Databricks Runtime 12.x migration guide

This guide helps you migrate your Azure Databricks workloads to the latest version of Databricks Runtime 12.x. To do so, Databricks recommends that you migrate your workloads in the following order:

• If your workloads are currently running on Databricks Runtime 11.3 LTS or above, you can migrate directly to the latest version of Databricks Runtime 12.x, as described later on this page.
• If your workloads are currently running on Databricks Runtime 11.3 LTS or below, do the following:
  1. Migrate to Databricks Runtime 11.3 LTS first. See the Databricks Runtime 11.x migration guide.
  2. Follow the guidance on this page to migrate from Databricks Runtime 11.3 LTS to the latest version of Databricks Runtime 12.x.

Note

LTS means this version is under long-term support. See Databricks Runtime LTS version lifecycle.

Apache Spark migration guidance

Use the following table to find the Apache Spark version for your Databricks Runtime version. Each link opens the Apache Spark migration guide for that release on archive.apache.org (Spark-specific changes between versions).

Databricks Runtime version	Apache Spark version
13.3 LTS	3.4.1
13.2	3.4.0
12.2	3.3.2

Databricks Runtime behavior changes

There are no notable behavior changes in Databricks Runtime 12.x releases at this time.

Databricks Runtime system environment properties, features, and libraries

For information about system environment properties as well as new, changed, and deprecated features and libraries in Databricks Runtime releases from Databricks Runtime 11.3 LTS to the latest version of Databricks Runtime 12.x, see the following:

• Databricks Runtime 13.2 for Machine Learning (EoS)
• Databricks Runtime 13.2 (EoS)
• Databricks Runtime 12.2 LTS for Machine Learning (EoS)
• Databricks Runtime 12.2 LTS (EoS)
• Databricks Runtime 11.3 LTS for Machine Learning (EoS)
• Databricks Runtime 11.3 LTS (EoS)

Post-release maintenance updates are listed in Databricks Runtime maintenance updates.