What is a vector database, and how does Milvus fit that category?

A vector database is a system purpose-built to store high-dimensional numerical vectors and efficiently find the vectors most similar to a given query vector, using approximate nearest neighbor (ANN) search rather than exact matching. This is a fundamentally different retrieval pattern than a traditional database's exact-match or range-based queries.

Milvus fits this category directly: its core operation is similarity search over embedding vectors, typically produced by a machine learning model (an embedding model turning a sentence, image, or product into a fixed-length vector), where "similar" means geometrically close in that vector space according to a chosen distance metric.

What distinguishes Milvus from a minimal ANN library is that it wraps that core similarity search capability with database-like features a production system actually needs: durable storage, horizontal scaling, replication, access control, and the ability to combine vector similarity search with traditional scalar filtering (like "similar products, but only in stock and under $50") in a single query.