Deployment compatibility

Cypher uses set-edge relationship identity: a relationship is identified by its source, type, and target. Importing parallel relationships with identical endpoints and type therefore changes multiplicities compared with databases that give each relationship an independent identity. See Cypher syntax and semantics.

SPARQL supports the query forms listed in SPARQL. ShrikeDB-specific SERVICE endpoints run graph algorithms or SpaceKey searches locally.

Reopening the same data directory recovers persisted data. Include the entire directory in a backup, including the durable Cypher schema catalog. Declared label/property indexes are rebuilt from the recovered graph, and uniqueness and property-existence constraints resume guarding node writes before the database acknowledges them.

Single-tenant mode uses an implicit system identity. Multi-tenant packages add authenticated users, roles, and policies. See Security Model.

HA transaction replication includes graph, vector, KV, and geospatial writes. Distributed planner support still varies by query shape; consult HA deployment status and configuration when designing clustered query workloads.

Measure the requests your application sends, including result transfer, on representative data. Record the query, data size, concurrency, cache conditions, returned row count, latency, and errors. Compare correctness before adjusting performance settings. Improve query performance explains the diagnostics available for that work.