IBM Powers Next Phase of UK Big Data Research

By Charles King, Pund-IT, Inc.  June 4, 2015

UK Universities and Science Minister Jo Johnson announced a partnership with IBM to boost big data research in the UK. Following on a £113mn ($173mn) government commitment to expand the Hartree Centre at Daresbury over the coming five years, IBM will provide additional technologies and onsite expertise valued at up to £200mn ($306mn). These contributions will include:

  • Access to IBM’s latest data-centric and cognitive computing technologies, including the first UK-based deployment of IBM’s Watson platform. Via the Watson Developer Cloud, the STFC plans to work with Watson Discovery for Life Sciences, Watson Oncology Advisor and Watson Clinical Trials Matching.
  • At least 24 IBM researchers who will work onsite, side-by-side with Hartree researchers.
  • Joint commercialization of intellectual property (IP) assets produced in partnership with the Scientific and Technology Facilities Council (STFC) which operates the Hartree Centre.
  • The STFC will employ solutions from IBM’s Big Data platform, such as Hadoop, Content Analytics, SPSS and Cognos. Initial projects will focus on the areas, such as multiphysics, computer-aided formulation of compounds, life sciences and machine learning.

According to STFC and IBM, the Hartree Centre will also benefit from innovations captured by the OpenPOWER Foundation (backed by IBM, NVIDIA, Mellanox and over 100 other global Foundation members). The big data research program will leverage OpenPOWER high performance computing (HPC) technologies to enable complex analytics on massive data sets.

In fact, the STFC will be accessing the first OpenPOWER high performance computing server — an advanced data-centric system leveraging technologies from OpenPOWER Foundation members IBM, NVIDIA, and Mellanox. The STFC and IBM will also collaborate with third parties to develop software solutions that address challenges in academia, industry and government.

The pitch

UK’s STFC sees the future of big data in IBM’s cognitive/analytics hardware and software technologies.

Final analysis

Big data, analytics and even cognitive computing have become somewhat synonymous, and for good reason. The first two disciplines are certainly complementary, and you would be hard pressed to find anyone willing to argue that computational cognition can occur without analysis. But the technologies required to power such processes are anything but common, let alone equal.

It’s a fact that many powerful analytics and HPC solutions are based on commodity x86-based systems. In fact, Intel silicon powers numerous commercial and scientific HPC and supercomputing installations globally, including a large majority of systems ranked by . But x86 is nowhere to be seen in Watson, the IBM cognitive computing system that captured the world’s attention and imagination in 2011 by thoroughly thrashing two former grand champions on the Jeopardy! game show.

Like many other IBM HPC and analytics offerings, Watson leverages the company’s homegrown POWER processors and Power Systems solutions. There are good reasons for that, such as POWER’s ability to support 4X more hyper threading performance per core than x86 processors. That’s an especially beneficial feature in processing complex queries like those that are so common in analytics workloads.

But IBM has also made substantial commitments to exploit Watson to its full potential. In 2014 the company established a new Watson-specific business unit and pledged $1bn to fund solutions and develop commercial markets around the platform. That is a substantially larger sum and more serious commitment than any of IBM’s analytics/cognitive computing competitors have placed on the table. Partly as a result of those efforts, the current POWER8-based Watson platform is 75% smaller than the system used on Jeopardy! while delivering 240% better performance.

But there’s another significant link between Power and Watson technologies. In August 2013, IBM open sourced its POWER processor architecture and launched the collaborative OpenPOWER Foundation with founding members, Google, NVIDIA, Mellanox and Tyan. Today, there are over 113 OpenPOWER Foundation members, and commercial solutions are expected to reach the market this year.

The U.S. Department of Energy recently selected an OpenPOWER technology similar to the high performance computing server slated for the Hartree Centre effort for next generation supercomputers at the Oak Ridge and Lawrence Livermore national labs. But the deep integration of OpenPOWER and Watson capabilities, along with access to IBM’s Watson Cloud, means that innovations in cognitive computing and big data are also likely to arise from efforts at the Hartree Centre.

What does all this have to do with IBM and the STFC’s collaboration? In short, call it betting on a winner. Per my earlier comments, there is certainly no shortage of interest in or solutions for big data and analytics in today’s IT markets. But it is also safe to say that no other vendor has as broad and deep a portfolio of solutions for these processes as IBM. That includes offerings that support core big data technologies like Hadoop but it also covers more traditional enterprise-class solutions like the company’s Content Analytics, SPSS and Cognos.

Perhaps most important to the Hartree Centre effort are the strengths of the IBM’s hardware investments. The foundation of those solutions is the current POWER8 architecture and associated Power Systems servers. But the latest POWER8-based Watson commercial solutions are essentially unique in the IT market—no other vendor has anything that is remotely similar to it or as flexibly powerful for enhancing commercial applications.

That last point is worth remembering in relation to the IBM and STFC collaboration. Though the STFC is dedicated to supporting and pursuing leading edge computing efforts, the organization works closely with numerous businesses and government agencies with interests in developing dependable, replicable computing solutions. In other words, the Hartree Centre effort will certainly involve scientific computing but it is anything but a science project for the sake of science.

Both IBM and the STFC have the experience, resources and wherewithal to make this effort work, and both are likely to benefit commercially from the IP derived from the project and from the engagements with businesses, universities and government agencies who participate in Hartree Centre projects. Overall, this partnership appears likely to benefit everyone involved. That certainly includes IBM and the STFC but over time the communities around Hartree Centre, other organizations involved in the project and a wide range of businesses and consumers will also be positively impacted by these efforts.

© 2015 Pund-IT, Inc. All rights reserved.