The following are changes in Oracle Big Data Appliance release 4 (4.3):
Software Updates
CDH (Cloudera's Distribution including Apache Hadoop) 5.4.7
CDM (Cloudera Manager) 5.4.7
Oracle Big Data Connectors 4.3
Oracle Big Data Discovery 1.1.1
Oracle Big Data SQL 2.0
Oracle NoSQL Database 1.3.4.7 (Community and Enterprise Edition)
Oracle Table Access for Hadoop and Spark
Perfect Balance 2.5
JDK 8u60
See Oracle Big Data Appliance Software User's Guide.
New Features
Automatic Installation for Oracle Big Data Discovery
Customers can download Big Data Discovery 1.1.1 and then use the bdacli command line utility to install the software on a designated node of the primary CDH cluster.
Oracle Table Access for Hadoop and Spark
Oracle Table Access for Hadoop and Spark is an Oracle Big Data Appliance feature that converts Oracle Database tables into Hadoop or Spark data sources. This feature enables fast and secure access to data in the Oracle Database.
HDFS Transparent Encryption
Oracle Big Data Appliance 4.3 provides the option to use HDFS Transparent Encryption. This replaces the eCryptfs on-disk encryption software provided with previous releases. Customers can enable HDFS Transparent Encryption for both new and pre-existing CDH clusters. When enabled, HDFS Transparent Encryption secures Hadoop operations running on the cluster (including HDFS, MapReduce on YARN, Spark on YARN, Hive, and Hbase tasks).
The Oracle Big Data Appliance Configuration Generation Utility provides an option to include HDFS Transparent Encryption when a new cluster is created.
HDFS Transparent Encryption can be enabled or disabled on a cluster via the bdacli command line interface.
HTTPS / Network Encryption
Provides HTTPS for Cloudera Manager, Hue, Oozie, and Hadoop Web UIs.
Enables network encryption for other internal Hadoop data transfers, such as those made through YARN shuffle and RPC.
Like HDFS Transparent Encryption, HTTPS/ Network Encryption is an option in the Oracle Big Data Appliance Configuration Generation Utility, and can also be enabled via bdacli.
Zero Downtime for Upgrades, One-Off Patches, and Cluster Extensions
In Release 4.3, Oracle Big Data Appliance leverages Cloudera’s Rolling Upgrades functionality to keep clusters operational during Mammoth upgrades, patches, and cluster extensions. This is an installation option that allows certain services on a cluster to remain continuously available while each node in the cluster is upgraded and rebooted. Zero Downtime is an option for the following tasks:
Upgrades of the Mammoth software (including Cloudera's Distribution Including Apache Hadoop, Cloudera Manager, and the Mammoth software itself).
One-off patches of Mammoth-installed software.
Cluster extensions. (For cluster extensions within a single rack, rolling upgrades are not optional. These extensions are always done as rolling upgrades.)
Deprecated Features
The following features are deprecated in this release, and may be desupported in a future release:
MapReduce 1 (MRv1)
YARN (MRv2) supersedes MRv1. Users who want to continue to use MRv1 on Oracle Big Data Appliance versions 3.x and 4.x should contact Oracle Support before using Mammoth to patch or upgrade the software.
Desupported Features
The following features are no longer supported as of this release:
eCryptfs On-Disk Encryption
This has been replaced by HDFS Transparent Encryption.