MapR Technologies Inc. is rolling out a new version of its flagship Converged Data Platform, with a focus on DataOps, a variation on the DevOps rapid application development process that focuses on reducing time and improving the quality of results in data analytics projects.

The 6.0 release folds in automatic platform health and security features with a new control system that administers all data sources and types from multiple database backends from a single point. The new MapR Change Data Capture uses an Apache Kafka-based publish-and-subscribe model to listen for changes in a database or stream and trigger actions based upon them.

“If you look at how people are trying to get real-time access to data, it’s extremely hard for them to have microservices that listen to everything in the organization,” said Anoop Dawar, vice president of product management. “We’ve built an [application program interface] to listen for changes in the database, so that anyone who wants to subscribe can start to listen for changes or get a firehose of all changes from Day 0.”

MapR describes its data model as a “fabric,” in which data from files, streams and tables are ingested once and made accessible as a single source. It calls its new platform a Data Science Refinery for self-service access to all data from within the same cluster. When combined with recently announced database indexing in its core database, the company said the new features deliver automated propagation, scaling and management.

Dawar said the intention is to solve the data scientist’s common problem of “finding the right data, cleaning it, securing it and putting in a form to look at. This takes weeks, so when you’re done it’s not the latest data anymore.”

Unified security

The new release also features extensive security enhancements bundled into what MapR calls “single-click security.” Essentially, it’s applying whatever authentication tactics the customer uses to big data tools such as Apache Spark, Apache Drill, Hadoop and Apache Mahout without requiring each to be configured separately. Wire-level encryption, which means just above the physical layer of a system, is now standard, and the company said default security standards have been significantly tightened.

Role-based policy configuration is available across a multitenant model with containerized access that can be set for a role or a user and managed by policies, Dawar said. “Every access is now authenticated so users can only get the data they’re allowed to access,” he said.

The process of harmonizing security across a host of open source data analytics modules is complex, said Mitesh Shah, senior analyst for industry solutions at MapR. “It’s not a simple process to turn on security for those components, but we’re making it a single-click operation,” he said. “Each system runs as a secure system service.” For example, if an organization is running multiple Apache “Drillbits” on different nodes in a cluster, they can authenticate with each other through MapR’s single security platform, he said.

Also shipping with this release is an update to the MapR Expansion Pack that folds in support for OpenStack Manila to enable OpenStack-based clouds to access data on the MapR-XD cloud data store, a new Apache Myriad 0.2 release with security improvements and the ability to handle Mesos GPU bids, a new MapR Container for Developers, enhanced support for Apache Hive on MapR-DB JavaScript Object Notation or JSON tables and improved support for Apache Spark DataFrames and Datasets.

Image: MapR