What is a Vector Database?
A vector database stores embeddings and makes it fast to find similar items by meaning, not just keywords.
Similarity search
k‑NN and ANN search for top‑k relevant vectors.
Metadata filters
Filter results by product, version, permission, or region.
Hybrid search
Combine BM25 and vector for better short‑query performance.
Re‑indexing & updates
Efficiently update embeddings as content changes.
How to choose
- Scale and latency requirements (QPS, p95 latency)
- Filter support and access control integration
- Hybrid search, re‑ranking, and observability
- Operational model: hosted vs self‑managed, backups, SLAs