Google Cloud Serverless for Apache Spark runtime versions

This document lists supported and unsupported Serverless for Apache Spark runtime versions and describes their characteristics.

Runtime node location

This section describes how Google Cloud Serverless for Apache Spark runtime versions allocate nodes within a region to execute batch workloads and interactive sessions.

• Runtime versions 3.0+: To help with obtainability and avoid zonal stockouts, Serverless for Apache Spark might allocate nodes to execute a batch workload or interactive session in multiple zones within the workload or session region. There is no additional cost for cross-zonal network traffic between nodes in a region.

  Runtimes versions prior to 3.0, Serverless for Apache Spark allocates nodes to execute a batch workload or interactive session in a single zone within the workload or session region.

Runtime availability and support

• Serverless for Apache Spark LTS (Long-Time-Support) runtime versions are supported for 30 months. Serverless for Apache Spark non-LTS runtime versions are supported for 12 months.

• Serverless for Apache Spark runtime versions continue to be available for two years after their end-of-support date.

• Runtime subminor version pinning is not supported.

Supported Serverless for Apache Spark runtime versions

Version	Last updated	Released on	Supported until	Available until	Notes
Spark runtime 3.0	2025年10月16日	2025年09月08日	Preview release.
Spark runtime 2.3 LTS	2025年10月06日	2025年05月28日	2027年11月26日	2029年11月26日	General Availability release.
Spark runtime 2.2 LTS	2025年09月11日	2024年03月27日	2026年09月30日	2028年09月30日	General Availability release. This is the default version.
Spark runtime 1.2 LTS	2025年09月11日	2024年03月27日	2026年09月30日	2028年09月30日	General Availability release.

How to choose a Serverless for Apache Spark runtime version

The current default Serverless for Apache Spark runtime version is listed in Supported Serverless for Apache Spark runtime versions. You can choose a different major.minor runtime version when you submit a batch workload or create a interactive session or session template.

Unsupported Serverless for Apache Spark runtime versions

The following Serverless for Apache Spark versions are unsupported.

Version	Includes	Released On	Last Updated	Available Until	Notes
Spark runtime 2.1	Apache Spark 3.4.0 Cloud Storage connector 2.2.20 BigQuery connector 0.28.1 Java 17 Conda 23.3 Python 3.11 R 4.2 Scala 2.13	2022年01月19日	2024年5月30日	2026年08月28日	Unsupported as of 2024年05月30日. 2.1.50 was the final released version.
Spark runtime 2.0	Apache Spark 3.3.4 Cloud Storage connector 2.2.20 Java 17 Conda 22.9 Python 3.10 R 4.1 Scala 2.13	2022年10月21日	2024年08月22日	2026年08月28日	Unsupported as of 2024年08月22日. 2.0.84 was the final released version.
Spark runtime 1.1	Apache Spark 3.3.4 Cloud Storage connector 2.2.28 BigQuery connector 0.28.1 Java 11 Conda 22.11 Python 3.10 R 4.2 Scala 2.12	2023年01月27日	2025年07月31日	2027年07月31日	Unsupported as of 2025年07月31日. 1.1.111 was the final released version.
Spark runtime 1.0	Apache Spark 3.2.3 Cloud Storage connector 2.2.11 Conda 4.11 Java 11 Python 3.9 R 4.1 Scala 2.12.17	2022年01月19日	2023年2月03日	2025年05月30日	Unsupported as of 2023年02月03日. 1.0.29 was the final released version.