October 2020
These features and Databricks platform improvements were released in October 2020.
Note
Releases are staged. Your Databricks account may not be updated until up to a week after the initial release date.
New Databricks Power BI connector available in the online Power BI service (Public Preview)
October 28, 2020
The new Databricks Power BI connector, released in Power BI Desktop in September (Public Preview), is now also available in the Power BI service (also known as Power BI online).
For more information, see Connect Power BI to Databricks.
Databricks Runtime 7.4 (Beta)
October 21, 2020
Databricks Runtime 7.4, Databricks Runtime 7.4 ML, and Databricks Runtime 7.4 for Genomics are now available as Beta releases.
For information, see the full release notes at Databricks Runtime 7.4 (EoS) and Databricks Runtime 7.4 for ML (EoS).
Expanded experiment access control (ACLs)
October 20-27, 2020: Version 3.31
The expanded experiment permissions introduced with Databricks platform version 3.29 (Sept 23-29, 2020) are now enabled for all deployments.
For more information, see MLflow experiment ACLs.
High fidelity import and export of Jupyter notebook (ipynb) files
October 20-27, 2020: Version 3.31
Databricks now provides high fidelity import and export of notebooks to the Jupyter notebook (ipynb) file format. When you import Jupyter notebook files that were originally exported from Databricks, results and dashboards are preserved. This was previously possible only with the DBC external format. With this upgrade to Jupyter notebook handling, Databricks notebooks are now compatible with tools like GitHub, nbformat, nbdime, and nbconvert.
SCIM API improvement: both indirect and direct groups returned in user record response
October 20-27, 2020: Version 3.31
User records returned by the SCIM API now comply with the SCIM RFC-7643 standard in the way that the response lists groups to which the user belongs. SCIM RFC-7643 defines two types of group membership: direct and indirect. The API was returning only direct membership. It now returns both direct and indirect membership. It also includes a new type
field to distinguish between the two. See Groups API.
Databricks Runtime 6.5 series support ends
October 14, 2020
Support for Databricks Runtime 6.5, Databricks Runtime 6.5 for Machine Learning, and Databricks Runtime 6.5 for Genomics ended on October 14. See Databricks support lifecycles.
Self-service, low-latency audit log configuration (Public Preview)
October 14, 2020
Audit log delivery is now supported as a self-service configuration for accounts on the Premium plan or above and above. As a Databricks account owner, you can use the Account API to configure Databricks audit logs to be delivered to your preferred S3 storage location. In addition, if you have a multi-workspace Databricks deployment, you can create a single audit log delivery configuration that is shared by all workspaces in your account. The new audit log delivery framework logs events with low latency, typically delivering logs within less than 15 minutes of an auditable event, providing a big improvement over legacy audit logging.
For details, see Configure audit log delivery.
SCIM API improvement: $ref
field response
October 7-13, 2020: Version 3.30
In SCIM responses, the $ref
field was returning an internal hostname in the URI. In accordance with SCIM RFC-7463, SCIM responses now return a relative URI in the $ref
field.
Databricks Runtime 7.3, 7.3 ML, and 7.3 Genomics declared Long Term Support (LTS)
October 8, 2020
Databricks Runtime 7.3, Databricks Runtime 7.3 for Machine Learning, and Databricks Runtime 7.3 for Genomics have been declared Long Term Support (LTS) releases.
Databricks provides two full years of support for LTS releases. These releases are supported until September 24, 2022.
For more information about these Databricks Runtime versions, see the Databricks Runtime 7.3 LTS (EoS), Databricks Runtime 7.3 LTS for Machine Learning (EoS), and Databricks Runtime 12.1 (EoS) release notes.
Render images at higher resolution using matplotlib
October 7-13, 2020: Version 3.30
You can now render matplotlib images in Python notebooks at double the standard resolution (also known as retina resolution), providing users of high-resolution screens with a better visualization experience. See Render images at higher resolution.