Graph Databases

A graph database stores a domain as nodes and relationships. A node represents an entity. A relationship records how two entities connect. Both can carry properties. The model makes the connections first-class data, so you can ask questions by following them.

The central design move is to start with questions, not tables. Write the questions your application must answer. Identify the entities and the verbs that connect them. Then model the minimum nodes, relationship types, and properties needed to answer those questions. Neo4j's modeling guide uses exactly this sequence: domain, use cases, data model, sample data, and testing.

What graph databases are for

Graph databases fit data where the relationships are part of the question. Examples include access relationships, dependency maps, fraud rings, recommendations, network topology, and knowledge graphs. In each case, you care about a path or neighborhood, not only one record.