Neo4j Certified Professional study materials
Neo4j Certified Professional study resources

Neo4j Certified Professional: Neo4j Fundamentals

Neo4j Certified Professional · Work through your exam, one topic at a time.

All guide sections

Neo4j Fundamentals

The Fundamentals domain checks the mental model behind the product: what a property graph is, how Neo4j stores and traverses it, and how a deployment behaves under concurrency and failure. Expect questions on editions, clustering, databases, and the shipped tooling.

The property graph model

Data is nodes, relationships, and properties. Nodes carry zero or more labels; relationships have exactly one type, exactly one start node, one end node, and a direction. A relationship is stored once and can be traversed in either direction, so direction is meaningful for reading but never duplicated in storage. Nodes and relationships both hold key/value properties. There are no dangling relationships and no join tables: connectivity is stored as pointers, so traversal cost depends on the subgraph touched rather than total graph size - the property usually called index-free adjacency. Indexes only help find entry points.

Deployment and editions

Community is a single server. Enterprise adds multi-database support (including the system database holding metadata, where SHOW DATABASES runs), role-based access control, and clustering. In a causal cluster a Raft consensus group elects one leader per database; only that leader accepts writes, while followers serve reads and can be promoted. Read scaling comes from adding followers; there is no write scaling within a database.

Worked example

A reporting job reads immediately after a write and sees stale data. Reasoning: the read was routed to a follower that had not yet applied the transaction. The fix is causal consistency - capture the bookmark returned by the writing session and pass it to the reading session, so the read waits until at least that transaction is visible. Bookmarks, not sleeps, express read-your-writes in a cluster.

Traps

Assuming every cluster member accepts writes. Confusing bolt:// (direct connection to one server) with neo4j:// (routing, topology-aware) - a cluster client must use the routing scheme. Believing relationships can have multiple types or that labels are mandatory. Expecting a node with relationships to be deletable without DETACH DELETE.