Amazon Neptune
Managed graph database — relationships as first-class data.
❓ What is it?
A managed graph database supporting property graphs (Gremlin/openCypher) and RDF (SPARQL), with Neptune Analytics adding vector search over graph data for GraphRAG patterns.
💡 Why does it exist?
Some questions are ABOUT connections — fraud rings, knowledge graphs, recommendations by relationship — and joins in relational stores collapse under multi-hop traversals that graphs answer natively.
⏱️ When should you use it?
Use it for knowledge graphs grounding AI answers, fraud-ring detection, and identity graphs; consider GraphRAG when answers require traversing relationships, not just similar text.
🗺️ Where does it fit?
In the data layer beside your other stores: applications and ML pipelines traverse it via graph queries; Neptune ML integrates with SageMaker for graph neural network predictions.
🔌 How do you integrate it?
Create a cluster, load data in bulk from S3, query with Gremlin/openCypher/SPARQL; enable Neptune ML or Analytics when predictions or vector search over the graph are needed.
🧩 Commonly integrated with
🎯 Exam angle (AIF-C01)
- "Highly connected data / relationships" → graph database → Neptune. It is a common wrong-answer trap for generic storage questions.
- Neptune appears in GenAI contexts as a knowledge-graph/GraphRAG grounding source.