# Vector database

> A database built to find the stored items whose embeddings are closest to a query's, quickly, across millions of rows.

Comparing a query against every stored embedding is exact and, past a few tens of thousands of rows, too slow to sit in front of a user. Vector databases trade a little accuracy for a great deal of speed using approximate nearest-neighbour indexes, and they add the operational parts a real system needs: filtering by metadata, updating documents in place, and keeping the index consistent while it is being written to.

The category is younger than it looks and the boundaries are moving. Purpose-built stores compete with extensions to databases teams already run, and for a corpus in the low millions the extension is often the better answer — one system to back up, one to monitor, one set of credentials. Reaching for a separate vector database before the existing one has been tried is a common and expensive habit.

## Related terms

- https://dfieldsolutions.com/en/glossary/embedding.md
- https://dfieldsolutions.com/en/glossary/rag.md

---

Source: https://dfieldsolutions.com/en/glossary/vector-database
DField Solutions — Dunakeszi, Hungary — dezso@dfieldsolutions.com
Booking: see https://dfieldsolutions.com/en/contact