October 23, 2012 – MapR has retooled its big data platform with automated architecture and data recovery meant to bring enterprise reassurance to Hadoop and NoSQL work.
The new offering, MapR M7, doubles information throughput speeds, cuts manual splits for tables and compaction, and features an integrated recovery and snapshot function, according to the product announcement. It builds upon Hadoop distribution on the open source non-relational database HBase for a more integrated platform. HBase scalability enables creation of more than 1 trillion tables and vastly increases the number of column families and row and cell sizes. M7 is also compatible with Apache HBase.
From a storage standpoint, John Webster, senior partner at analyst firm Evaluator Group, says there are plenty of capabilities in M7 that resonate with IT administrators managing production data centers, such as “automated load balancing across Hadoop cluster nodes, the ability to use data snapshots to recover applications quickly, and automated failover to recover from a number of different failure modes without restarting the cluster from the beginning of a process.”
The bigger problem with any big data release may be from the business side of the house, where it may be hard to find “enterprise-level acceptance of Hadoop beyond that of an interesting pilot project,” says Webster.
Even as big data capabilities have drummed up nonstop chatter on possibilities from analysts and enterprises alike, actual business adoption of capabilities for Hadoop and other frameworks remains on the low-level of the spectrum, according to a recent survey of IT and business information managers. Jack Norris, VP of marketing for MapR, acknowledges that big data tools are playing catch-up with persistent business-side concerns over outages and backup.
“Part of it is that you have this separate architecture for HBase that sits on top of Hadoop, and if you’ve got a data outage, the reassignment process … takes time. It can be as long as two hours. That really impacts the SLAs and the types of applications that are really successful on HBase,” says Norris, who estimates that 45 percent of Hadoop users are running HBase. The product release was made Tuesday during the first day of Strata Conference and Hadoop World in New York City.