What is a vector database?

In practice

For most business workloads a dedicated vector database is not required: the Postgres, SQL Server or search engine already in your stack can hold vectors competently at the scale you actually have. The question to ask is not which vector database is fastest, but whether you need a new piece of infrastructure to operate, secure and back up at all.

Where it fits in a build

It becomes genuinely necessary at large scale, with heavy query volume, or when you need advanced filtering and hybrid search across millions of passages. Below that, adding one is often a cost and a maintenance burden bought in exchange for a benchmark nobody will notice.

Common mistakes

• Adding a specialist database for a corpus of a few thousand documents, which any existing store would handle.
• Storing vectors without the permissions metadata, so the system happily retrieves a document the asker is not allowed to see.
• Forgetting that vectors need backing up and re-indexing like any other data.

Related terms

• Embedding
• Retrieval-augmented generation (RAG)
• Least privilege

Working on something that involves this?

ASTACKRA designs and builds AI systems, automation and custom software for businesses that need technology shaped around their own workflow. If this term turned up in a proposal and you want a straight answer about whether it applies to your situation, ask us — no obligation, and we will tell you if the answer is no.