Cray Makes Big Data Move With Analytics Appliance

Cray Inc has taken the first step away from its supercomputer heritage after it announced the official launch of its new ‘big data’ graph appliance.

The appliance has been created by Cray’s newly formed YarcData division, which is dedicated to offering big data solutions for enterprise customers.

Big Data Analytics

According to the company, the YarcData uRiKA graph appliance is a purpose-built solution that is designed to tackle big data relationship analytics, by the use of graphs.

Cray said that while many critical big data problems are based on graphs, most current big data solutions are based on partitioned data structures that scale out on clusters. Apparently the problem with the current big data approaches, including graph databases, is the poor performance on graphs, “since graphs are hard to partition across cluster nodes, are non-deterministic, and are highly dynamic.”

Cray believes that its uRiKA (pronouced Eureka) graph appliance addresses the challenge of delivering insightful analytics on graphs, not only in terms of its ability to handle size and complexity of relationships, but also in terms of its response time and speed of processing.

“With the launch of the uRiKA graph appliance, YarcData now enables data scientists to quickly get to what we call the uRiKA moment – the ability to discover unknown, unforeseen and hidden relationships in Big Data,” said Arvind Parthasarathi, General Manager, YarcData.

“By combining high-performance, graph optimised hardware with an industry-standard, open-source software stack, YarcData’s uRiKA graph appliance enables enterprises to have a rapid time to value on their Big Data relationship analytics initiatives,” he added.

Graph Approach

Cray explained that the uRiKA graph appliance has been designed from the outset for a graph-based approach to big data analytics.

“By bringing together data and relationships from multiple sources, and enhancing these relationships through automated inference and deduction, the uRiKA graph appliance builds up a relationship warehouse,” the company explained. “The system supports both real-time visualisation of relationships for interactive discovery and real-time searches for relationships based on partially-specified patterns and templates.”

The issue of graphs being used to help with big data analytics has been backed by analysts.

“Graphs are an important segment of the Big Data market with increasingly important applicability to problems in areas such as social networking, healthcare, finance, life sciences, and telecommunications,” said Tony Baer, Principal Analyst at Ovum Research.

“Given the importance of real-time interactive analytics on graphs, it’s not just about Big Data but also Fast Data,” he said.

Supercomputer Heritage

Cray said the uRiKA graph appliance is available now, and that it has already been successfully deployed at multiple customer sites, including the Institute of Systems Biology (ISB), Mayo Clinic, Noblis, Swiss CSCS and a US government organisation.

Cray is of course mostly famously associated with high-end supercomputers and this is thought to be the first time the company has ventured outside the supercomputer arena.

Last November it confirmed it would build the supercomputer at the University of Illinois National Centre for Supercomputing Applications (NCSA).

That supercomputer, codenamed Titan, is expected to offer a peak performance of 10 to 20 petaflops, which would exceed Japan’s Fujitsu-built K Computer, which last June hit number-one on the Top500 list of the world’s most powerful supercomputer with a peak performance of 8.8 petaflops.