Designed for customers with mission-critical analytics challenges, the Cray Urika-XA system reduces the analytics footprint and total cost of ownership with a single-platform consolidating a wide range of analytics workloads. It also provides superior performance for latency-sensitive analytics with an architecture that is optimized for compute-heavy and memory-intensive workloads, with acceleration through SSDs, parallel file systems, fast interconnects and unique software in the Cray Adaptive Runtime for Hadoop. The Cray Urika-XA also eases the system management burden with a single point of support, a standards-based software stack and compliance with enterprise standards.
"The massive explosion of data has created significant challenges for enterprise customers as they face critical decisions on how to extract value from all of that data and determine the ideal tool for the job", stated Peter Ungaro, president and CEO of Cray. "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. With a pre-tested, pre-configured analytics system free from technology lock-in, the Cray Urika-XA is the right product for a growing set of customers who have analytics needs not met by existing solutions."
The Cray Urika-XA system has been in beta testing for the past few months in both commercial and government customer environments. One of the first customers is the Department of Energy's Oak Ridge National Laboratory (ORNL). Researchers and scientists at ORNL will use the system for data-analytic workloads in climate science, materials science and health care.
"Our researchers are enthusiastic about the opportunities that the Cray Urika-XA system presents in tackling some of our more demanding data-intensive challenges such as those in materials science", stated Galen Shipman, data systems architect and director of the compute and data environment for science at ORNL. "Understanding the causal relationship between nano-scale material structure and desired material features is a big-data challenge. We look forward to working with Cray on this and other Big-Data challenges, leveraging the unique capabilities of the Urika-XA platform."
The Cray Urika-XA system features up to 48 compute nodes with integrated SSDs in a single 42U rack, including Intel Xeon processors, an InfiniBand interconnect, and a Cray Sonexion storage system. The software stack for the Urika-XA system includes Apache Hadoop, Apache Spark, Cray Adaptive Runtime for Hadoop and the Urika-XA management system.
"We are past the point in time where big data is only for the few willing and able to experiment, and quickly evolving to a period where organisations are trying to leverage technology and analysis of this Big Data across their many mission critical processes", stated Dan Vesset, programme vice president, business analytics and big data, IDC. "The Cray solution not only combines supercomputing technologies and performance with the Hadoop and Spark ecosystems, it also allows companies to consolidate various analytics pipelines into one platform. It's a compelling product that should draw the interest of companies in various vertical markets with advanced analytics needs."
With the launch of the Cray Urika-XA system, Cray now has two big data analytics solutions in its product portfolio designed to work together or separately, depending on the workload. The YarcData Urika appliance for graph discovery has been renamed the Cray Urika-GD system. Cray now offers customers a Big Data solution for real-time discovery using graph analytics, and an advanced analytics solution supporting Hadoop, Spark and a wide range of analytic applications.
The Cray Urika-XA system will be generally available in December 2014.
The Cray Urika-XA system is a pre-integrated, open platform for high-performance Big Data analytics. The platform enables organizations to gain insight and capture business value rapidly through advanced analytics, using either the pre-configured Hadoop and Spark frameworks or user-installed analytic tools. The Urika-XA platform is engineered for superior performance on the most demanding analytic workloads, possessing a dense compute footprint of over 1,500 processor cores in a single rack, 38 TB of high-speed SSD storage, and InfiniBand interconnect. With its ability to support both low-latency and batch analytics, ease of management, and standards-based design, Urika-XA provides enterprises a multi-use, turnkey analytics environment at a low total cost of ownership.
The Urika-GD system is a purpose-built, big data appliance for real-time data discovery using graph analytics. The appliance helps automate the surfacing of unknown relationships and non-obvious patterns in diverse data sets without the need for pre-modeling, partitioning or knowing all the queries in advance. The Urika-GD appliance includes graph-optimized hardware that provides up to 512 terabytes of global shared memory; massively-multithreaded graph processors supporting 128 threads/processor; highly scalable I/O with data ingest rates of up to 350 terabytes per hour; and an RDF/SPARQL database optimized for the underlying hardware. Data-center ready and standards based, Urika-GD complements an existing data warehouse or Hadoop cluster and is easy to integrate with existing analytical and visualization tools.