NSF XSEDE

NSF XSEDE
Name	XSEDE
Full name	Extreme Science and Engineering Discovery Environment
Established	2011
Predecessor	TeraGrid
Funding	National Science Foundation
Country	United States

NSF XSEDE XSEDE provided a distributed high-performance computing and data infrastructure that connected researchers to supercomputers, data collections, and software across the United States. It integrated resources from multiple centers to support computational science workflows for investigators working on problems in areas including climate modeling, genomics, astrophysics, and materials science. XSEDE offered user support, allocation management, and training to facilitate access to national-scale computing platforms.

Overview

XSEDE linked investigators to large-scale systems such as Intel-based clusters at facilities analogous to those run by organizations like National Center for Supercomputing Applications, service offerings similar to Oak Ridge National Laboratory, and data resources comparable to Los Alamos National Laboratory repositories, while coordinating policies among institutions like University of Illinois at Urbana–Champaign, Purdue University, University of California, San Diego, and Texas Advanced Computing Center. The environment supported workflows used in projects associated with agencies such as the National Institutes of Health, Department of Energy, National Aeronautics and Space Administration, and collaborations with laboratories including Argonne National Laboratory and Lawrence Berkeley National Laboratory.

History and Development

XSEDE succeeded the TeraGrid program and grew from investments by the National Science Foundation to expand capabilities built earlier at centers like the San Diego Supercomputer Center and the Purdue University Research Computing Center. Foundational technologies and management practices drew on experience from initiatives tied to Teragrid, partnerships with regional consortia such as the Ohio Supercomputer Center and projects related to middleware development from groups like Condor Project. Over time, XSEDE evolved to integrate capabilities similar to those in international efforts exemplified by PRACE and Gauss Centre for Supercomputing collaborations.

Services and Resources

XSEDE provided services including allocation review and issuance modeled after practices at NSF large facilities, user portal and single sign-on systems comparable to InCommon, workflow tools similar to Globus, and support for software stacks like GNU Compiler Collection-based toolchains, OpenMPI, and scientific libraries used in communities served by centers such as NCSA and TACC. Resource types included capability computing on leadership-class systems analogous to IBM Summit-scale allocations, high-throughput computing approaches akin to those used by Open Science Grid, data-intensive storage comparable to HPSS systems, visualization resources reminiscent of VizHub deployments, and training programs modeled on courses at institutions like Carnegie Mellon University and Stanford University.

Organizational Structure and Funding

XSEDE operated as a coordinated alliance of resource providers, user services teams, and partner campuses, with a central project office hosting management staff, technical liaisons, and training coordinators drawn from member institutions such as University of Chicago, University of Michigan, and Cornell University. Funding originated from grants administered by the National Science Foundation with oversight practices similar to Office of Management and Budget guidelines and coordination with program officers who engaged stakeholders from Department of Energy laboratories, private partners, and academic user communities. Governance included advisory committees influenced by models at organizations like ACM and IEEE.

Major Projects and Collaborations

XSEDE supported major scientific campaigns in areas overlapping projects such as the Human Genome Project-inspired genomics initiatives, climate work related to Intergovernmental Panel on Climate Change model intercomparison efforts, astrophysics simulations like those pursued by teams associated with Large Synoptic Survey Telescope planning, and materials discovery consortia similar to Materials Genome Initiative activities. Collaborations included partnerships with resource providers at institutions like Argonne National Laboratory, software collaborations with groups resembling the Apache Software Foundation-hosted projects, and interoperability efforts with international infrastructures such as European Grid Infrastructure and PRACE.

Impact and Usage Metrics

XSEDE tracked metrics comparable to those used by large facilities: core-hour allocations consumed by projects from investigators at universities such as MIT, Harvard University, California Institute of Technology, and Princeton University; numbers of supported users analogous to those reported by NCSA; publications acknowledging resource support appearing in journals associated with societies like American Physical Society and Royal Society Publishing; and training outcomes similar to workforce development measures reported by NSF-funded programs. Usage patterns reflected demand from domains including bioinformatics teams at centers like Broad Institute, climate modelers from agencies like NOAA, and computational chemists collaborating with Argonne.

Challenges and Future Directions

Operational challenges included integrating heterogeneous architectures including GPU-accelerated systems comparable to NVIDIA-based platforms, ensuring secure access policies aligned with standards such as those promoted by NIST, and sustaining funding models similar to those debated within National Science Foundation portfolios. Future directions considered transitions to cloud-compatible models exemplified by deployments at Amazon Web Services research collaborations, alignment with data management principles advocated by organizations like Research Data Alliance, and continued interoperability with international infrastructures such as PRACE and European Grid Infrastructure to serve emerging communities at institutions like University of Washington and University of Texas at Austin.

Category:Supercomputing