Understanding DETACH DELETE in Cypher

DETACH DELETE in Cypher is an example of why Cypher is one of my favorite ways of interacting with the Neo4j graph database. The declarative graph query language is constantly evolving to ease the requirements of querying Neo4j. This benefit in ease of interaction, however, can often further

remove the query writer from needing to understand the inner workings of the Neo4j graph database being queried.

Whenever I cover node deletion in my Neo4j trainings I always mention DETACH DELETE as a convenience method for what we just covered because just like in mathematics, you don’t always want to be stuck plugging in numbers to a formula. To grow past that and deepen your understanding it becomes necessary to understand why the formula works the way it does. So let’s dive in…

Neo4j Referential Integrity Guarantee

Within the Neo4j graph database, there’s a guarantee that each relationship will always have a starting and ending node. This guarantee of referential integrity is core to the principles of reliability in which Neo4j concentrates its write architecture. To achieve this, ever time a relationship is being created or modified, Neo4j locks the nodes on both sides of the relationship.

Deleting Relationships with DETACH DELETE

This guarantee on write will also need to be considered whenever a node is being deleted. Neo4j carries out a constraint (upon commit) in which relationships must possess a valid start and end node. This is one reason why, before a node is removed, all its relationships must first be deleted. If a node were to be removed without deleting all relationships, there would be dangling relationships left behind.

Such relationships, considered as first class citizens within the graph, would already be violating their referential guarantee, since they would no longer possess a start and end node. Being unable to cross a relationship and get to the node on the other side will lead to

Read More......