Continuing its drive to bring the batch-oriented Hadoop Big Data platform into the real-time world, MapR technologies, Inc. today is announcing a host of new features designed to support on-the-fly decision-making. MapR 5.0 is also crafted to support bigger workloads, responding to what MapR says is a trend toward customers running more applications on individual clusters.

The new release automatically synchronizes storage, database and search indices for real-time transactions and includes improved security auditing, an area that is considered to be a MapR forte. Release 5.0 also adds support for Apache Drill 1.0 and the latest 2.7 release of Hadoop and YARN.

The biggest news in the expansion of real-time features and support for integrated search, initially through a partnership with Elasticsearch BV. Elasticsearch is used primarily for search applications involving internal records such as customer service data and log files. MapR said the new integration will enable synchronized full-text search indexes to be created automatically without writing custom code. Other search engines will be added in the future.

The addition of support for Hadoop 2.7, including YARN 2.7, enables new features like YARN application rolling upgrades to complement the platform-level rolling upgrades already supported by MapR, as well as integrated Docker container support.

For auditing purposes, log files are now available in JSON, a data interchange format that is designed to be lightweight and easy to understand. Support for Drill 1.X also enhances security by providing for secure access to field-level data files to ensure only authorized data can be analyzed by specific analysts and permissions can be set without the need to IT involvement.

MapR is also continuing its push to make Hadoop clusters easier to configure with the addition of auto-provisioning templates that use a wizard-like format to create clusters according to the most common configuration options. They can be used, for example, to provision data lakes with services deployed in a typical Hadoop cluster, or alternatively for schema-free interactive exploration using Apache Drill. MapR said the templates automate layout, and server provisioning while also executing a suite of tests to ensure that the template deployments will perform as expected.

The company hopes that real-time processing will be its key differentiation point, particularly as user interest moves toward rapid decision-making using popular new technologies like Apache Spark. The company announced support  for the new release from a host of security and data analytics partners. “We’re seeing a lot of automation at the edge that makes adjustments or takes action based upon real-time data,” said Chief Marketing Officer Jack Norris. “Increasingly, it’s about incorporating data flows into interactions.”

In contrast to the popular perception of Hadoop as being a batch-only technology, Norris said MapR has manufacturing customers that are tracking sensors in real time on the factory floor.

MapR also aims to position its Hadoop distribution as a real-time data transport layer between multiple data stores, including relational DBMS, network-attached storage, HBase, and data processing engines like Spark. Drill 1.0 adds self-service data exploration, and Spark 1.3 integration provides for rapid application development and execution. The combination of JSON and Drill, in particular, position MapR as a source for end-user data exploration across multiple back ends. “From the beginning we’ve focused on eliminating the batch limitations of Hadoop,” he said.

Version 5.0 of the MapR Distribution will be available in 30 days.

MapR’s Jack Norris joined SiliconANGLE’s John Furrier and Wikibon’s Jeff Kelly on theCUBE at Hadoop Summit 2014 (19:40).

Photo by Pandu Adnyana via Flickr