Storage
Durable data without a storage cluster.
Persist Parquet snapshots, commits, manifests, and indexes on local disk. Keep recovery snapshots in optional S3-compatible storage.
GGraphDB 2.3.9 · Local-disk property graph
Build graphs that stay operational.
Continuously update your graph and query its latest state on local disk. Schedule verified S3 snapshots and restore when needed.
Query
GraphQL + JSON DSL
Persist
Local disk + S3 backups
Operate
One process, many tenants
Build
One Go binary
A graph core shaped for production.
Flexible entity modeling meets bounded queries, explicit freshness, and the controls needed to run the database.
Bound every graph walk.
Match, traverse, inspect impact, and find shortest paths with explicit limits.
match
pattern
neighbors
traverse
impact
shortest_path
Make read visibility explicit.
Carry a committed version into the next query when a workflow must observe its own write.
min_version: committed_version
Protect every tenant boundary.
Isolate entity data, indexes, cursors, and write pressure by tenant while preserving idempotent ingestion.
Request header
X-Tenant-ID: catalog
Write identity
Idempotency-Key: sync-1842
Keep the graph inspectable.
Compact, audit, repair, back up, restore, and inspect index health through supported operational paths.
Integrity audit
Backup and restore
Index health
Metrics and traces
Update and query in one process.
Each process owns its local directory. Choose standalone or independent Raft replicas; concurrent reads see committed versions while updates continue.
Clients
Applications and collectors
GraphDB process
Write and query API
GRAPHDB_MODE=all
Durable state
Local persistent disk
Parquet + manifest
Optional recovery
S3 snapshot repository
Backup → restore
Recovery boundary
Standalone or optional Raft
Opt in to scheduled snapshots, retries, retention, and restore drills. For high availability, the same binary also supports independent Raft replicas and tenant-sharded groups. Cross-host and production capacity acceptance remain pending.
Query the same graph at the right level.
Use GraphQL or JSON DSL for new integrations. The text DSL is also available for concise graph exploration.
Read the query guideStart locally. Keep the same data model in production.
Run the Go binary directly, or start the bundled Docker Compose topology.
Terminal
go run ./cmd/graphdb serve
curl -fsS http://127.0.0.1:8080/v1/health