Release Apache Hadoop 2.6.0--heterogeneous storage, long-running service and rolling upgrade support

Publish Apache Hadoop 2.6.0
--heterogeneous storage, long-running service and rolling upgrade support

I am pleased to announce that the Apache Hadoop community has released the Apache 2.6.0:http://markmail.org/message/gv75qf3orlimn6kt!

In particular, we are pleased with the three major films in this release: heterogeneous storage using SSDs and memory tiers in HDFs, support for long running in yarn services and rolling upgrades, upgrade your cluster software, and then restart upgraded nodes without shutting down the cluster or losing the work in progress. Yarn as its architecture center, Hadoop is constantly attracting new engines to run in the data platform as an organization that wants to efficiently store data in a single repository and interact with it in different ways at the same time.

Thank you very much for all the contributors and the people who have worked with this release, there are nearly 900 Jira issues solved in four ways:
? Hadoop generic: 231 Jira Problem Solving
? hdfs:305 a Jira problem solving for Hadoop
? yarn:290 a Jira problem solving for Hadoop
? The MapReduce of Hadoop: 70 Jira Problem Solving

The highlights of Apache Hadoop2.6.0

Here are some details about the most important features. For a complete list of features, improvements, and bug fixes, see Release Notes: Http://hadoop.apache.org/docs/r2.6.0/hadoop-project-dist/hadoop-common/releasenotes.html.

Enhanced HDFs support for heterogeneous storage tiers

Administrators can store data to these different tiers of storage in a qualified datanode across disk storage tiers, as well as APIs that the application can take advantage of. This means that administrators can optimize their applications by using Hadoop to run:
? In SSD storage tier to increase read/write Latency
? memory storage layer for fast read/write applications that either have temporary data or fail (such as Spark, Tez, etc.)
? archive storage tiers to improve storage efficiency.

Support for long-running services in yarn

Apache Hadoop2.6.0 includes enhancements to the core Apache Hadoop yarn platform to make long-lived services (such as Apache Storm,apache Samza,apache Kafka or Apache HBase), Can run in yarn and take full advantage of its benefits of fault tolerance, security and ease of maintenance.

Apache Hadoop originally architected to support batch processing of data. However, some applications are "always online" and ready to process input data. For example, Apache Storm must be prepared to process data streams in real time at any time of the day, any day of the year.

With Hadoop2.6.0, clusters can now take advantage of the same infrastructure arrangements to execute and manage multiple workloads for all deadlines. Long-lived services such as Storm and hbase can coexist peacefully together at a specific point in time (such as Apache Hive or Apache Pig) for ad hoc working applications.

Rolling upgrade works in yarn, leaving a reboot

The new work, maintenance Restart feature allows the application to keep its complete and ongoing state in the face of a node failure or reboot. Yarn can now provide scrolling with minimal quality of service degradation used to run the application's upgrade support. The progress of the application work node that has completed or is in progress remains unchanged during the restart process, without having to restart all tasks from the beginning.

Outlook Apache Hadoop2.7 version

The main driving force for the next version of Apachehadoop is the jdk7+ that we now require to use JDK7 (hadoop-10530:https://issues.apache.org/jira/browse/ HADOOP-10530) Apachehadoop forward, also supports JDK8 as a runtime (hadoop-11090:https://issues.apache.org/jira/browse/hadoop-11090).

Other important activities undertaken in the Apachehadoop community are:
? erasure code support in HDFS-hdfs-7285:https://issues.apache.org/jira/browse/hdfs-7285
? resources that support disk YARN scheduling and isolation-yarn-2139:https://issues.apache.org/jira/browse/yarn-2139
? Container Resource Delegation extended YARN resource Management-yarn-1488:https://issues.apache.org/jira/browse/yarn-1488

As always, you can follow along with the development of Wiki:http://wiki.apache.org/hadoop/roadmap Apache Hadoop by tracking the roadmap.

Thanks

Thank you very much for everyone who contributed to this version, and the entire Apache Hadoop community.

RELATED LINKS
Download Apache Hadoop2.6.0 version: Http://hadoop.apache.org/releases.html#18+November%2C+2014%3A+Release+2.6.0+available.
Read the hadoop-2.6.0:http://hadoop.apache.org/docs/r2.6.0/hadoop-project-dist/hadoop-common/releasenotes.html of the release notes.

This article translated from: http://zh.hortonworks.com/blog/announcing-apache-hadoop-2-6-0/

Release Apache Hadoop 2.6.0--heterogeneous storage, long-running service and rolling upgrade support