Hadoop distribution vendor MapR announced the latest release of its distro that includes the latest version of the distributed computing system, Apache Hadoop 2.2, and YARN, a resource management system that was added in 2.2.

YARN enables Hadoop applications to share compute resources of a cluster they are running on, which improves utilization of the cluster, according to MapR. It frees applications up to use underlying distributed file system and database tables outside Hadoop's MapReduce framework.

“The beauty about YARN is that you're not tied to the MapReduce framework,” Jack Norris, chief marketing officer at MapR, said. The new capability frees up a compute cluster to run workloads other than Hadoop.

MapReduce jobs can run side by side with YARN schedulers on the same nodes. YARN-based applications can read write and update data in the cluster's distributed file system and database tables.

The end goal is to give developers the ability to write a broader set of big data applications, not limited to MapReduce.

Another feature added in the release is support for the open Network File System (NFS) protocol in addition to Hadoop's native HDFS (Hadoop Distributed File System). This means an enterprise storage application no longer has to be able to use HDFS to access data on the Hadoop cluster.

Through MapR's NFS API (application programming interface), an application can run directly on the Hadoop cluster, through YARN, without any changes.

The advantage of this is that the cluster has access to more and more recent data, paving the way for analytics applications that use streaming data sets, Norris explained. Applications are no longer limited to batch processing.

In addition to the latest release of the distro, MapR rolled out a free virtual environment (sandbox) for Hadoop developers can use to write and test applications. “It's really about spotlighting the advanced capabilities of the MapR platform,” Norris said.

Finally, MapR announced the release of HP's business intelligence solution Vertica Analytics Platform built on top of its distribution. The SQL-on-Hadoop solution is available as an early access release.