Release Notes
Also available as:
PDF

New Features

This section highlights new features in HDP 2.6.0.

Apache ComponentsFeature
Atlas

Enterprise Readiness

Cloud

  • Shared Services HDC: Atlas (AMBARI-19007)

  • Install & configure Atlas hook components independently of Atlas admin components

Core Capabilities

  • Manually create entities (ATLAS-1193)

  • Higher level API - Simplify and Streamline usecases

  • Integration of ATLAS with Hive2 (ATLAS-1280)

Integration

HDFS

Enterprise Readiness

  • Detect slow disks and slow Datanodes.

Hive

HDP 2.6.0 includes two versions of Apache Hive, one based on the Hive 1 branch, meant for batch processing and one based on the Hive 2 branch, meant for interactive processing. Hortonworks applies stabilizations, bug fixes and features from the trunk code base as deemed appropriate, so neither of these is identical to any specific released build in Apache. Specific details on the differences are provided here.

HDP deployments driven by Apache Ambari can deploy both Hive 1 and Hive 2 side-by-side. In this setup, Hive 2 features are accessed through the Hive LLAP interface.

  • Hive LLAP (Hive 2 only)

  • ACID Merge (Hive 1 and 2)

Hue

Ease of Use

  • GA criteria for Hue migration tool

Knox

Security Features

  • Knox support for Hadoop User-group mapping (KNOX-237)

  • Support for Linux user authentication via PAM (KNOX-537)

Integration

  • Add support for WebSockets to Knox

Phoenix

Reliability

  • Phoenix index stability improvements

Ranger

Security Features

  • Tag Based security expand for existing type common to Atlas + Ranger : Kafka, HDFS, Hbase (RANGER-1233)

  • Ranger Authorization for "show Databases/Tables/Columns/Views" or "describe database/schema/table" permissions on HiveServer2 (RANGER-1195, RANGER-1204)

  • Support for $username variable in Ranger policies (RANGER-698)

Integration

  • SmartSense Rules: Product Analytics for Ranger (RANGER-1304)

Enterprise Readiness

  • Support Governance and security features for Beacon - Ranger

  • Policy versions in use for each Ranger plugin (RANGER-1230)

  • UI hints for Audit search for Ranger Audit Logs & Policy Search on Report Manager or Policy Listing Page (RANGER-1227)

  • Change Ranger Admin start script to setup db for ranger (AMBARI-18065)

  • Export/Import of policies in Ranger (RANGER-1214)

  • Ranger UserSync improvement - Incremental, group filtered usersync with additional attributes (RANGER-1211)

Spark

Ease of Use

  • [Codev 2016-17] Support virtualenv in PySpark (SPARK-13587)

Integration

  • Umbrella - Provide Spark 2 GA on HDP: Features across Spark, Livy & Zeppelin components

  • Spark HBase connector enhancements

Enterprise readiness, security

  • Spark Thrift Server doAs support for identity propagation, for Spark SQL queries

  • Livy API (version 0.3) for REST-based access to Spark and Spark2 jobs, including support for Kerberos authentication and high availability

Core capabilities

  • Support for user-defined functions in SparkR

  • Support in SparkR for additional large-scale machine learning algorithms

Storm

Integration

  • Connectivity Enhancements: Kinesis spout, openTSDB bolt, new Kafka Consumer spout

YARN

Enterprise Readiness

  • Queue Priorities to allow High Priority Services (e.g. LLAP) run in a shared cluster (YARN-5864)

  • Fault tolerance of Slider apps

  • Support port range for YARN apps

Zeppelin

Core capabilities

  • Zeppelin to work with Spark version 2.x

Ease of Use

  • Improve JDBC interpreter

  • Improve Zeppelin Livy integration

  • Support multiple SQL statements in one notebook paragraph

Enterprise readiness, security

Integration

  • Umbrella - Provide Spark 2 GA on HDP: Features across Spark, Livy & Zeppelin components

  • Support for Phoenix interpreter