Support for Hadoop

SAS 9.4M0 (TS1M0) through SAS 9.4M6 (TS1M6), including Base SAS, SAS 9.4 Foundation, and the versions of SAS solutions that run on those maintenance releases, moved to Limited Support from SAS on February 1, 2025.

If you are running SAS 9.4M3 through SAS 9.4M5, contact SAS Technical Support for assistance with your SAS deployment or for any questions about Hadoop support.

SAS 9.4 M8

Details about the minimum supported versions for Hadoop distributions and Kerberos are provided in the following table.
Unless noted otherwise:

• SAS software listed below is for SAS 9.4 (9.4 M8) with the latest hot fixes applied.
• SAS 9.4M8 and later provide Limited support for Cloudera CDH and Hortonworks HDP. Per the Cloudera Support lifecycle policy, these offerings are no longer supported.
• SAS/ACCESS Interface to Hadoop on SAS 9.4M8 supports Cloudera Data Platform (CDP) 7.2 for Public Cloud and CDP 7.1 with CDP Private Cloud Data Services 1.5.0 for Private Cloud. Bulk loading (BULKLOAD=YES) is not supported in Private Cloud.
• The HADOOPPLATFORM=SPARK option requires Spark 2.4. This option is available for use with SAS In-Database Code Accelerator for Hadoop, SAS Scoring Accelerator for Hadoop, DATA Step Processing in Hadoop, and SAS High-Performance Analytics.
• Read SAS Support for Alternative Releases of Hadoop Distributions below to understand SAS support for later versions of Hadoop distributions.
• Information about Hadoop JAR files and SAS environment variables for Hadoop is provided in the SAS 9.4 Hadoop Configuration Guide for Base SAS and SAS/ACCESS. Multiple editions of this guide are available.

Footnotes:

[1] In addition, Apache Hadoop 0.23, 2.4.0, and 2.7.1 and later versions are supported as the Hadoop cluster that is co-located with SAS LASR Analytic Server for access to SASHDAT on HDFS.
[2] Includes a REST API for job submission that results in better performance.
[3] On a public cloud, HDFS operations through the REST API are supported.
[4] Supported only with the HADOOPPLATFORM=SPARK option.
[5] Version 18 or later of the SAS Embedded Process for Hadoop is required.
[6] The latest SAS hot fixes are required.
[7] Version 19 or later of the SAS Embedded Process for Hadoop is required.
[8] Not supported for use in a public cloud environment because of the limitations of org.apache.hadoop.fs.FSDataOutputStream.
[9] Not supported for use in a public cloud environment.
[10] “And later” versions of CDP are not supported.
[11] For in-database processing in Amazon EMR, SAS must be deployed in Amazon Web Services (AWS).
[12] Only SAS/ACCESS Interface to Hadoop supports CDP 7.1 with CDP Private Cloud Data Services 1.5.0.

Hadoop Security

Details about the supported Hadoop Security configurations

Footnotes:

[1] MapR security and Kerberos are not supported
[2] Kerberos not supported if connecting to MapR from Windows host

Hadoop High Availability

SAS currently supports HDFS HA and Hive HA with Cloudera distributions.

SAS 9.4 M7

Details about the minimum supported versions for Hadoop distributions and Kerberos are provided in the following table.
Unless noted otherwise:

• SAS software listed below is for the seventh maintenance release of 9.4 (9.4 M7) with the latest hotfixes applied.
• The HADOOPPLATFORM=SPARK option requires Spark 2.4. This option is available for use with SAS In-Database Code Accelerator for Hadoop, SAS Scoring Accelerator for Hadoop, DATA Step Processing in Hadoop, and SAS High-Performance Analytics.

• Read SAS Support for Alternative Releases of Hadoop Distributions below to understand SAS support for later versions of Hadoop distributions.
• Information about Hadoop JAR files and SAS environment variables for Hadoop is provided in the SAS 9.4 Hadoop Configuration Guide for Base SAS and SAS/ACCESS. Multiple editions of this guide are available.
• Cloudera Data Warehouse - Data Services is supported (for Hive). SAS 9.4M7F or later is required. Bulk loading (BULKLOAD=YES) is not supported.
• SAS/ACCESS Interface to Hadoop on SAS 9.4M7 supports Cloudera Data Platform (CDP) 7.1 for Public Cloud and CDP 7.1 with CDP Private Cloud Data Services 1.5.0 for Private Cloud. Bulk loading (BULKLOAD=YES) is not supported in Private Cloud.

Footnotes:

[1] In addition, Apache Hadoop 0.23, 2.4.0, and 2.7.1 and later versions are supported as the Hadoop cluster that is co-located with SAS LASR Analytic Server for access to SASHDAT on HDFS.
[2] Includes a REST API for job submission that results in better performance.
[3] On a public cloud, HDFS operations through the REST API are supported.
[4] Supported only with the HADOOPPLATFORM=SPARK option.
[5] Version 18 or later of the SAS Embedded Process for Hadoop is required.
[6] The latest SAS hot fixes are required.
[7] Version 19 or later of the SAS Embedded Process for Hadoop is required.
[8] Not supported for use in a public cloud environment because of the limitations of org.apache.hadoop.fs.FSDataOutputStream.
[9] Not supported for use in a public cloud environment
[10] Only SAS/ACCESS Interface to Hadoop supports CDP 7.1 with CDP Private Cloud Data Services 1.5.0.

Hadoop Security

Details about the supported Hadoop Security configurations

Footnotes:

[1] MapR security and Kerberos are not supported
[2] Kerberos not supported if connecting to MapR from Windows host

Hadoop High Availability

SAS currently supports HDFS HA and Hive HA with Cloudera and Hortonworks distributions.

SAS Support for Alternative Releases of Hadoop Distributions

SAS documents the specific set of Hadoop distributions supported with each SAS product release. SAS also documents the minimum release version of each distribution supported with that SAS product release. These represent the specific distribution versions that SAS used to test and validate the associated SAS product release. Each supported Hadoop distribution release is documented using the "dot" version used to test and validate the associated SAS product.

Unless otherwise noted, SAS only supports minor releases within the documented major release.

Please note that if a customer elects to use a version that is later than a documented Hadoop version, different configuration steps than those described in the associated SAS product’s installation documentation are likely to be required. These can include but are not necessarily limited to

• Hadoop client runtime file sets and locations.
• Hadoop configuration files settings and locations.
• Hadoop and SAS environment variable and system option settings.

Some SAS products that integrate with Hadoop may be delivered to customers with a fixed product architecture and deployment configuration design. Such a design cannot be modified to allow use with alternative Hadoop release versions.

SAS expects customers to have the appropriate skills to resolve differences between the supported release and the alternative release being used. By electing to use an alternative release, the customer acknowledges that they have made the appropriate investment into resolving the differences inherent in that alternative release.

All attempts to re-create any problematic customer scenario at SAS will be done using an officially supported Hadoop version release.

If SAS is unable to reproduce the problem, the customer will be required to perform further diagnostics on their own to isolate the problem up to and including reproducing the problem using a supported Hadoop version release.

Support for Apache Hadoop Software Distributed with SAS® Software

Certain SAS software requires a Hadoop distribution as a pre-requisite. Prior to SAS 9.4 TS1M4, as a convenience for customers, Apache Hadoop software was distributed with that SAS software, as documented in License Information for Third-Party Software Distributed with SAS^® Software.

SAS provides support for the integration of Apache Hadoop delivered with SAS software. SAS does not provide support for the installation, or other aspects of the administration and operation, of Apache Hadoop. For production environments, customers should seek out a well-supported third-party distribution of Hadoop. This ensures that they can turn to a dedicated Hadoop vendor for assistance with their production Hadoop needs.