Real-Time Analytics: 10 Ways To Get It Right

It seems that every company working with Hadoop is chasing after the Holy Grail of real-time analytics. MapR, no less competitive, is adding support for JavaScript Object Notation (JSON) to its MapR-DB.

A free developer preview is now available from the MapR website. An enterprise edition is expected next quarter, with pricing to be determined.

"Increasingly, the company is looking at how to incorporate analytics in operations," said Jack Norris, MapR's chief marketing officer.

The challenge comes down to reducing process overhead, measured in time. Traditional database practices were relatively static. Data had to be collected, then moved from one server/application to another server/application, while whoever was doing that made sure the data was in its correct state, Norris explained.

Hours or days could pass before the analysis of a data set yielded an answer or an insight, but by that time the data was old and the world had changed.

[Learn why the data warehouse is history. See Big Data Moves Toward Real-Time Analysis.]

With MapR-DB, the enterprise user can update data in real-time. Retrieving that data is where JSON support comes in. JSON provides a human-readable query that only takes one line of code to "get an answer" from the data.

In contrast, "a typical schema with records and attributes results in a complex selection statement," Norris said. The user may have to input 22 lines of code to retrieve the same information that a JSON query gets in one.

JSON gains its flexibility through MapR's support of the Open JSON Application Interface (OJAI), which forms an access layer across databases, file systems, and message streams.

"[JSON] enables development of applications and helps simplify complex environments." Norris said. "In JSON data being produced can be leveraged immediately without interrupting a step."

Data can be worked on in real-time as it is coming off the Web. Users will be working with the data in its native format. "You are not depending on IT to set up the schema," Norris said. "This adds to the capability of how data is stored."

The community version of MapR-DB with JSON will enable users to test and develop their own apps on the platform, Norris said. "When it becomes part of the business backbone, the company will be glad to pay for the enterprise features," he said. That version will have the governance and security features needed for corporate IT use.

The addition of JSON support to MapR's product line is yet another step of adding utility to Hadoop's capability. Last month, MapR announced it was integrating its Hadoop Distribution 5.0 service with Amazon Web Services.

In February, MapR announced that its latest release of MapR Distribution would enable active-active clusters across the database, with no table replication or bidirectional updates.

One recent study showed typical corporate users applying Hadoop for ETL (extraction, transformation, and load). But recent trends show the major players in Hadoop/big data adding utilities to better enable real-time analytics, in effect placing their products and services one step ahead of the market.