Dedicated compute requirements and limitations

This page outlines requirements and limitations for dedicated compute. Most dedicated compute limitations are runtime-dependent, as feature support has been added over time.

important

Init scripts and libraries have different support across access modes and Databricks Runtime versions. See Where can init scripts be installed? and Compute-scoped libraries.

Dedicated compute assigned to a group has additional limitations. See Group access limitations.

Fine-grained access control support

Fine-grained access control is supported on dedicated compute with certain requirements:

• Your workspace must be enabled for serverless compute.
• Read operations are supported on Databricks Runtime 15.4 LTS and above.
• Write operations are supported on Databricks Runtime 16.3 and above. See Support for write operations.

• If your workspace was deployed with a firewall or has outbound network restrictions, you must open ports 8443 and 8444 to enable fine-grained access control on dedicated compute. See Security groups.

If your dedicated compute is running on Databricks Runtime 15.3 or below:

• You cannot access a table that has a row filter or column mask.
• You cannot access dynamic views.
• To read from any view, you must have SELECT on all tables and views that are referenced by the view.

Streaming and materialized view requirements on dedicated compute

• To query a table that another user created using Lakeflow Declarative Pipelines, including streaming tables and materialized views, your workspace must be enabled for serverless compute and your dedicated compute must be on Databricks Runtime 15.4 or above.
• Asynchronous checkpointing is supported on Databricks Runtime 12.2 LTS and above.
• Using StreamingQueryListener to interact with objects managed by Unity Catalog is supported on Databricks Runtime 15.1 and above.