Jetbrains YoutrackDB
YouTrackDB landed on GitHub in 2024 under the JetBrains banner, and the headline reads: object-oriented graph database, Gremlin/TinkerPop, ACID transactions, O(1) link traversal. Reasonable specs, modest star count, easy to scroll past.
YouTrackDB’s story starts when one of the lead developers of OrientDB forked its source code after SAP acquired OrientDB. The YouTrack team forked OrientDB but explicitly wants to preserve OrientDB’s original goals: a general-use open-source database that excels at supporting object-oriented domain models, combining document and graph features with inheritance and polymorphism at the storage level.
Most graph databases force you to flatten your domain model on the way in. Entities become nodes, type hierarchies become label conventions, polymorphic queries become application-side union operations you write yourself. YouTrackDB pushes back on that: classes, subclassing, and polymorphic traversal are first-class constructs in the engine itself, not workarounds in the client layer.
The query surface is layered deliberately. Gremlin is available for graph traversal, YQL (a SQL-based extension with dot-notation link traversal and a MATCH statement for pattern matching) handles structured querying, and GQL integration is in progress. The isolation model is worth noting too: snapshot isolation by default, with each transaction seeing a stable snapshot of the database as of its start time.
It raises a genuine question. Most pipelines I see in practice convert rich domain models into property graphs and lose the type structure in the process. The mapping is done once, treated as a solved problem, and then quietly becomes a maintenance burden. A database that keeps inheritance semantics native to storage means that burden shifts or disappears. Whether YouTrackDB grows into a serious contender or remains a JetBrains-internal infrastructure choice is uncertain. JetBrains has shaped it around the database needs of YouTrack, its issue and project management product, and it’s used internally in production. That’s a meaningful and strong signal.
By the way, if you want to see a more grand and solid fork of OrientDB, take a look at ArcadeDB. It sits somewhere in my top LPG databases.