Amazon DocumentDB

❓ What is it?

A managed, MongoDB-compatible JSON document database with separated compute/storage scaling — and native vector search, so embeddings can live beside the documents they describe.

💡 Why does it exist?

Teams already storing JSON app data in DocumentDB can add semantic/RAG features without introducing a separate vector store: one database, one operational model, embeddings next to source documents.

⏱️ When should you use it?

Choose it when the application is document-centric (profiles, catalogues, content) and you want vector similarity search in the SAME store rather than syncing to OpenSearch.

🗺️ Where does it fit?

In the application data layer: services read/write JSON via MongoDB drivers; an embedding pipeline adds vector fields; similarity queries power search or retrieval for a Bedrock app.

🔌 How do you integrate it?

Create a cluster, connect with standard MongoDB tooling, store embedding arrays in documents, build a vector index (HNSW/IVFFlat), and run $vectorSearch-style queries.

🧩 Commonly integrated with

🎯 Exam angle (AIF-C01)

• It is on the exam's list of AWS vector-capable stores: OpenSearch, Aurora/RDS pgvector, Neptune, DocumentDB, MemoryDB.
• "MongoDB-compatible" is the unmistakable identifier in questions.

📚 Study it in a learning path

More in Data & Analytics