User Tools

Site Tools


cluster:7


Home

TITLE OF PROPOSED PROJECT

“Acquisition of Cluster Computing Facilities for Research and Education at Wesleyan University”

NATIONAL SCIENCE FOUNDATION PROPOSAL AUTHORS

  • Francis W Starr
  • David L Beveridge
  • Kathryn V Johnston

PROJECT SUMMARY

This proposal describes a request for support from the NSF Major Research Instrumentation Program for establishing a state of the art 128-CPU cluster computing facility for research and educational purposes in the sciences at Wesleyan University. There are established extramurally-funded research programs at Wesleyan in theoretical astrophysics, liquid state chemical physics, nanotechnology, quantum chemistry, molecular biophysics, and the emerging field of neuroinformatics and structural bioinformatics, all of which depend on high-end computing to be competitive. Courses that involve computers are offered in each of these areas. However, there is presently no university based computer facility available for intensive research computing and corresponding initiatives in undergraduate and graduate education. An extensive network of computer workstations and personal computing facilities is now in place at Wesleyan, but those involved in large-scale research computing projects have needed to maintain and upgrade their computing clusters individually. There have been three such clusters set up over the last five years. One of these is defunct, one is obsolete, and the active one – situated in the physics department – is currently saturated and will soon be out of date. Two new hires have requested PC Clusters as part of their start-up requests, and we anticipate others to follow. These additional requests, plus the problems with systems management and obsolescence in the previous clusters have led us to conclude that building and maintaining a science division-wide cluster computer would be a more efficient use of both Wesleyan and individual faculty time and funds.

Working with a University Vice president and Information Technology Services (ITS), the PI’s have formulated a plan to establish and maintain a dynamic state-of-the-art PC cluster computing facility. This purpose of this proposal is to obtain funding for a new 128-CPU cluster. The equipment will be situated in ITS space, maintained by a designated ITS systems manager and made available for research and instructional needs over our current network, WESNET. A faculty Computer Advisory Committee will provide academic oversight, manage allocation requests, and coordinate with ITS. In support of this proposal, the university has also agreed to maintain this cluster at state-of-the-art with a systematic program of upgrades at a rate that will totally replace the hardware on subsequent 3-4 year time frames. Thus the problems of system management and obsolescence are mitigated, and requests for clusters from new faculty will be dealt with by providing an appropriate addition and allocation of CPU access and time on the shared facility.

Realistically, the proposed cluster will not meet 100% of anticipated needs, and it is expected that the highest-end users will make appropriate use of the proposed facility, but also continue to request allocations on national supercomputers for projects that will outstrip local capabilities. Hence the intellectual merit of the proposed computing facilities is that they will provide a thoughtful solution to the problem of providing a reliable scientific computing resource at a small university based on a coherent, university-wide strategic plan. The broader impact of the facility will be on both the faculty research and student education.

Specifically, since faculty will no longer be burdened with maintaining computational resources, more attention can be given to the actual research activities. The cluster will also impact the ongoing instruction in the university, serving as a learning tool to develop students’ scientific computing skills, both in existing courses and assisting faculty with research. Such training is invaluable to prepare students for the expanding field of information technology. Overall, this revision in Wesleyan’s institutional strategy towards high-level computing fits naturally within the university’s mission of achieving excellence in undergraduate education while supporting an active research faculty.

IMPACT OF INFRASTRUCTURE ON PROJECTS

Providing a central facility for high-level computations represents a dramatic shift in the approach that Wesleyan University has taken to supporting science faculty research programs. Currently, computational resources are distributed in an ad-hoc fashion across the campus, with what computing resources we have being maintained by individual faculty with a particular interest in computational work. The impacts of the proposed general purpose cluster computing on performing research are as follows:

  • 1. The facility will enable faculty to produce new science in diverse research projects including the structure and formation of galaxies, molecular dynamics of proteins, elucidating activity patterns in cortical circuits, DNAs and protein DNA recognition, methods developments and applications in molecular quantum mechanics, complex quantum dynamics and mesoscopic transport phenomena, computer simulations of the clustering of nanoparticles and studies of the assembly and properties of soft materials
  • 2. Distributed resources currently are maintained by individual faculty who aim to have enough computing resources to meet their peak needs. As result, computational resources sit idle during non-peak usage periods. A shared facility would allow users to take advantage of computing time that would otherwise go wasted, meaning that the total aggregate computing resources needed not be as large as if they are distributed.
  • 3. A central computing facility and internal computing workshops would provide an enabling environment to bring together researchers from different areas of the sciences and foster collaborative activities. The current distributed model does not encourage collaboration, since individuals need to assure adequate resources for their specific projects.
  • 4. A central facility would not only facilitate the present computational research, but would lower the barrier to initiate new computational projects so that faculty and students can become quickly involved and explore new approaches to their research.
  • 5. Removing the burden of maintaining computational facilities from faculty members will free them to focus on the effective use of resources to strengthen research and educational activities. Moreover, access to such facilities is vital to maintain the competitiveness with larger universities.
  • 6. The cluster will serve as a learning tool to develop student scientific computing proficiency both through existing courses and though assisting faculty with research. Such training is invaluable to prepare students for the expanding field of information technology.
  • 7. Computational facilities quickly become obsolete with the furious pace of technological development. Often, individual faculty are not able to keep up with the pace of innovation lacking either the time needed to stay informed about the latest innovations or funds necessary to buy them (or both). Wesleyan’s ITS is committed to the maintenance and regular upgrading of facilities once they are in place. This is a truly major matching commitment and provides a longevity, continuity and stability to research computing that is currently missing in the current model of distributed resources.
  • 8. Six faculty research groups involving postdoctoral research associates, graduate students and undergraduate students pursuing honors thesis research comprise the primary cadre of users of the cluster. Nine additional groups are expected to be involved in significant but smaller scale computer related research initiatives. There are a number of inter-group collaborations and projects that involve both experiments and computations. In total, there will be roughly 50 regular users of this facility.

Clearly, a centralized cluster computer will introduce a new era to the quality and inclusiveness of computationally intensive research at Wesleyan, affecting both faculty programs and the undergraduate and graduate students involve in those programs. Overall, this revision in Wesleyan’s institutional strategy towards information technology fits naturally within the university’s mission of achieving excellence in undergraduate education via the effective integration of teaching and scholarship.

PROJECT BUDGET

$ 190,000


Home

cluster/7.txt · Last modified: 2019/12/13 08:43 by hmeij07