GQL for connected data
Match relationships, explore paths, aggregate results and update a property graph with one language.
Graph queries. Columnar execution.
An embedded graph database in Rust. Express relationships in GQL, execute with Apache DataFusion, and work with Arrow and Parquet.
From a pattern to a result
Section titled “From a pattern to a result”MATCH (person:Person {name: 'Alice'})-[:Knows]->(friend)-[:Knows]->(next)RETURN next.name AS friend_of_friend;GraphFusion turns graph patterns into DataFusion plans. Scans, joins, recursive paths and aggregates run in the same query engine; results come back as Arrow batches.
Embedded in your application
Use the Rust API or the CLI. Start in memory, then open a local database directory when you need persistence.
Arrow in memory, Parquet on disk
Import columnar graph tables and consume Arrow results. Catalog and graph changes commit together.
An explicit support boundary
The runtime implements a growing subset of GQL. Every reference page distinguishes executable features from syntax-only support.
Learn by doing
Section titled “Learn by doing”Create your first graphInsert Alice → Bob → Cara and query friends of friends.
Find shortest pathsUnderstand selectors, ties, repeated groups and search limits.
Embed in RustRun queries and consume Arrow record batches.
Make writes durableUse transactions, Parquet storage and checkpoints.
GraphFusion is under active development. It does not yet claim complete ISO GQL conformance. Start with the support matrix and operational limits when evaluating it for a workload.