

The newest version of the MapReduce framework has been released, featuring several improvements for the security branch of Hadoop 0.20.205.0, the current stable version. It’s notable that one of the main enhancements arriving with today’s release is the added stability in large scale deployments, the main type of environment for which Hadoop is used to carry out analytics jobs. Several companies are already making use of MapReduce v1.0.
“Hadoop 1.0 is based on the security branch of Hadoop 0.20, specifically Hadoop 0.20.205.0. The new 1.0 release is said to be stable and reliable at scale, with 50,000 node installations already deployed. HBase, the Hadoop “big data” database modelled after Google’s Bigtable, is now integrated with the 1.0 release and the support also includes “performance enhanced access” to local files for the database.”
In addition to the lengthy list of bug fixes and minor tweaks to the code, extra security has been added in the form of Kerberos authentication. The release also supports webhdfs via a read/write HTTP access layer.
This most recent software update follows several other ecosystem developments in the past couple of months. In December Hadoop startup MapR updated its commercial distribution of the big data engine, adding things such as C/C++ apps and an installer for Windows and Mac. Version 1.2 of MapR’s offering contained several other new functions as well and is based on Hadoop 2.3, which was rolled out a month earlier.
This release marks the first Hadoop update in over a year, a time which the various contributors made good use of available resources. Among of the things MapR and other users can leverage and build on is better scalability thanks to federation, and a written version of MapReduce codenamed YARN that’s currently in early stages of development.
THANK YOU