New Features

« Prev

	Important
	Hortonworks strongly recommends that all users running HDP 2.3.4 upgrade to HDP 2.3.4.7.

This section highlights several new features in HDP 2.3.4.

Spark 1.5.2 is the default Spark version with HDP 2.3.4. With a new HDP 2.3.4 cluster install, Spark 1.5.2 is installed. With the upgrade of an existing HDP cluster to 2.3.4 using Ambari, the Spark component is automatically upgraded to 1.5.2. For clusters not managed with Ambari, Spark can be manually upgraded from 1.3.1 or 1.4.1 to Spark 1.5.2; see the Optional Spark Manual Upgrade Procedure for instructions.

If you have upgraded to Spark 1.5.2 and want to downgrade to Spark 1.4.1, follow the Optional Spark Manual Downgrade Procedure. The Spark downgrade is only available as a manual step.

Table 1.1. New Features

Component	Feature
Atlas 0.5.0	UI Enhancements (RMP-4872) HBase Integration (RMP-4834) SolrCloud Integration (RMP-4835)
Falcon 0.6.1	Hive hook Addition of two-part upgrade for HDP 2.1 to 2.3 (user must upgrade Falcon from 2.1 to 2.2 and then 2.2 to 2.3)
HDFS	Log tracing (RMP-4750). Tag jobs or workflow with id that can be used for tracing.
Kafka 0.9.0	Kafka 0.9.0 GA
Knox 0.6.0	Honor X-forwarded headers (RMP-3621) LDAP diagnostic tools (RMP-3645)
Ranger 0.5.0	Store audit in Azure data store (Azure SQL Database) (RMP-4360) LDAP configuration tool (RMP-4649) HA blueprint Ranger Kafka Plugin -- add support for additional access types (BUG-49278) Kafka Ranger Plugin -- implement updated Apache Kafka Authorizer interface (BUG-45593)
Spark 1.5.2	Spark 1.5.2 GA Spark streaming (RMP-4640) (formerly available as a Technical Preview) Spark SQL GA (RMP-4639) (formerly available as a Technical Preview) Spark SQL Thrift JDBC/ODBC Server (formerly available as a Technical Preview) DataFrame API (SPARK-5097) (formerly available as a Technical Preview)
Tez 0.7.0	Standalone UI (BUG-43333)
YARN	ATS (Application Timeline Server) 1.5

​New Features