Memgraph

Description

Memgraph is an in-memory graph database for realtime and scalable analytics which speaks Cypher over the Bolt protocol. Hop connects to it with a graph database connection of the Memgraph type. See Graph databases for the other graph databases Hop works with.

Run Memgraph in docker

Example command to start a Memgraph server in a container:

docker run -it -p 7687:7687 memgraph/memgraph

Memgraph Lab, the browser of Memgraph, runs in a container of its own: docker run -p 3000:3000 memgraph/lab.

What works

Use a graph database connection of the Memgraph type. With it, these work:

Graph query, including Return graph
Graph output, including the validation against the graph model and its indexes
Cypher output
Cypher builder
Split graph and Generate graph CSVs
Graph script
Graph index and Graph constraint, in Memgraph’s syntax, see Memgraph syntax
Check graph database connections
Execution logging (HOP_GRAPH_LOGGING_CONNECTION), the Get graph logging info transform and the Graph logging perspective
The graph database location for execution information

What doesn’t work

Neo4j import: it runs neo4j-admin. Load the files of Generate graph CSVs with plain column names and LOAD CSV instead.
Named indexes and constraints: Memgraph identifies them by their label and properties. The Graph index and Graph constraint actions ignore the names you give them, and drop them by label and properties.
Node key constraints (a combination of properties which has to exist and be unique, a Neo4j Enterprise feature) and constraints on relationships: Memgraph doesn’t have them. The Graph constraint action only offers unique and existence constraints on nodes.
Edge indexes on more than one property, and existence constraints on more than one property.

Memgraph syntax

The Memgraph connection type makes Hop generate Memgraph’s syntax, like a SQL dialect does for relational databases:

Operation Statement

Operation	Statement
Graph index: node index	`CREATE INDEX ON :Person(id)`, with several properties `CREATE INDEX ON :Person(a, b)`
Graph index: relationship index	`CREATE EDGE INDEX ON :KNOWS(since)`
Graph index: drop	`DROP INDEX ON :Person(id)`, `DROP EDGE INDEX ON :KNOWS(since)`
Graph constraint: unique	`CREATE CONSTRAINT ON (n:Person) ASSERT n.id IS UNIQUE`
Graph constraint: existence	`CREATE CONSTRAINT ON (n:Person) ASSERT EXISTS (n.name)`
Graph constraint: drop	`DROP CONSTRAINT ON (n:Person) ASSERT n.id IS UNIQUE`
Graph output and Cypher output with Create indexes	A unique constraint for a single key property, an index for several

Graph index: node index

CREATE INDEX ON :Person(id), with several properties CREATE INDEX ON :Person(a, b)

Graph index: relationship index

CREATE EDGE INDEX ON :KNOWS(since)

Graph index: drop

DROP INDEX ON :Person(id), DROP EDGE INDEX ON :KNOWS(since)

Graph constraint: unique

CREATE CONSTRAINT ON (n:Person) ASSERT n.id IS UNIQUE

Graph constraint: existence

CREATE CONSTRAINT ON (n:Person) ASSERT EXISTS (n.name)

Graph constraint: drop

DROP CONSTRAINT ON (n:Person) ASSERT n.id IS UNIQUE

Graph output and Cypher output with Create indexes

A unique constraint for a single key property, an index for several

Memgraph doesn’t run some statements in an explicit transaction: information queries like SHOW INDEX INFO, and index and constraint changes. Hop runs them on their own, in auto-commit transactions: in the Graph index and Graph constraint actions, in the Graph script action, and in the Graph query transform when every statement of a batch is one of them.

Known issues

With a Neo4j connection (protocol bolt) instead of a Memgraph connection, Hop uses the Neo4j syntax and transactions: creating indexes and constraints fails, for example with Index manipulation not allowed in multicommand transactions.. Change the type of the connection to Memgraph to solve this: convert the Neo4j connection to a graph database connection and select the Memgraph type.

Check the Memgraph docs for more information on indexing.