Skip to content
anythingai.space
Retrieval & RAGbeginner12 min · 10 steps

The Vector Store Landscape

From an in-process library to a fully managed service: what FAISS, Chroma, Postgres with pgvector, Milvus, Weaviate and Pinecone each take off your hands, and when to pick which.

YOU OWN LESS OPERATIONAL WORK →LibraryFAISSEmbeddedChromaExtensionPostgres + pgvectorDedicatedMilvus · WeaviateManagedPineconelibrary → embedded → extension → dedicated → managedEVERY OPTION MUST COVERvectors + textmetadata filtershybrid searchupdates · deletesscale
requestdataheavy work
1 / 10
Key ideaStep 1 of 10

A RAG database does more than find neighbours

Every RAG database stores each chunk's vector, text and metadata, and answers “nearest k to this query, where tenant = A and year ≥ 2024” in milliseconds. Production adds hybrid keyword search, updates and deletes, replication and growth to millions of chunks.

The options differ in how much of that they do for you. They form a spectrum from a library you embed to a service you call.

InsightThe question is rarely “which is fastest?” — it's “which operational work do I want to own?”

Lesson 23 of 37 in Retrieval & RAG