Cray Rolls Out Urika-XA Hadoop Analytics Appliance

By Jeffrey Burt  |  Posted 2014-10-16 Print this article Print
big data

The system, which includes Intel chips and a storage combination that includes SSDs and HDDs, comes two years after Cray's first such appliance.

Supercomputer maker Cray is unveiling the second of its appliances designed to address big data in the high-performance computing space.

Cray officials two years ago rolled out their YarcData Urika system—now called Urika-GD—which offers real-time data discovery using graph analytics. In December, the company will begin selling the Urika-XA appliance, a system that comes integrated with Apache Hadoop and Apache Spark frameworks to address big data analytics applications leveraging those open-source technologies.

Urika-XA, which offers an open and flexible platform that can be used with future analytics workloads, is a natural next step in the company's efforts around analytics, according to Ramesh Menon, vice president of analytics products at Cray. The company saw that customers were using Urika-GD for a wide range of workloads, from fraud detection and cyber-security to helping Major League Baseball teams optimize pitcher-batter matchups.

"What we also noticed was that customers who were using Urika-GD also are using Hadoop," Menon told eWEEK.

The Urika-XA, introduced Oct. 16 at the 2014 Strata + Hadoop World conference in New York, comes with up to 48 compute nodes in a single 42U (73.5-inch) rack. It features Xeon processors from Intel—more than 1,500 cores—and an InfiniBand interconnect. It also includes up to 200 terabytes of solid-state disk (SSD), hard drive and Cray Sonexion 900 storage, which is compatible with the Hadoop Distributed File System (HDFS). The software, along with Hadoop and Spark, includes Cray's Adaptive Runtime for Hadoop, the Urika-XA management system and Cloudera's enterprise offering.

Analytics is becoming increasingly important as organizations look for ways to not only collect and store the massive amount of data being created, but also organize and analyze it to derive usable information that quickly lead to good business decisions. The challenge is creating the technology that is flexible and cost-effective enough to enable enterprises to fully embrace Hadoop, Menon said.

Appliances like the Urika-XA give businesses a single packaged solution with the necessary technologies that can reduce the footprint needed for analytics and drive down the overall cost of ownership, according to Cray officials. A broad range of tech vendors, such as Oracle and IBM, offer similar appliances to deal with workloads such as Hadoop. However, Cray's platform is more open and is designed to not only run Hadoop applications, but also to handle other analytics technologies when they become available, he said.

In addition, Cray's Urika-XA can address the various types of analytics workloads that are found in the high-performance computing (HPC) and enterprise spaces, from stream processing and data mining to ETL and interactive queries.

"Enterprises like the pre-integrated, pre-tuned approach," Menon said.

The Urika-XA addresses the weaknesses that can be found in other approaches to data analytics, including the cloud (fast time to value, but expensive and a lack of control over data), other appliances (pre-integrated hardware and software, but with the risk of getting locked into a vendor's software stack), and customer-developed solutions (good flexibility, but difficult to set up and Hadoop isn't always well-integrated), he said. And that Hadoop capability is important.

"Hadoop is getting to the point where it's not just an experiment at these enterprises," Menon said, adding that organizations are moving on to production implementation.

"The convergence of big data and high performance computing is creating a demand for an open analytics system built on a supercomputing architecture—a solution that allows customers to realize the benefits of advanced analytics techniques, better deal with data complexities, lower TCO and realize faster time-to-value results," Cray President and CEO Peter Ungaro said in a statement.

Menon said the system has been beta tested over the past several months by both commercial and governmental customers, including the Department of Energy's Oak Ridge National Laboratory, which will use Urika-XA for data analytics in climate science, materials science and health care.



Submit a Comment

Loading Comments...
Manage your Newsletters: Login   Register My Newsletters

Rocket Fuel