corpus grows beyond RAM capacity, making production-scale similarity search expensive and impractical. pgvectorscale brings DiskANN (Disk-based Approximate Nearest Neighbor) to PostgreSQL, enabling vector indexes that scale beyond memory limits by using SSD-optimized graph algorithms. This results in dramatically faster index builds, lower storage costs, and reliable query performance at any scale — all within the familiar PostgreSQL ecosystem.

INSTALL
docker pull timescale/timescaledb-ha:pg17 psql -d "postgres://<user>:<pass>@<host>:<port>/<db>" -c "CREATE EXTENSION IF NOT EXISTS vectorscale CASCADE;"

INTEGRATION GUIDE

1. Build production RAG pipelines with millions of embeddings stored and queried directly in PostgreSQL 2. Replace dedicated vector databases with PostgreSQL + pgvectorscale for simpler infrastructure 3. Run real-time semantic search on document corpora too large to fit in memory 4. Deploy cost-effective vector search at scale using commodity SSD storage instead of RAM-heavy solutions 5. Combine vector search with full PostgreSQL querying — joins, filters, aggregations, and transactions

TAGS

postgresqlvector-searchpgvectordiskannextensionrusttimescale
pgvectorscale — AI Tool | Agentic AI For Good