HDP 2.2 includes the following new features:
Accumulo
Accumulo multi-datacenter replication
Accumulo on YARN via Slider
Falcon
Authorization
Lineage Enhancement
HCat Replication / Retention
Archive to Cloud
Flume
Flume streaming to Hive for secure and unsecure clusters
HBase
HBase HA: Timeline-consistent replicas with realtime replication
HBase block cache compression
HBase on YARN via Slider
HDFS
Heterogenous Storage: Support for SSD Tier
Heterogenous Storage: Support for Archival Tier
Operating secure DataNode without requiring root access
AES support for faster wire encryption
Hive
Support for SQL Transactions with ACID semantics
Support for SQL Temporary Tables
Better optimize queries using Cost Based Optimizer
Add performance improvements for various queries
Security additions such as Grant, Revoke with a choice of using Native Security or Ranger security with both integrated completely in Hive
Knox
Support for HDFS HA
Installation and configuration with Apache Ambari
Service-level authorization with Apache Ranger
YARN REST API access
Oozie
Oozie HA on secure clusters
Phoenix
Subquery support
Robust secondary indexes
Build secondary index during bulk import
Pig
Pig on Tez
Including DataFu for use with Pig
Ranger
Storm Authorization and Auditing
Knox Authorization and Auditing
Deeper integration with HDP stack: Hive Auth API support for grant/revoke commands, grant/revoke commands in HBase, and Windows support
REST API's for policy manager, local audit log storage in HDFS, and support for Oracle DB for policy store and audits
Sqoop
Support for all Hive types via HCatalog import/export
Support for multiple partition keys
Integration with Hadoop Credential Management Framework
Storm
Storm clusters can now be provisioned via Ambari
Storm can now run on YARN based clusters using Apache Slider
Storm now supports kerberos based authentication and pluggable authorization
Pre-built Spouts for JMS, HBase lookups and Bolts for Kafka, Hive.
Real-time visualization of running topologies and its associated metrics
REST APIs for topology stats
Tez
Tez Debug Tooling & UI
YARN
Support long running services: handling of logs, containers not killed when AM dies, secure token renewal, YARN Labels for tagging nodes for specific workloads
Support for CPU Scheduling and CPU Resource Isolation through CGroups
Work-preserving restarts of ResourceManager and NodeManager
Support node labels during scheduling
Global, shared cache for application artifacts
REST API for YARN application submission and termination
Application Timeline Server is supported in a Secure (Kerberized) cluster
