December 09, 2005
Being the new editor for Pacific Northwest National Laboratory's magazine, I was a little apprehensive about meeting the force behind the Laboratory's newest research directorate. But within minutes, Associate Laboratory Director Dr. George Michaels put all concerns to rest as he laid out his vision for the Computational and Information Sciences Directorate (CISD), PNNL's fastest growing research directorate.
In a matter of hours, it became clear why George, as he prefers to be called, is considered a visionary in his field. As I dug into his background, I found that his youthful interests in electrical engineering hinted at the early makings of a computer engineer. However, graduate studies in biochemistry and molecular biology took him down another path. The years that followed engaged George in the fundamental sciences, which allowed him to test an emerging research techniquerelying on computers as a primary tool of scientific discovery.
Much of his career has focused on computational analysis and applying statistical models for fundamental science research. Coincidentally or not, that is PNNL's approach. Every day, researchers push the boundaries on scientific understanding with the strength of high-performance and data-intensive computers behind them. As CISD's leader, George Michaels is leading the charge to provide the best-in-class tools for the next generation of discovery.
Teske: Why did PNNL establish the Computational and Information Sciences Directorate?
Michaels: PNNL has had pockets of expertise in this area scattered across the Laboratory for years. With CISD, we centralized our expertise and created a critical mass in a variety of areas. This approach allows PNNL to advance the sciences using computation as a tool to better serve the Department of Energy, the Department of Homeland Security (DHS) and other clients.
The team delivering this capability is impressive on several levels. First, the staff has achieved a high level of art in practicing teamwork. The other thing about this group is that it is driven to impact the big picture. Many members are key national figures. They lead initiatives for the government at the lab system-wide level and, of course, within PNNL. We have experts in homeland security, materials science and engineering, mathematics, and information technology and infrastructure. I could spend hours talking about the talent in this group. For the kind of diverse projects we do at PNNL, it is essential to have world-class multidisciplinary contributors as part of the mix. And we do.
Teske: What is unique about PNNL's work in this area?
Michaels: What sets PNNL apart is that everything we do focuses on taking science to solutions. Whatever mission we apply it tonational security, environmental technologies, energy sciencescomputation is an integral piece of delivering science-based solutions. It has become the foundation upon which the other mission areas build and advance their work.
With the current national challenge of data-intensive computing, we have refocused our computational expertise there. The fact is that the nation has spent a lot of time generating a lot of data. One of the fundamental problems with preventing 9/11 had to do with the fact that we had data scattered across multiple agencies. We didn't have the tools to bring it all together. Bringing large data sets together for analysis requires a different computing approach; it requires tools that transform data into information that gives us knowledge we can use. At PNNL we are focused on data-intensive computingit's one of our central activities.
Teske: How is computation directly affecting PNNL's initiatives?
Michaels: One of the ways we are making a difference is in the area of information analytics. PNNL excels in developing sensors and collecting data. Computational sciences provide the tools that allow us to understand that data in real time; we use high-performance computing and data-intensive computing to do it. So whether we are looking at climate change scenarios or cloud physics or doing threat and vulnerability analysis for DHS, we provide the tools that allow that important work to move ahead.
We also contribute substantially to the Lab's emerging strengths in predictive biology and energy sciences, nanoscience, and energy conversions, all which are central to much of the work going on in the fundamental sciences and the environmental technology groups. There is an overlap in the analytics methodologies developed for bioinformatics with those used for national security and homeland security analytics. So what we are doing is very synergistic across the board.
Teske: What was accomplished in the first year?
Michaels: Strategy. We sought to answer the questions: "How are we going to be a world-class computational sciences contributor? What do we need to do to deliver the big picturefor PNNL's mission?"
One of the early challenges of starting a new directorate was identifying who needed to be a part of it. We pulled together folks from several research areasFundamental Science, National Security, Energy Sciences and Technologyto create this group. We started out with about 420 people, and we've hired 50+ more since October 2004.
An exciting development this year is winning a large DOE grant for a new multiscale mathematics program. Our plan is to take mathematics to a level where it can help researchers break through barriers in understanding complex physical processes involving extremely long scales of time or distance. It has potential applications in fuel cell research, efficient engine design, and design of materials atom by atom.
Teske: So now you have a strong team in place and a clear mission. What's next?
Michaels: The big challenge I have put to the team is to double our business volume by 2009. That means we must initiate a variety of new programs and take on a larger client base while maintaining our high level of service. An example of that growing volume and diversity is the recent $3 million project awarded to us by the National Institutes for Health for computational biology work. We are modeling protein properties and particular bacteria that are problematic for folks who have cystic fibrosis.
I also foresee that we will lead the charge into the petascale computing arena. Currently, we're in the terabyte scale. The Environmental Molecular Sciences Laboratory (EMSL) supercomputer has 11.8 teraflops of performance. By 2009, I think supercomputers will be 10,000 times more powerful and be able to address very large, data-intensive work.
With that capability, I see us advancing information-based science, which will revolutionize how scientific research is done. From a computing standpoint, technology currently cannot manage the large-scale and data-intensive enterprises. That means we will have to address the need for new approaches in computing, databases, and knowledge discovery, which is what we're doing right now. Information-based sciences will allow researchers to holistically address very complex problems.
Teske: How do partnerships play into the team's success?
Michaels: We broadly partner with government and industry entities. Partnerships are synergistic because no single entity has all of the expertise to cover every mission area. It makes sense to partner across the lab system because we have common funding and management functions.
For example, we worked with Hewlett-Packard to develop what was the fastest computer in DOE's Office of Science for about three years. We haven't really done much to that machine since and yet it is still among the 20 fastest in the world. We're working on upgrading it through other partnerships so that we can increase its power and applicability to bigger science challenges undertaken by EMSL.
We partner extensively with other national laboratories. Right now, we're working with two laboratories in the Office of Science network to define the needs for the next generation of bandwidth and the most effective use of it.
So partnerships are very useful. I think the joint ownership of our collaborations motivates every partner to own the problem and to work effectively towards an integrated solution.
Teske: What is it about leading the Lab's CISD business that motivates you?
Michaels: Good question. The truth about me is that I really like taking on new challenges. I'm something of a risk takerin terms of going in new directions. So this new direction of data-intensive computing motivates me to make a significant contribution. Another factor is the talented people here; they work well together and like working together. One of my career mantras has been that work needs to be focused and fun. If not, there are plenty of other things to do. The fact is that we are enjoying our work. In taking on new things, we are making a difference in the scientific community...and the world. That's what drives me and the directorate.
This article originally appeared in the Fall 2005 issue of Breakthroughs Magazine, a publication of Pacific Northwest National Laboratory.
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.
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.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
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?