What are the Apache Hadoop components and versions available with HDInsight?
Learn about the Apache Hadoop ecosystem components and versions in Microsoft Azure HDInsight, as well as the Enterprise Security Package. Also, learn how to check Hadoop component versions in HDInsight.
Apache Hadoop components available with different HDInsight versions
Azure HDInsight supports multiple Hadoop cluster versions that can be deployed at any time. As of April 4, 2017, the default cluster version used by Azure HDInsight is 3.6.
The component versions associated with HDInsight cluster versions are listed in the following table:
The default version for the HDInsight service might change without notice. If you have a version dependency, specify the HDInsight version when you create your clusters with the .NET SDK with Azure PowerShell and Azure Classic CLI.
|Component||HDInsight 4.0||HDInsight 3.6 (Default)|
|Apache Hadoop and YARN||3.1.1||2.7.3|
|Apache Hive||3.1.0||2.1.0, 1.2.1|
|Apache Tez Hive2||-||0.8.4|
|Apache Spark||2.3.1, 2.4||2.3.0, 2.2.0, 2.1.0|
|Apache Livy||0.5||0.4, 0.4, 0.3|
|Apache Kafka||1.1.1, 2.1||1.1, 1.0 * (See Note below)|
Due to system performance considerations, support for Kafka version 0.10 was expired in March 2019.
Check for current Hadoop component version information
The Hadoop ecosystem component versions associated with HDInsight cluster versions can change with updates to HDInsight. To check the Hadoop components and to verify which versions are being used for a cluster, use the Ambari REST API. The GetComponentInformation command retrieves information about service components. For details, see the Apache Ambari documentation.
See HDInsight release notes for additional release notes on the latest versions of HDInsight.
Supported HDInsight versions
The following tables list the versions of HDInsight. The support expiration and retirement dates are also provided, when they're known.
The following table lists the versions of HDInsight that are available in the Azure portal as well as other deployment methods like PowerShell and .NET SDK.
|HDInsight version||VM OS||Release date||Support expiration date||Retirement date||High availability||Availability in the Azure portal|
|HDInsight 4.0||Ubuntu 16.0.4 LTS||September 24, 2018||Yes||Yes|
|HDInsight 3.6||Ubuntu 16.0.4 LTS||April 4, 2017||December 31, 2020||December 31, 2020||Yes||Yes|
Spark 2.1, 2.2 & Kafka 1.0 support will expire on June 30, 2020.
After support for a version has expired, it might not be available through the Microsoft Azure portal. However, cluster versions continue to be available using the
Version parameter in the Windows PowerShell New-AzHDInsightCluster command and the .NET SDK until the version retirement date.
The following table lists the versions of HDInsight that are not available in the Azure portal.
|HDInsight version||HDP version||VM OS||Release date||Support expiration date||Retirement date||High availability||Availability on the Azure portal|
|HDInsight 3.5||HDP 2.5||Ubuntu 16.0.4 LTS||September 30, 2016||September 5, 2017||June 28, 2018||Yes||No|
|HDInsight 3.4||HDP 2.4||Ubuntu 14.0.4 LTS||March 29, 2016||December 29, 2016||January 9, 2018||Yes||No|
|HDInsight 3.3||HDP 2.3||Windows Server 2012 R2||December 2, 2015||June 27, 2016||July 31, 2018||Yes||No|
|HDInsight 3.3||HDP 2.3||Ubuntu 14.0.4 LTS||December 2, 2015||June 27, 2016||July 31, 2017||Yes||No|
|HDInsight 3.2||HDP 2.2||Ubuntu 12.04 LTS or Windows Server 2012 R2||February 18, 2015||March 1, 2016||April 1, 2017||Yes||No|
|HDInsight 3.1||HDP 2.1||Windows Server 2012 R2||June 24, 2014||May 18, 2015||June 30, 2016||Yes||No|
|HDInsight 3.0||HDP 2.0||Windows Server 2012 R2||February 11, 2014||September 17, 2014||June 30, 2015||Yes||No|
|HDInsight 2.1||HDP 1.3||Windows Server 2012 R2||October 28, 2013||May 12, 2014||May 31, 2015||Yes||No|
|HDInsight 1.6||HDP 1.1||October 28, 2013||April 26, 2014||May 31, 2015||No||No|
Highly available clusters with two head nodes are deployed by default for HDInsight version 2.1 and later. They are not available for HDInsight version 1.6 clusters.
Enterprise Security Package for HDInsight
Enterprise Security is an optional package that you can add on your HDInsight cluster as part of create cluster workflow. The Enterprise Security Package supports:
Integration with Active Directory for authentication.
In the past, you can only create HDInsight clusters with a local admin user and a local SSH user. The local admin user can access all the files, folders, tables, and columns. With the Enterprise Security Package, you can enable role-based access control by integrating HDInsight clusters with your own Active Directory, which include on-premises Active Directory, Azure Active Directory Domain Services, or Active Directory on IaaS virtual machine. Domain administrator on the cluster can grant users to use their own corporate (domain) user-name and password to access the cluster.
For more information, see:
Authorization for data
Integration with Apache Ranger for authorization for Hive, Spark SQL, and Yarn Queues.
You can set access control on files and folders.
For more information, see:
View the audit logs to monitor accesses and the configured policies.
Supported cluster types
Currently, only the following cluster types support the Enterprise Security Package:
- Hadoop (HDInsight 3.6 only)
- Interactive Query
Support for Azure Data Lake Storage
The Enterprise Security Package supports using Azure Data Lake Storage as both the primary storage and the add-on storage.
Pricing and service level agreement
For information on pricing and SLA for the Enterprise Security Package, see HDInsight pricing.
Service level agreement for HDInsight cluster versions
The service level agreement (SLA) is defined in terms of a support window. The support window is the period of time that an HDInsight cluster version is supported by Microsoft Customer Service and Support. If the version has a support expiration date that has passed, the HDInsight cluster is outside the support window. The support expiration date for a specified HDInsight version X (after a newer X+1 version is available) is calculated as the later of:
- Formula 1: Add 180 days to the date when the HDInsight cluster version X was released.
- Formula 2: Add 90 days to the date when the HDInsight cluster version X+1 is made available in Azure portal.
The retirement date is the date after which the cluster version can't be created on HDInsight. Starting July 31, 2017, you can't resize an HDInsight cluster after its retirement date.
Hortonworks release notes associated with HDInsight versions
The section provides links to release notes for the Hortonworks Data Platform distributions and Apache components that are used with HDInsight.
HDInsight cluster version 4.0 uses a Hadoop distribution that is based on Hortonworks Data Platform 3.0
HDInsight cluster version 3.6 uses a Hadoop distribution that is based on Hortonworks Data Platform 2.6.
HDInsight cluster version 3.5 uses a Hadoop distribution that is based on Hortonworks Data Platform 2.5. HDInsight cluster version 3.5 is the default Hadoop cluster that is created in the Azure portal.
HDInsight cluster version 3.4 uses a Hadoop distribution that is based on Hortonworks Data Platform 2.4.
HDInsight cluster version 3.3 uses a Hadoop distribution that is based on Hortonworks Data Platform 2.3.
HDInsight cluster version 3.2 uses a Hadoop distribution that is based on Hortonworks Data Platform 2.2.
HDInsight cluster version 3.1 uses a Hadoop distribution that is based on Hortonworks Data Platform 2.1.7. HDInsight 3.1 clusters created before November, 7, 2014, are based on Hortonworks Data Platform 2.1.1.
HDInsight cluster version 3.0 uses a Hadoop distribution that is based on Hortonworks Data Platform 2.0.
HDInsight cluster version 2.1 uses a Hadoop distribution that is based on Hortonworks Data Platform 1.3.
HDInsight cluster version 1.6 uses a Hadoop distribution that is based on Hortonworks Data Platform 1.1.
Default node configuration and virtual machine sizes for clusters
For more information on which virtual machine SKUs to select for your cluster, see Azure HDInsight cluster configuration details.