Databricks and Google Cloud let two catalogs read one copy of data
Databricks says that two catalogs can read the same data copy at this time. The tags move with the data.
Databricks, a data platform company, and Google Cloud announced access in two directions between their catalogs. A catalog is the list of data tables and the rules for them. Customers can use one copy of data from Unity Catalog, the Databricks catalog, and from the Google Lakehouse. Databricks gave no test result or customer number.
How it works
Databricks can attach Iceberg tables that the Google catalog manages. Teams then query them in Databricks, or use Genie, a tool that answers data questions in words that people use. In the other direction, BigQuery can read Unity Catalog tables with no copy.
Shared labels
The two companies helped add labels to the Iceberg project. A label is a tag, for example “domain=finance” or “pii=ssn”. Databricks says that the tags stay on a table when it goes to a different catalog. Tools can use the tags to hide secret data and to help AI agents find data.
What is not known
The source gives no speed or precision numbers. It does not tell how many customers use this tool. It does not tell if all table types work.
Insight: Data access is open, but business meaning is a rival fight
On 11 November, Databricks released Genie Ontology, a layer of business definitions that agents make. Google says its Knowledge Catalog gives agents business context. This reads as a race to hold the business meaning, not the data. Tags move, but each vendor definition possibly does not. Select where your definitions will live.
Sources
Posted