Unified context and governance for analytics and AI agents
One catalog over every engine, cloud, and agent you already run. Shared definitions, shared permissions, no migration.
Ask one question in four places. Get one answer.
Dashboards, notebooks, finance models, and AI agents each compute revenue their own way. In 2.0, one governed definition compiles to SQL for every engine, runs as the caller, and cites where the number came from.
Ossie semantic models
A vendor-neutral spec for metrics and joins, in the Apache Incubator. Gravitino compiles it, permissions it per model and per table, and certifies it.
Glossary and domains
Terms bound to live columns and metrics, so "active customer" means one thing. Domains scope what an agent searches before it answers.
The agent's path
Resolve the term, compile the metric, run as the caller with row filters and masks, cite the model version. Every step audited.
Storing data is solved. Governing how every engine, user, and agent uses it is not.
Open by default. Iceberg REST and Gravitino REST, no lock-in.
Start with the project you already have.
Get off Hadoop without a cutover weekend
Federate Hive and Iceberg under one catalog, move tables at your own pace, and keep one permission model and one audit trail the whole way through.
For Hadoop migrationsOne Iceberg REST Catalog across every cloud
A single endpoint for Spark, Trino, Flink, and Python. Query data where it lives: no per-cloud catalog, no copies, no egress.
For hybrid and multi-cloud estatesOne governed namespace for a regulated estate
Oracle, Teradata, filesets, and cloud catalogs behind one access model and one lineage trail. Built for BCBS 239, GDPR, SOX, and data residency.
For banks and regulated industriesGoverned agentic analytics
Agents assume the user's permissions, resolve "revenue" the same way everywhere, and run through one governed path on any engine. Semantic layer and MCP Gateway arrive in 2.0.
For AI platform teamsRoku narrowed roles across every engine without touching a pipeline.
Roku contributed role narrowing to Gravitino 1.3 and is building tag-based permissions for 2.0, on top of a Hive migration that moved nothing.
Governed self-service on Iceberg at Roku
How role narrowing landed in Gravitino 1.3, and what tag-based permissions change for 2.0.
The Hive migration underneath it moved no data and rewrote no pipelines.
With Bharath Murali, Roku · Mark Hoerth, Datastrato
Built on Apache Gravitino. Datastrato is its original creator.
The open core is an Apache Top-Level Project under the Apache 2.0 license. Datastrato Enterprise is what you run in production: supported, with the core unchanged.
What Enterprise adds
SCIM and OIDC identity · supported Helm chart · audit export to Splunk and Datadog · certified connectors · standard and premium support
Compare editions →Ready to Federate Your Lakehouse?
Whether your next workload is a Trino query, a Lance training run, or an AI agent asking for metadata, Datastrato gives you one catalog and one governance plane to handle it.
Subscribe to our newsletter
Stay up to date with all things Datastrato and the Gravitino Ecosystem