Research Report: Campus bridging software and services

The Arlington Memorial Bridge in Washington, DC, crosses the Potomac River. © Christopher Allen, 2009

How should the university campus interface with national cyberinfrastructure (CI)? What role should the university play? These are questions that matter, which is why iSGTW staff were pleased to see the recent report on campus bridging. The report, edited by John McGee (Renaissance Computing Institute), Von Welch (Indiana University), and Guy Almes (Texas A&M University) is based on an August 2010 workshop hosted by Indiana University and held in Denver, Colorado. Read on for a condensed version of the report's executive summary.

The workshop took a broad view of software and services, defining services to include user support, information technology services, and everything in between. The workshop focused on two goals:

• Finding common elements of software stacks widely usable across the nation/world, in order to promote interoperability/economy of scale; and
• Suggested policy documents that any research university should have in place.

Prior to the workshops, we designed an online user survey to capture user experiences with CI, and distributed it to 5,000 scientists who have served as National Science Foundation Principle Investigators.

Nearly half of the respondents indicated they used some CI besides their own workstation or locally controlled CI. The responses we received helped us to identify how CI is used as a bridge, and how well various aspects of CI are working. They also helped guide our workshop discussions.

A number of findings emerged from the survey and workshop, which we group into the following four categories:

Challenges related to software and services

1. Scientists have no coordinated mechanism to discover CI resources and services, and, once discovered, it is a challenge to figure out how to use those resources, determine their policies, find users support, etc.
2. Conversely, it is difficult for CI projects to discover small communities and discern the needs of those communities (as opposed to their vocal minorities).
3. There are significant challenges in measuring the effort spent on and impact of campus bridging and campus-level CI due to the distributed nature of these activities and lack of clear metrics.
4. Scientists are hampered in their use of CI by a lack of coordination and interoperability between CI and campus support mechanisms.

Importance of campuses in CI education and workforce development.

1. A trained workforce is critical to mature, usable CI.
2. The new generation of scientists and students are accustomed to commercial and other computing infrastructure, and they expect similar performance from CI. These scientists also tend to be less accustomed to the low-level usage modalities used by their predecessors.
3. Campuses, as educators of the future CI workforce and scientists, can have huge impacts on both of the previous findings.

Relationship of NSF CI, administrative computing, and commercial CI

1. Computing infrastructure within campuses tends to be split between administrative and research computing. Research computing is more readily integrated as part of a coordinated national CI, but the administrative IT tends to get more attention and funding from campus leadership.
2. Administrative and commercial computing infrastructure tend to be highly polished, emphasizing attributes like reliability and usability. This has led to significant adoption of commercial infrastructure by scientists, according to our CI user survey.
3. While CI should become more polished in order to meet scientist’s increasing expectations, CI also needs to maintain enough flexibility to adapt to the still-evolving needs of collaborative science.

User support and fostering expertise sharing for CI

Effective use of CI for science requires good user support and access to CI expertise. The reward system for many faculty PIs, however, does not motivate supporting scientists. Changing this reward system would be very difficult. Other ways to increase support for scientists using CI would be to provide for more “peer-to-peer” support through user forums and the like, and increasing the CI expertise of campus support staff close to scientists and giving that support staff cognizance of the scientist’s problems with CI outside the campus.

Emerging from these findings were the following recommendations:

• NSF must lead the establishment of a coordinated, national cyberinfrastructure support system to provide user support and expert guidance for using cyberinfrastructure. This system should be constructed in concert with and with contributions from campuses, regional providers, and CI projects. This support structure must be constructed with the backing of campus leadership and coordinated with campus support services to bring support as close to scientists as possible. The system must be neutral with respect to any given CI project or technology in order to put the need of the scientists first.
• NSF must lead the community in establishing a blueprint for a National Cyberinfrastructure. A “National Cyberinfrastructure” composed of CI from state, regional, and federal sources must be coordinated to be effective. Campuses and other CI providers need a better understanding of what the National CI is in order to effectively integrate their activities. A blueprint would provide architecture for this National CI, showing how the different contributions contribute. This blueprint must be trusted to be neutral to any particular project or technology viewpoint, and focused on furthering domain science rather than CS research. There are significant challenges in selecting an appropriate body to create this blueprint to meet these goals as CI expertise tends to be integrated with CI projects and hence seen as biased.
• NSF must continue to emphasize maturity (reliability, usability, etc.) in its review process for cyberinfrastructure. The new generation of scientists and students are accustomed to more mature commercial and administrative campus computing infrastructure. CI needs to increase its level of robustness towards this new bar, while maintaining enough flexibility to adapt to evolving demands of collaborative science.
• NSF must continue to provide leadership towards a national cyberinfrastructure. NSF’s technical vision was just as critical as its funding for the establishment of the NSFNET. While providing such leadership is significantly more challenging at this time, NSF’s voice has impact beyond its funding programs and by continuing to provide a vision and guidance, NSF can significantly advance CI outside of direct funding efforts.

For the full report, visit the report's page in the Indiana University document repository, here.