Follow Datanami:
May 2, 2013

MapR Revs HBase with M7; Plots Search Integration

Isaac Lopez

Last year when MapR previewed their M7 edition for Datanami, they said they would drive a path down HBase lane towards making the non-relational, distributed database more enterprise ready. On Wednesday, they announced that their HBase-infused M7 is now being pressed into production, and dispensed details of their integration of search technology with partner, LucidWorks – a project that is currently in beta.

Looking at the M7 release, MapR claims that it has removed the trade-offs that organizations face when deploying a large-scale NoSQL solution. Through the implementation of a collapsed NoSQL layer with no Java dependency, Jack Norris, VP of Marketing at MapR, says that M7 is a unified data platform that eliminates compactions, increases performance, and simplifies administration.

“It’s an architectural play that drives these features so we can make claims like eliminating trade-offs,” explained Norris. With other NoSQL on Hadoop distributions, Norris claims that other distros are overly complicated – “they have HBase running on top of a java virtual machine that stores its data in the distributed file system, which in its own turn is running on a java virtual machine, which stores its data in a Linux file system.”

“It’s a convoluted architecture,” charges Norris, who further notes that there are a lot of administrative tasks involved with how HBase writes data from point A to point Z. “It creates compactions where you are doing batch updates into the file structure which makes HBase unavailable for updates while those compactions are taking place.”

In the MapR implementation, says Norris, they have collapsed that layer (into their proprietary M7 vehicle), which he says delivers an array of features, including snapshots, mirroring, instant recovery, unlimited tables, automatic merges, and other database candy. Addressing how this translates to business value, Norris claims that M7 opens up applications that are currently being done with different point solutions, which he says can now be done in a single cluster.

 “We’ve got companies in the digital advertising space that have Hadoop, and then running alongside Hadoop, they have an MPP database, and running alongside that, they have a NoSQL solution. On top of that you’ve got data being moved and copied and translated across these different systems that’s part of their product and service that they’re providing to their customers. Now they can replace that with a single platform that has the integration of Hadoop with this powerful NoSQL database that’s integrated on the same cluster.”

In conjunction with the M7 release, MapR also announced that they are entering into the beta stage of the integration of search capabilities of partner, LucidWorks. Norris noted that this integration drives again toward the business value of having a single cluster solution where discovery and analytics can take place without having to worry about added synchronization, or data location. “It’s all in the same cluster.”

Related Items:

MapR Traces New Routes for HBase 

New Ropes for Scaling the SQL Wall 

On the Couch with a NoSQL CEO