Cray Adds Supercomputer for Big Data
Cray is attempting to democratize supercomputing with its latest product, a supercomputer that is both agile and open, and designed for big data analytics applications.
Although aimed at the enterprise market, one of the first customers is an unnamed CDN evaluating the possibility of using the system to offer cyber security as a service. Other customers include service providers using the system to analyze network telemetry.
Cray Inc. (Nasdaq: CRAY) claims its new Urika-GX is the first agile supercomputer. In this context, agility means the user can run multiple analytics workloads (e.g., Hadoop, Apache Spark, Cray Graph Engine, etc.) concurrently on a single platform. The alternative has been running separate systems for each workload and hoping they work together -- a Franken-cluster, as Cray's senior vice president of products Ryan Waite referred to it.
The new supercomputer includes enterprise tools, such as OpenStack for management and Apache Mesos for dynamic configuration. The inclusion of these tools gives customers the option of trying out different innovations. It also helps with scaling to size, so that users can deploy as few as 16 nodes, or 48 nodes (one full rack), or a multi-rack system, Waite said.
As noted above, the system includes the Cray Graph Engine, a tool for analyzing multi-terabyte datasets comprised of billions of objects. It is the first time Cray has installed the Graph Engine in a general purpose machine, Waite told Light Reading.
The proposition is to combine the scale and throughput capabilities of Cray supercomputers with the convenience of an appliance, the flexibility of industry-standard hardware and an open software framework that enables customers to innovate as they run existing and emerging analytics workloads.
Cray Urika-GX systems are currently being used by multiple Cray customers across the life sciences, healthcare and cyber security industries. The Broad Institute of MIT and Harvard, a non-profit research institute aimed at advancing the understanding and treatment of disease, is currently using the Cray Urika-GX system for analyzing high-throughput genome sequencing data.
The Cray Urika-GX system features Intel Xeon Broadwell cores, 22 terabytes of memory, 35 terabytes of local SSD storage capacity and the Aries supercomputing interconnect. Three initial enterprise-accessible configurations featuring 16, 32 or 48 nodes delivered in an industry standard 42U 19-inch rack will be available in the third quarter of 2016, and larger configurations will be available in the second half of 2016.
— Brian Santo, Senior Editor, Components, T&M, Light Reading