The intent of the TIGER project is to implement and evaluate a "campus bridge" model that addresses the growing need for Cyber-Infrastructure (CI) support for researchers at campuses of every size requiring resources that may not be of petascale size, but that outstrip the infrastructure that can be supported at most institutions. The project will support regional access to HPC/HTC facilities, computational and visualization expertise as well as leveraging of expertise from each participating institution. Computational nodes for regional use have already been purchased via prior EPSCoR and DoE awards. The two HBCU partners will be connected to the regional R&E network via previously awarded ESPCoR RII 2 and ESPCoR CI R2 awards. A key task is to provide expert implementation and integration services and also increase the size of the workforce having this skill sets.
The project will extend the campus experience to a regional one, funding a collaboration of Clemson expertise and faculty at comprehensives and HBCUs and will expand and grow cyber-enabled knowledge discovery communities. A successful outcome is potentially transformative in providing a new model shown to enable and grow a broad and diverse cyber-enabled research community at a regional scale, leveraging strengths at a regional center of CI expertise.
The TIGER project implemented and evaluated a "campus bridge" model. The term "campus bridging" has multiple definitions – our use of the term campus bridging describes activities that address the growing need at campuses of all sizes for (1) computational resources that may not be of petascale size but that outstrip the desktop and infrastructure that can be supported at the local campus and (2) human expertise in how to use the available resources to (3) address the scientists’ specific computational challenge at hand via collaboration with computational scientists. CI resources are moving into "cloud services" which can potentially be located anywhere and in some ways may be considered to be 'ubiquitous'; success in using such resources increasingly depends on expertise in use and integration of these services in a collaboration of technology and domain scientists. The key finding is that providing expert cyberinfrastructure implementation/integration services in conjunction with access to computational resources will increase use of computational approaches in multiple disciplines and increase the size of the workforce having this skill set. The TIGER project had a four-prong strategy in successfully completing the campus bridge model goal. 1. Enable access to existing campus resources, including Clemson’s Palmetto high performance computing cluster. Participating users (both faculty and students) at Claflin University, Francis Marion University (FMU), and South Carolina State University (SCSU) were provided with user accounts on Clemson’s Palmetto HPC cluster. Claflin is a private HBCU, Francis Marion is a public Minority Serving Institution, and SCSU is a public HBCU. Accounts included 100 GB of backed-up storage space in user home directories, as well as unlimited access to a shared 6 TB backed-up storage space. A high-priority job queue on the cluster was established exclusively for these users. Upon account creation, each user was introduced to the comprehensive collection of online user documentation available for the Palmetto cluster and to a computational scientist who would serve as that user’s primary point-of-contact for user support. The guidance and support was customized for each user’s unique needs, enabling quick and effective adoption of the available resources. Flexible customization of computational scientist support was key to enabling the target user community to find value in HPC resources and make use of those resources in their research activities. Use of the Palmetto cluster and computational scientist support was funded by NSF EPS-0919440 (EPSCoR Track II). 2. Connect to XSEDE facilities using regional network facilities funded by NSF ARI-R2 and EPSCoR RII funds. Network connections for SCSU and Claflin were provided using EPSCoR C2 funds (NSF Award EPS-1006833). FMU’s connection was funded by DOE XXX. These connections are not in a regional connected model due to State regulation passed during the award period, but we are able to pass traffic to these locations going through the SoX GigaPOP in Atlanta. This has not caused any issues for this project. The sharing of data and access to computational resources has been at lower bandwidth and slower speeds than we would have liked but were adequate for the High Performance Computing (HPC) access needed for this project. 3. Extend the campus experience to a regional model, facilitating a collaboration of Clemson expertise and faculty who interact with regional researchers and technical staff to grow cyber-enabled knowledge discovery communities There are now over 130 external users involved in this bridging model. The TIGER activity further built a regional community with interest in scientific visualization. Dr. Vetria Byrd, a visualization scientist, and Dr. Jill Gemmill (project co-PI) built a collaboration that included faculty in Clemson’s Visual Computing Department in the School of Computing, and faculty at the three partner schools plus Clemson. Dr. Byrd worked with Dr. Jerry Tessendorf to apply his particle model based visualization software to visualizations in Astronomy and Mechanical Engineering. Dr. Jeannette Myers of FMU achieved new insights into her simulation of two galaxies colliding through this visualization. The group collectively presented their findings through presentations during the Supercomputing 2012 conference, at the Clemson exhibit booth. 4. Expand visualization capability and outreach to South Carolina campuses Effective use of scientific visualization was expanded during this award, includingpresenting visualization as a tool for pedagogical purposes (Internet2 Day at Francis Marion University, February 2012). These activities revealed a desire and need for hands-on experience with visualization tools to aide in understanding the important role visualization plays in knowledge discovery. Dr. Byrd has developed a series of scientific visualization that expose participants to scientific and information visualization by way of hands-on experience with visualization tools specific to their knowledge domain. This approach allows participants to explore various visualization paths at their own pace and help to identify the path most closely related to their data. Once this path is identified, participants are better equipped to choose visualization tools that address their visualization needs.
