cancel
Showing results forย 
Search instead forย 
Did you mean:ย 
Data Governance
Join discussions on data governance practices, compliance, and security within the Databricks Community. Exchange strategies and insights to ensure data integrity and regulatory compliance.
cancel
Showing results forย 
Search instead forย 
Did you mean:ย 

Metastore Location

VRaj
Visitor

Does removing the ADLS path metastore location affect the data or cause any critical issues? the ADLS path tied to a project and if people from other projects wanted to use the metastore for catalog creation. We do not want to incur any payments for their data storage or processing. I believe only one metastore is allowed for a tenant in a region. How could this be solved. 

5 REPLIES 5

nayan_wylde
Esteemed Contributor

Youโ€™re correct: only one Unity Catalog metastore per region per account is allowed.

If the metastore root is tied to Project Aโ€™s ADLS account, and Project B uses the same metastore, then:

Any managed tables created by Project B will store data in Project Aโ€™s ADLS path.

This means Project A incurs storage and possibly compute costs for Project Bโ€™s data.

VRaj
Visitor

Does removing this ADLS path from the metastore cause any issues? 

Our idea is to remove the ADLS path from the metastore and have the metastore in a common resource group or subscription. So, all projects or departments using the databricks service in a tenant can attach to it. Does deleting or removing the existing ADLS path cause any critical concerns like losing the hierarchy in existing catalogs or the accesses for AD groups.

nayan_wylde
Esteemed Contributor

 

Yes there will be few impacts.

  • Managed Tables Impact

    • The catalog hierarchy (catalogs โ†’ schemas โ†’ tables) is stored in the metastore database, not in ADLS.
    • So removing the ADLS path does NOT delete catalogs or permissions.
    • However, managed tables depend on that root path for their physical data. If you remove it:
      • Managed tables will break because their data location becomes invalid.
      • External tables remain intact since they point to their own storage.
  • Access Control (AD Groups)

    • Permissions for catalogs, schemas, and tables are stored in Unity Catalog, not in ADLS.
    • So AD group access policies remain unaffected by removing the root path.
  • Critical Risk

    • If any managed tables exist, you must migrate their data to a new managed location before removing the old path.
    • Otherwise, those tables will become inaccessible.

 

Coffee77
Contributor III

When you configure the metastore you can indeed set up the default ADLS for catalogs. However, it is very important to know that you can create other external locations pointing at different ADLS in which placing new desired catalogs. This way you can segmentate your storage in many different ways, all of them under same metastore.

Your scenario seems to match different ADLS holding project catalog(s) but not sure.

On the other hand, there are times Databricks allows to create exceptions to create several metastores in a region. You must reach out to Databricks support contact and ask for requirements.


Lifelong Learner Cloud & Data Solution Architect | https://www.youtube.com/@CafeConData