Talk with our engineers by joining the new Onehouse community Slack!
Core capabilities
Catalog interoperability
Onehouse registers your tables in the catalogs your other engines already use, so those engines read the data in place with no copy. Data privileges sync the same way, translated between each catalog’s privilege model.
Catalog sync
OneSync, the metadata sync service, registers your tables in external catalogs, including Delta Lake tables in catalogs such as Databricks Unity Catalog. The format translation underneath is Apache XTable (Incubating), the project Onehouse originated. Microsoft OneLake uses the same project: Microsoft and Snowflake announced that data written by either platform is available in both Iceberg and Delta Lake formats in OneLake through XTable translation.
| Catalog | AWS | Google Cloud | Azure / on-prem |
|---|---|---|---|
| Hive Metastore, DataHub, XTable | Yes | Yes | Yes |
| Databricks Unity Catalog | Yes | Yes | Yes |
| Snowflake Horizon Catalog | Yes | Yes | Yes |
| Snowflake Open Catalog (Iceberg REST) | Yes | Yes | Yes |
| OneLake Catalog | — | — | Azure |
| AWS Glue Metastore and Glue IRC | Yes | — | — |
| Dataproc Metastore | — | Yes | — |
| BigQuery and BigLake | — | Yes | — |
A project can nominate default catalogs. They pre-fill when you create a Flow, and apply automatically to tables created through SQL and to newly registered external tables.
OneSync keeps one copy of the data, exposes it in every format through Apache XTable (Incubating), and registers it in every catalog.
Adding a catalog with XTable multi-format sync enabled.
Permission sync
Data privileges sync in both directions between catalogs. You import privileges from a source catalog into a Onehouse data role, edit them there, and push them out to target catalogs. Onehouse translates table and database privileges between each catalog’s privilege model. Row filters and column masks stay in the Onehouse catalog.
| Catalog | Principals | Hierarchy |
|---|---|---|
| AWS Glue via Lake Formation | IAM user, group, role | Database, table |
| Databricks Unity Catalog | User, service principal, group | Catalog, schema, table |
| Snowflake Horizon | User, role | Database, schema, table |
| Microsoft OneLake | Entra user, Entra group, OneLake role | Item, schema, table |
Query engines
Onehouse documents read paths for Snowflake, Databricks, Amazon Athena, Trino, StarRocks, DuckDB, and Redshift Spectrum. Each engine reads the table in place through a catalog, with no copy. Agents reach the same tables through the AI Gateway: Lakegres serves them natively, and another platform such as Snowflake can be added as an MCP server under the same tool controls.