Rackspace Launches OnMetal Cloud Big Data Platform

logo-WHIR

This article originally appeared at The WHIR

Rackspace announced the launch of the OnMetal Cloud Big Data Platform at the Strata + Hadoop World conference this week. The new offering provides big data analytics by running Apache Hadoop with Spark on bare metal servers.

Previous Hadoop solutions offered by Rackspace were virtualized, while bare metal is more conducive to consistent performance from big data applications. Rackspace progeny OpenStack’s released its new version “Juno” this week with a big data processing service which also manages clusters using Hadoop and Spark.

“This solution breaks new ground for the world of big data,” Rackspace CTO John Engates said, according to Forbes. “For the first time, Hadoop and Spark can have the best of both worlds: bare metal performance with cloud agility, all backed by Fanatical Support.”

With the hardware and software infrastructure already fully integrated, the offering is meant to be a convenient, turn-key service. According to Rackspace, it can be provisioned in three clicks. The company also says that in preliminary DFSIO and Terasort performance benchmark testing, the new OnMetal Cloud Big Data Platform beat Rackspace’s legacy Cloud Big Data Platform by an average of between 50 and 100 percent.

OnMetal Cloud Big Data Platform is available for a free trial, and Rackspace is offering a $250 credit for new OnMetal Cloud Servers customers as a promotional offer which ends with calendar 2014.

Forbes reports that Rackspace demonstrated the new platform at Strata + Hadoop World with an analysis of sentiments on all tweets with a certain hashtag. The service gave snapshots every five seconds almost in real time, whereas previous virtualized Hadoop installations would theoretically take 2 or 3 minutes, or possibly even 10 to 15 minutes to deliver the analysis, according to Forbes.

While running Apache’s big data platforms on bare metal is part of the new product’s value proposition, Spark is becoming a valued tool in its own right since becoming a top-level Apache project in February. A Wired article on the then-incubating Spark noted that it is “about 100 times faster than the mighty Hadoop — and could very well replace Hadoop as the stuff that fuels the modern web.”

For now Spark compliments Hadoop with operations like classification and stream processing, and as such competes with tools like MapReduce and Storm.

Rackspace expanded its enterprise offerings earlier this month with managed Google Apps for Work, a few weeks after Taylor Rhodes was named CEO and said the company had been repositioned and had “many levers to pull.”

This article originally appeared at: http://www.thewhir.com/web-hosting-news/rackspace-launches-onmetal-cloud-big-data-platform