The Portland Group
CSCS Top Right Frontpage

Since 1986 - Covering the Fastest Computers
in the World and the People Who Run Them

Language Flags

Visit additional Tabor Communication Publications

Enterprise Tech
HPCwire Japan

From TeraGrid to Clouds

The TeraGrid, the National Science Foundation's evolving program of cyberinfrastructure for U.S. science and education, held its third annual conference June 9-13 in Las Vegas. Observing three years of TeraGrid full-production operation, TG08 opened with a presentation from Dan Reed, one of the people most instrumental in TeraGrid's 2001 genesis as NSF's flagship cyberinfrastructure.Click for full size photo
After founding and directing the Renaissance Computing Institute (RENCI) at the University of North Carolina, Reed moved to Microsoft, where he is Scalable and Multicore Computing Strategist. Before RENCI, he was director of NCSA in Illinois. There, building on the PACI Alliance notion of a distributed grid of shared resources, he helped to develop the TeraGrid vision. The idea, Reed reminded his audience of about 350 researchers, educators and TeraGrid staff, was "to begin escaping the tyranny of data captured at single supercomputing sites."

After looking back to TeraGrid's origins, Reed focused on the future. "What can we learn from the TeraGrid experience, technically and politically? Where is the technology going and what are the research implications?" He referred to a recent special issue of Nature that explores the state of science in 2020, noting that science in the 21st century is inextricable from computing.

Quoting from the study, "From sequencing genomes to monitoring the Earth's climate, many recent scientific advances would not have been possible without a parallel increase in computing power -- and with revolutionary technologies such as the quantum computer edging towards reality, what will the relationship between computing and science bring us over the next 15 years?"
As befitted the Las Vegas setting, Reed asked his audience to ponder risk versus reward. "What probability of successful return would you accept to be the first human to set foot on Mars?" Twenty years ago, he noted, grids were research curiosities and a terabyte was many disks of data. "The future depends on vision and context."

The context has radically changed from not that long ago, he noted, in that bulk computing has become almost free relative to software and power. "Nowadays you can buy a lot of computing on your credit card. We still don't have terabit transcontinental networks for research use; moving lots of data is still hard. The big cost is people. The cost of a professional software developer for a year is now more than a teraflop computing cluster."

Today's context, says Reed, is a Five-Fold Way comprising 1) many-core on-chip parallelism, 2) big "really big" datacenters, 3) web services, 4) ubiquitous sensors (producing huge data volumes), and 5) "clouds" as an evolving model of computational service. Today, further increases in computer performance require embracing multicore parallelism; hardware progress has outstripped progress in software to exploit it.

An important goal, Reed emphasized, is context-aware information. Referring to Vannevar Bush's vision of a national research enterprise, which led eventually to the National Science Foundation, Reed called for services, including datacenters, and the concept of cloud computing that has the ability to put the right information in the right heads at the right time.Click for full size photo
Data models, noted Reed, are in rapid flux because of larger and larger data volumes. This is especially pronounced in some fields, such as biomedical research, where large databases are subject to distributed analysis. A big challenge, probably underappreciated, says Reed, is the scale of the data deluge. "We will be running queries on 100,000 servers," said Reed. "And research is moving from being hypothesis driven ("I have an idea, let me verify it.") to exploratory ("What correlations can I glean from everyone's data?"). This kind of exploratory analysis will rely on tools for deep data-mining." Massive, multi-disciplinary data, said Reed, is rising rapidly and at unprecedented scale.

In discussing next-generation applications and cyberinfrastructure investment, Reed noted that the historical model of "punctuated competitions" is not optimal in that it tends to stress a culture of competition among research centers over long-term collaboration. Research and infrastructure, he noted, mix badly since "it takes a long time to identify appropriate practices and software." Sustainability really matters because software and organizations take time.

Grids and clouds, says Reed, will tend to fuse with time. The rapid growth in the size and capability of commercial computing clouds, as exemplified by work underway at Microsoft, is driven by economics. Reliable, centrally hosted infrastructure provides commercially-based services. While grids are more tailored for academic agendas, economic factors will tend to bring these two related service models together in a fusion that is more than the sum of its parts.

Returning to the ratio of risk and reward as he concluded, Reed stressed the need to ask big questions. In Reed's view, there are basically three: 1) biology, understanding of life and nature, 2) the universe, how matter came to be and cosmic structure, and 3) the human condition, where biology and the universe intersect in the sphere of human creativity and social life. Answering the big questions requires boldness and interdisciplinary partnerships. With the three-fold way of science -- theory, simulation and experiment -- now proven, says Reed, "Great things are ahead. We are positioned to do amazing things."

Most Read Features

Most Read Around the Web

Most Read This Just In

Most Read Blogs

Sponsored Whitepapers

Breaking I/O Bottlenecks

10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.

A New Ultra-Dense Hyper-Scale x86 Server Design

10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.

Sponsored Multimedia

Xyratex, presents ClusterStor at the Vendor Showdown at ISC13

Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.

HPCwire Live! Atlanta's Big Data Kick Off Week Meets HPC

Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?


HPC Job Bank

Featured Events

HPCwire Events