Feature - From High Performance Computing to Grid
Computing power continues to grow exponentially. Users are demanding ever-more computing power as their applications become increasingly complex.
Marie-Christine Sawley, co-director of the Swiss National Supercomputing Center (CSCS), has the job of matching the rapid expansion of computing technology with the needs of CSCS Users.
“Communities with many different portfolios are using the CSCS: from biology and environmental science, fluid dynamics and material science to physics and astronomy,” Sawley explains. “Our facilities are used by the whole scientific community in Switzerland. Proposals to use our computers are scientifically reviewed and then they get their resources for free.”
Supporting universities and research institutions in Switzerland is only one of CSCS’ areas of activity. Other units are dedicated to system administration, software development, visualization and now grids.
“When I came to the CSCS as general manager in 2003,” says Sawley, “there were less than 20 people working here—a medium-sized center. Today we have almost double the number of people and it is constantly growing. Finding the right people has really been a challenge.”
Sawley came to the CSCS after working at the Ecole Polytechnique Fédérale de Lausanne as head of the User Consulting Group at the Scientific Computing Centre, and Vice President for Research, where she was responsible for maintaining strategic partnerships with IT companies and institutions. With her background in plasma physics and as a graduate of the Mastering the Technology Enterprise program (a collaboration of the IMD Business School in Lausanne, Switzerland and the Swiss Federal Institutes of Technology, in Zurich and Lausanne) she understands both what it takes to run a large center and how to satisfy their clients: the scientists.
Using grid technology is a relatively new development for the CSCS. Two years ago, CSCS became the first Swiss partner of the Enabling Grids for E-sciencE project and is currently involved in a number of other projects such as the SwissBioGrid.
“We see a great synergy between theses two pillars: Grid and HPC,” says Sawley. “Our center started some 15 years ago with a unique machine and today HPC is still represented by a few big mainframe machines. With the Grid, we now have another major system that complements high-performance with high-throughput computing.”
All of these developments are ultimately driven by the science using them and the large-scale computing needs faced by today’s researchers.
“This has also changed the way science is done in large virtual, distributed communities, of which the Large Hadron Collider experiments at CERN are a good example,” says Sawley. “The Grid is going further than just providing an IT infrastructure; it provides a platform for researchers to work together in a distributed environment.”
Sawley predicts the next big developments will be in new tools for data analysis and visualization: “Scientific validation by intuitive tools—that’s another good way for collaboration, sharing and creating innovation.”