HydraDB
A distributed graph database that stores its data in object storage such as S3, separating durable storage from a compute tier that serves queries and builds traversal indexes.
HydraDB is an object-store-native graph database that keeps graph records, write-ahead logs, and traversal indexes durably in S3-compatible storage, while compute runs as two independent roles: data nodes that serve queries and canonical mutations, and indexers that build immutable traversal indexes in the background. Both roles keep only disposable state locally, so they can be replaced or scaled without moving the graph itself.
It combines durable graph storage on SlateDB with snapshot-consistent OpenCypher queries, GraphBLAS traversal, Neo4j-compatible Bolt connectivity, and an HTTPS query API, so applications can connect with existing Neo4j drivers or the database's own typed JSON and streaming HTTP API. Object-store leases and SlateDB writer epochs coordinate which node may write to a given part of the graph at a time.
It is written in Rust and released under the AGPL-3.0 license.