Infrastructure · Established · Intermediate

Vector Database

A database optimised for storing embeddings and finding the nearest vectors to a query at scale.

What Vector Database is

Vector stores answer 'what is most similar to this' in milliseconds across millions of items, using approximate nearest-neighbour indexes rather than exact scans.

How it works

Indexes such as HNSW and IVF trade a little recall for large speed gains. Production systems combine vector search with metadata filters and keyword search, and support incremental updates and namespace isolation per tenant.

Why it matters

It is the storage layer of the RAG stack, and the component where multi-tenant data isolation must be enforced carefully.

Common uses

→RAG retrieval backends
→Semantic product and content search
→Image similarity search
→Deduplication at scale

Strengths

✓Fast similarity search
✓Scales to hundreds of millions of vectors
✓Hybrid filtering support

Watch for

✓Another system to operate
✓Recall and speed trade-offs
✓Reindexing cost on model changes

Continue exploring

Embeddings Retrieval-Augmented Generation Semantic Search Knowledge Graph