Big Data specialist MapR Technologies Inc. made waves at the Strata Data Conference on Tuesday, announcing a major overhaul of its flagship Converged Data Platform.

The company has introduced a slew of new capabilities for its MapR-DB NoSQL database – a key component of its Converged Data Platform – designed to power a new generation of applications that leverage in-place and continuous machine learning, real-time data integration and support for microservices.

With the update, MapR-DB can now handle a much broader set of use cases than before, the company said. This includes the Internet of Things, single view, personalization, metadata catalogs and artificial intelligence- or machine learning-infused applications.

The new capabilities suggest MapR is trying to position its platform as a kind of superset of other databases, including both relational and NoSQL. The update means MapR’s platform now can read every kind of database while incorporating streaming data for real-time analytics. MapR thinks the database market is fragmenting, and it’s trying to be the glue that puts all the pieces together.

Combining analytics with operations

The crux of the update is that MapR now can process data of all types from multiple sources, bringing together analytics and operations under one roof for the first time. Many database makers are now combining transactional and analytical features in the same engine, an architecture that Forrester Research Inc. calls “translytical,” but MapR said it can do so while reading data from many sources.

The list of new features is dizzying. Data access via native secondary indexes, self-service Structured Query Language data exploration, operational business intelligence, batch and real-time data processing, analytics and integration just some of the highlights.

Jack Norris, senior vice president of data and applications at MapR, told SiliconANGLE that the main issue the company is trying to solve is one of scaling existing databases. He explained that most organizations still have a tendency to lump all of their data into one big “data lake,” making it difficult to process and analyze.

“Organizations are realizing that they’re not winning by collecting data and putting it into a big lake,” Norris said. “The way you win is by adding context, by injecting intelligence as the page is loading. What are the offers and additional products I want to recommend to a shopper?”

MapR-DB’s new capabilities are designed to add this context and intelligence, enabling operations and in-place analytics to be performed at the same time. This allows customers to apply analytics in real-time, so for example, a retail company could quickly crunch its data on a customer to offer various discounts and promotions as the web page loads.

“It’s about integrating real-time flows both on the front end and back end,” Norris said.

Native SQL support

The addition of full support for native secondary indexes is key to enabling this. So to is the support for SQL queries, which previously could only be performed on MapR’s platform using third-party tools such as Apache Drill. With SQL, MapR is combining transactional and analytical functions in a single platform, in a similar fashion as SAP SE’s in-memory database Hana.

“It’s in the same family as Hana, but it’s not in-memory,” Norris said. “If you looked across apps, the ones that are more ERP data-intensive are where you’d find HANA. As you look at greater context, great scale, JSON directly, web, IoT, you’ll see the MapR play there.”

Another key new capability is batch and real-time data processing and analytics with native connectivity to Apache Spark and Apache Hive, which is what enables in-place machine learning. This feature also allows for advanced machine learning constructs to be based on higher-end algorithms rather than just SQL, Norris said.

MapR said the new capabilities for MapR-DB will be introduced with the release of the MapR Converged Data Platform 6.0 sometime in the fourth quarter.

Image: Pixabay