SGI, the high performance computer system vedor, plans to offer pre-built clusters running the Apache Hadoop data analysis platform, the company have announced.
SGI Hadoop Clusters will run fully supported versions of Cloudera's Distribution Including Apache Hadoop (CDH), atop of SGI's line of Rackable servers. SGI, which joined Cloudera Connect Partner Program, will offer a single telephone support line that can answer customer questions about either the SGI hardware or the Hadoop software.
The distribution would be best suited for organisations "that want a more enterprise Hadoop experience, instead of having to roll them on their own," said Bill Mannel, vice president of product marketing at SGI.
SGI will market the clusters to its traditional market of government agencies and financial institutions, focusing on those without the in-house talent to build such clusters by hand. Many of the early users of Hadoop, including giant internet companies like Yahoo and US intelligence agencies, have had the expertise to build their own Hadoop deployments in house. Most organisations, however, do not have this capability, but nevertheless would like to explore vast realms of their data using Hadoop, Mannel pointed out.
Such organisations "have smaller budgets and are looking for a more robust software stack and support services," Mannel said.
With its focus on serving the scientific high performance computing market, SGI is uniquely well suited for delivering large clusters to cost-sensitive customers, asserted Ed Albanese, Cloudera head of business development.
"SGI has had a long history on focusing on the price-performance ratio," Albanese said. "When you look at the larger clusters, the price-performance ratio will really pay dividends. So we expect that SGI will have a very strong offering in that market."
In a Terasort benchmark SGI executed this month, a 20-node SGI Hadoop cluster was able to sort through 100GB of data in 130 seconds. The cluster consisted of SGI Rackable C2005-TY6 half-depth servers, each running Intel Xeon E5630 processors, and sporting 48GB of memory, and four 1 terabyte SATA hard drives.
SGI will reveal pricing, reference configurations and other additional details of the offering when launches the line of clusters at the Hadoop World conference, to be held in New York on November 8 and 9.