Building Research Cyberinfrastructure at Small/Medium Research Institutions

Key Takeaways

• While cyberinfrastructure was initially seen as support for scientific and engineering research, scholars in nearly every discipline increasingly require the same range of support to enhance their studies.
• The nearly ubiquitous demand for cyberinfrastructure places an especially heavy burden on institutions not in the top tier of the research hierarchy.
• Defining research needs, setting priorities for research support, developing support strategies, developing a funding model, and building partnerships to support research are key steps in building research cyberinfrastructure at small/medium research institutions.

Along with teaching and service, research is a critical component of the mission at most universities. Creating and sharing new knowledge across a broad range of disciplines enhances the intellectual life of both faculty and students, and research productivity often serves as a yardstick by which university reputations are measured. At larger universities, research may be deeply embedded in the institutional culture, while at small/medium research institutions, a research agenda might require incubation, nurturing, and development of appropriate support. Small/medium research institutions might have fewer large projects, less indirect cost recovery, and fewer possible economies of scale than large universities. Nevertheless, research remains important to the well-being of those institutions, and their faculty expect and deserve the best support possible.

A 2006 ECAR study defined cyberinfrastructure as the coordinated aggregate of "hardware, software, communications, services, facilities, and personnel that enable researchers to conduct advanced computational, collaborative, and data-intensive research."¹ Further, "…IT professionals who view the future of IT in research merely in terms of network speed or computing cycles are missing the boat."² While cyberinfrastructure was initially seen as support for scientific and engineering research, scholars in nearly every discipline increasingly require that same range of support as they come to understand the power of computation to enhance their studies. This nearly ubiquitous demand for cyberinfrastructure places an especially heavy burden on institutions not in the top tier of the research hierarchy.

Small/medium research institutions often lack dedicated staff and departmental structures to support faculty research, so they might need to take a different approach to developing an adequate cyberinfrastructure. The campus culture at small/medium research institutions could result in inappropriate infrastructure to support research, inability to actively promote support for research, conflicting priorities for cyberinfrastructure funding, reduced agility in providing needed computing resources to researchers, and lack of awareness by researchers of the limits of institutional infrastructure. These realities at small/medium research institutions can result in dissatisfied researchers and delays in conducting research activities.

One of the first challenges for a discussion about building research cyberinfrastructure at small/medium research institutions is defining a "small/medium research institution." A loose definition is any institution with significant research activities that isn't classified as RU/VH: Research University (very high research activity) in the Carnegie Basic Classification. "Research cyberinfrastructure" also means different things to different people. A 2008 ECAR research study found varying levels of adoption of the five research cyberinfrastructure categories that the study considered.³ Both factors make it hard to present a single method for building research cyberinfrastructure at small/medium research institutions. Instead, this article presents five key steps for building research cyberinfrastructure.

To build a respectable cyberinfrastructure, the IT organizations at small/medium research institutions need to use creativity in discovering the needs of their researchers, setting priorities for support, developing support strategies, funding and implementing cyberinfrastructure, and building partnerships to enhance research support. This article presents the viewpoints of four small-to-medium-sized research universities who have struggled with the issue of providing appropriate cyberinfrastructure support for their research enterprises. All four universities have strategic goals for raising the level of research activity and increasing extramural funding for research.

Table 1. University Profiles

Our four universities employ a variety of approaches based on diverse institutional experience, but with similar challenges in supporting research. Action steps for creating a research support structure include:

1. Define research needs
2. Set priorities for research support
3. Develop support strategies
4. Develop a funding model
5. Build partnerships for research support

Defining Research Needs

Institutions trying to create a research culture will need a process of requirements discovery and elicitation to uncover and define the research requirements of the university community. In well-established research support operations, large research teams either support themselves or have departmental or college-level support. Less research-intensive institutions often lack that departmental support, hence researchers must rely on the central IT organization for support — but the culture has not developed clear processes for obtaining research support. Additionally, the institution might still be working on establishing an overall strategy for its research enterprise. So, in addition to discovering the needs of individual researchers and research clusters, the IT unit must understand the broad strategic directions the institution has in mind for its research programs. Because research needs constantly change as the institution emphasizes different discipline areas and as the disciplines themselves evolve, the discovery and definition of research needs must be an ongoing process, not a one-time exercise.

Miami University restructured the IT organization and hired its first CIO in 2003. One of his first initiatives was a strategic planning exercise during which faculty identified the need for more IT support for faculty research. This resulted in the formation of a Research Computing Support (RCS) group within the IT organization and the purchase of a small high-performance computing (HPC) cluster. At the same time a faculty group had received an internal grant for activities to advance the use of computing in faculty research. This faculty group provided important input into the structure of the RCS group, which initially aimed to provide HPC programming, general scientific programming, statistical support, and support for the use of databases in research. As the RCS group worked with faculty to further understand their needs, the support focus expanded. For example, Miami only has a limited number of graduate programs, so many students involved in research are undergraduates or master's students who might only be involved in a project for a year. The RCS group found that providing personalized training could reduce the time it took for students to become productive members of a research team, and management of software developed for a project could help smooth the transitions as students started and completed their work on a project.

Oakland University made an effort to engage faculty in discussion through a series of discovery meetings, but a lack of understanding about the goals and poor attendance led to cancellation of the meetings. Instead, the university IT organization, University Technology Services (UTS), requested from the Office of Grants, Contracts, and Sponsored Research a list of faculty who received National Science Foundation grants. A senior IT staff member familiar with a broad range of IT services was assigned to interview each faculty member. The interview questions focused on research information:

• Processing
• Storage
• Transit

and understanding the IT service needs of researchers that UTS could address. Also, the CIO met with the Research Office and faculty groups to determine other support needs, including administrative support and purchasing functions, that faculty members had identified as cumbersome.

As part of its strategic planning process, UMass Boston established a committee on research and graduate studies to create a vision and recommend goals for a comprehensive plan to enhance the research enterprise. The final report, issued in 2007, included recommendations to improve a number of support services that university researchers found lacking, including enhanced network bandwidth for data-intensive activities, expansion of central data storage, and increased support for multiple operating systems. Additionally, in 2006 the university had commissioned a study from the Battelle Memorial Institute to recommend potential focus areas for strategic expansion of the university's research enterprise. In 2008, four working groups prepared more detailed reports about specific research clusters the university wanted to emphasize: urban health and public policy, STEM education, computational sciences, and developmental sciences. Each of these working group reports identified specific staffing and technology needs for each area. Among the most consistently mentioned staffing needs were database support, statistical analysis support, and storage engineering. These two sets of reports provided the IT unit with a better understanding of the university's overall research strategy and a starting point for planning how to meet these research needs. Additionally, following the advice of some peer institutions with more established research support structures, the IT unit developed a short set of questions for new faculty, to obtain information about their specific research needs and expectations including networking requirements, security requirements, and research equipment that they might be planning to bring with them to the university.

Efforts at UWM started with creation of a vision statement by the CIO that was shared with and endorsed by two faculty governance committees, the Research Policy Committee and the IT Policy Committee. Results of a campus-wide survey established broad categories of cyberinfrastructure needs, but a deeper understanding of researcher needs was believed necessary. Therefore, one of the first activities for the newly established director of cyberinfrastructure was to collect information from campus researchers regarding technology needs across a broad range of disciplines. The CIO and director of cyberinfrastructure met with the deans and other appropriate individuals in each school to discuss their school's significant research directions and goals. At these meetings, the deans were solicited for their recommendations of faculty who should be interviewed regarding computing support for their research and scholarly activities. Each of the interviews with the recommended faculty focused on three goals: understanding the faculty member's research, analyzing the individual faculty member's cyberinfrastructure needs, and collecting information to help inform the campus dialogue regarding campus-wide needs. The interviews led to a number of successful outcomes, including:

• Identifying existing resources useful to researchers about which they had not previously been aware
• Facilitating collaborations between researchers with similar or complementary research interests
• Assisting the campus with identifying common cyberinfrastructure needs to support research
• Generating good will toward the central IT organization and "high touch" interaction opportunities for IT with faculty

Setting Priorities for Research Support

Given limited resources for research activities at small/medium research institutions, establishing funding priorities for larger scale or centrally provided research activities assumes great importance. However, the wide variety of research and scholarly activities carried out at higher education institutions can make obtaining consensus on priorities challenging.

At Miami, the RCS group's first priority was simply to find faculty projects on which to work, while today the group has a full workload. The university's strategic goals include involving students in research, increasing the level of scholarly accomplishments, contributing to larger scale communities, and maximizing resources. These goals translate into a focus on expanding the number of faculty with whom the RCS group works, helping students learn to use research cyberinfrastructure, providing support for collaborations with researchers at other institutions and in industry, and supporting grant applications. The RCS group also works to transfer skills and knowledge to faculty and students that allow them to expand their use of cyberinfrastructure resources with limited support from RCS.

Following up on what was learned in the private interviews, Oakland University established priorities through the existing University Senate Academic Computing Committee, the faculty committee charged with issues of academic and faculty IT. Further review of priorities was done by the Academic Council, a leadership group with deans and other academic area leaders, led by the provost. Using existing university governance structures supported the goal of building communications lines that expose and prioritize research needs within the university culture. Priorities from these organizations emphasized a need for improved research administration as well as ongoing capacity for network growth and operational stability.

At UMass Boston, knowing that all of the needs identified could not be met at once, the IT leadership team took the directions given in the reports, plus input from faculty researchers and the vice provost for research, and developed a five-year plan for improving research support. Having a written plan gave the CIO a vehicle for discussion and a structure for making budget requests to implement the plan. Two of the highest priorities were increasing available storage and providing more research support staff. As a result, one major initiative in the plan is enhancing storage; another is gradually building a team within IT dedicated to research support, starting with a budget request for a position to coordinate research support and give research faculty a single point of contact. An ongoing governance structure for research support is still in the planning stages.

UWM formed a cyberinfrastructure working group comprised primarily of faculty to identify and prioritize services needed to support campus research and scholarly activities, as well as to recommend best practices for funding cyberinfrastructure services. The working group presented its recommendations to the chancellor, provost, CIO, and vice chancellor for research and economic development. As a result, the central IT organization, in collaboration with two schools, has deployed the first shared HPC cluster service for campus and is actively pursuing the other priorities identified by the working group.

Developing Support Strategies

Support strategies need to strike a balance between providing a large amount of support for a small number of projects and providing more general support to a wider range of users. Faculty input or review can be useful when developing support strategies so that faculty have a clear understanding of all the routine activities for which the support staff is responsible and how the remaining support resources will be allocated.

At Miami, the RCS group was created at the same time that the university installed its first HPC cluster, and support strategies for both resources were established simultaneously. An initial focus of the RCS group was finding faculty projects to collaborate on, so large blocks of time were allocated for these projects, with the remaining time allocated for general support of the computing cluster and research software packages. An important aspect of any large faculty collaboration involving software development is identifying a method for developing the skills of faculty and students on the project so that they can take on much of the ongoing support and development work after the initial development work ends. Since many of the students involved in research at Miami are undergraduates or master's students, consideration was given to picking technologies with which students would already have experience. For example, for one mechanical engineering project, a complex parallel simulation framework developed using C++ linked with simulation code written in MATLAB. With this tool, the simulation code, which will need to change as research progresses, is written in a language that mechanical engineering students learn as part of their course work. Therefore ongoing support by the RCS group will be limited to support for the simulation framework and Matlab integration, which should not need to change very often. Another challenge has been that successful collaborations never end — they just lead to more projects. However, the group needs to respond to requests from other faculty members as well. This challenge has been addressed by having the RCS group staff member handle the high-level software and algorithm design while students work on implementation and testing.

At Oakland University, faculty members most frequently stated that their network connection, e-mail account, and desktop computer (or in a few cases a local mini-tower) were their most important computing resources. Some specialized areas in engineering, chemistry, and physics reported using slightly more resources, including data center hosting for research clusters. Faculty did not actively state security requirements, so IT staff members had to probe more about security. Oakland determined that the central IT organization could emphasize five support areas:

1. Edge client: Providing consistent support for network connectivity, e-mail, and desktop
2. Collaboration: Promoting available shared collaborative document systems including both the campus implementation of Xythos and off-campus tools in Google Apps, and determining when use of either service is appropriate
3. Hosting: Offering to host servers in the central data center and reviewing possible cloud opportunities for storage or processing services (see Figures 1–3)
4. Security: Providing assistance or implementation of special security requirements
5. Administrative: Identifying and implementing an improved research support system

Establishing connections with research faculty based on those support areas provides a foundation for building additional support later. Interviews with research faculty indicate that trust in IT services only extends to one service failure.

Figure 1. New Electrical and UPS Upgrades in the Oakland University Data Center

Figure 2. Working on the Last Cooling Upgrades at Oakland University, Spring 2010

Figure 3. Preparing the Oakland University Data Center for the Next Generation of Research Systems (New Electrical, UPS, HVAC, Floor Tiles, and Server Racks)

Since UMass Boston had very little IT support available for researchers, the IT organization started with some very basic steps — providing physical space for collocated servers and housing a small HPC cluster; both depended on refurbishing the existing data center, as detailed in the next section. IT also developed a collaborative workspace through Xythos that met at least some needs for shared online research materials. More recently, with assistance from the vice provost for research, IT added a modest storage array (4 terabytes) to support research data. The IT organization also worked with faculty to determine priorities for adding software licenses and modestly increased the availability of software to support research, adding, for example, licenses for NVivo, a tool that supports qualitative research. Adding more storage space and creating at least one position dedicated to research support are the next planned strategies, to be implemented as the budget allows. It is also worth noting that the Healey Library at UMass Boston has taken a lead role in increasing the availability and accessibility of electronic resources for researchers and in developing a digital repository that will make it easier for researchers to share either completed research or research in progress. These initiatives, which go beyond hardware and software, can also be key components in building a respectable cyberinfrastructure and a culture of research support.

UWM is developing a research cyberinfrastructure support model based on leveraging personnel from both college-level IT groups and the central campus IT organization as appropriate. The newly formed HPC cluster service is supported by a hybrid support structure, with use of cluster resources based on the level of resource investment in the cluster service by each college. Funding for cluster system administration resources are provided at the college level and supervised through the central IT organization. Cyberinfrastructure facilitators, who assist researchers with developing and running codes on the cluster, are provided by participating colleges and are affiliated either with colleges or the central IT organization.

Funding and Implementing Cyberinfrastructure

"Seize the day!" should probably be the motto for supporting research at small/medium research institutions, particularly state-funded institutions. Extra state dollars for infrastructure are few and far between, so the enterprising IT unit must be prepared to make the most of any such opportunity.

At Miami, funding for RCS staff is part of the base budget for the IT Services organization. Funding for Miami's initial HPC cluster came from central university funds. Funding for the recent HPC cluster replacement was discussed between IT, the provost, and the deans of the two academic units with the largest number of cluster users, and final funding was provided by IT Services and the two academic units. Several faculty members have included budget line items for "Research Computing Support" in grant proposals, but none of these requests have been funded so far. It is expected that any funds received through grants would be used for replacement of or upgrades to cluster hardware.

Oakland recognized that an updated data center was needed to best support research hosting services. The university has relied on incremental change through allocation of some small portion of the annual budget and occasional support in terms of one-time funds made available for specific projects. With a long-term plan that spanned multiple years, the data center was upgraded with a new fire suppression system, new HVAC units and better cooling plans, a new UPS, updated electrical and security infrastructure, leak mitigation, and standardized racks and floor design. With this investment, the facility is more attractive to researchers. Also, strong presentations for one-time funds resulted in the implementation of Xythos for storage of work-in-progress research papers and data files. A partnership with the university library resulted in the implementation of DSpace for research presentation and data file archiving. Faculty have been active in seeking research support funding for specific technologies.

UMass Boston managed to squeeze a data center upgrade out of a state building maintenance project and took advantage of the upgrade to add a caged collocation space where researchers could house servers in a secure and reliable environment. After years of leaking ceilings and inadequate power and cooling in the data center, researchers were understandably reluctant to trust their servers in that space (see Figure 4). However, after a multi-million dollar overhaul, the Information Technology Services Division (ITSD) finally had a space that could effectively handle at least some researchers' needs (see Figure 5). The 700 square foot collocation space offers lockable data cabinets and power feeds protected by UPS and a dedicated generator as well as a fault-tolerant cooling infrastructure. PIN access to the space and video surveillance provide security. As researchers request access to the space, IT staff meet with them and help assess whether the collocation area can meet their needs. Researchers manage their own systems in the collocation space and have 24-hour secure access. By early 2010, the collocation space housed 30 academic servers and associated back-up hardware.

Figure 4. UMass Boston Data Center Before Upgrade

Figure 5. UMass Boston Data Center After Upgrade

The new data center also provided space for the university's first HPC cluster, a 32-node machine with about 1 TB of storage. The IT unit provides support in the form of a half-time position to maintain the cluster and some troubleshooting support for users. The HPC is primarily used by researchers in physics (where faculty start-up money funded the cluster), but other faculty can get at least some computing cycles from it. The relatively small size of the cluster definitely limits the amount of research supported, and the research community is pursuing grant opportunities and partnerships with other institutions to expand the existing cluster and provide more computing cycles (see the next section).

UWM's CIO reallocated internal staffing resources to support cyberinfrastructure and brought in external revenue to fund the director of cyberinfrastructure position and a second data center. The new HPC cluster service was developed through a collaboration between the Office of the Provost and Vice Chancellor for Academic Affairs and the College of Engineering and Applied Science (CEAS). Initial purchase of the cluster hardware and renovations to the data center space to house the cluster were funded using CEAS faculty start-up funds and funds from Academic Affairs. Funding for personnel resources to support the cluster is provided by the participating colleges and the central IT organization. In addition, the provost and the vice chancellor for research and economic development recently agreed to allocate five percent of campus indirect funds per year to the library and central IT to support research.

Building Partnerships for Research Support

Developing an ongoing relationship with stakeholders within the institution as well as with outside agencies is a key to successfully providing and supporting research cyberinfrastructure for all institutions, but especially for those in the early stages of building a culture of research support. The challenges to building relationships will vary as much as the researchers involved, but the goal in all cases should be for IT to be seen as a productive partner or collaborator in the research activity of the institution.

Building partnerships has been the most important factor in the success of research support efforts at Miami. These partnerships have resulted in several interesting projects, but more important have been the efforts that the faculty partners have made to tell their peers how research support has advanced their research programs. While the research support group at Miami has made many efforts to inform faculty about available research support services, referrals by existing faculty partners have been the most effective way of expanding the reach of these services. Faculty partners have also been a key asset in helping department chairs, deans, and IT leaders understand how research cyberinfrastructure services have advanced research efforts at Miami. Partnerships can go beyond just efforts related to faculty research — the opportunity to discuss the research support available at Miami has been a notable factor in several recent faculty searches and candidate interviews.

The importance faculty members place on the network reinforced Oakland University's participation in the state-wide network operated by Merit Network. Ongoing participation allows the university to utilize high-performance research networking capability suited to higher education. Faculty appreciate a quality data center facility, with supporting staff, to which they have access.

At UMass Boston, collaboration has been a key to enhancing support for research. The vice provost for IT and the vice provost for research both share the goal of helping the institution develop a long-term plan for research support. Another element of cyberinfrastructure that requires enormous collaboration is information security. Researchers expect to have secure access and secure sharing of their data, even though the data is almost never under the direct control of IT. Besides taking steps to achieve network security, UMass Boston has also put together an Information Security Council that helps the academic units better understand and implement security measures for their research and other data that requires protection.

In addition, the UMass Boston CIO has worked with the CIOs of the other UMass institutions to facilitate collaborative efforts within the system to provide better support for all UMass researchers. Several UMass schools are jointly developing a virtual computing lab infrastructure that they can leverage to provide enhanced computing resources to researchers as well as better access to computing resources for students and faculty. Additionally, UMass Boston joined a consortium of Boston-area higher ed and health care institutions to facilitate data sharing and collaboration among bio-med researchers. Finally, the University of Massachusetts system joined forces with other Massachusetts institutions to develop a regional high-performance computing center that has the potential to provide enhanced resources for the smaller institutions in the consortium.

Collaboration has been and will continue to be key to UWM's research cyberinfrastructure support. Partnerships involving the Office of the Provost and Vice Chancellor for Academic Affairs, the Office of the Vice Chancellor for Research and Economic Development, the central IT organization, the UWM Libraries, college deans, and the faculty provide vital support. In addition, UWM is partnering with other institutions throughout southeast Wisconsin to form a campus grid that will link computing resources across multiple entities.

Lessons Learned

As the competition for students increases and the budget climate becomes more restrictive, small/medium research universities are looking to increase research activities to raise campus prestige and obtain higher levels of extramural funding. It is important to understand the challenges that small/medium research institutions face when planning effectively for building and maintaining an appropriate campus research cyberinfrastructure.

Investments in cyberinfrastructure can also help small/medium research institutions attract faculty with high-quality research programs. Effective cyberinfrastructure includes many elements, from hardware and software to staffing. Appropriately skilled and trained staff who can facilitate the use of technology by researchers are key to success. These individuals must not only possess IT knowledge and skills but also must be able to match researchers' needs to the appropriate technology services. Data security and library support are also part of this equation and should not be overlooked. Additionally, rather than investing in expensive, specialized research equipment and lab facilities for use by a small number of faculty, schools can invest in commodity computing hardware to support research in a wide range of disciplines.

Given the greater resource constraints, additional effort towards collaboration and resource sharing across multiple campus areas will likely be necessary. Although individual principal investigators or schools might be unable to provide sufficient resources for a given research activity, combining resources across multiple researchers or schools and units may yield adequate or superior resources. Additionally, "above-campus" resources, such as services offered by other higher education institutions, consortia, or public cloud vendors, should be leveraged to supplement locally provided services.

As with many other activities in IT, the most important and challenging aspects of building research cyberinfrastructure have little to do with technology and more to do with building trust and establishing good communications with the constituents. Effective outreach from central IT to faculty might be needed. The "build it and they will come" approach is not generally effective. Time must be invested in listening to faculty needs and building a line of trust for and openness about central IT facilities and systems. Priority setting is essential and may require new governance structures specifically responsible for research technologies.

Each of our institutions has made progress in establishing and advancing its state of cyberinfrastructure by nurturing relationships and working with the appropriate campus constituencies. Although our approaches differ, common strategies and themes exist. We believe that other small/medium research universities can benefit from our experiences to achieve success in their own research cyberinfrastructure efforts.

1. Harvey Blaustein, with Sandra Braman, Richard N. Katz, and Gail Salaway, "IT Engagement in Research" (Roadmap) (Boulder, CO: EDUCAUSE Center for Analysis and Research, July 2006), p. 2.
2. Harvey Blaustein, with Sandra Braman, Richard N. Katz, and Gail Salaway, IT Engagement in Research: A Baseline Study, (ECAR Research Study, Volume 5) (Boulder, CO: EDUCAUSE Center for Analysis and Research, August 2006), p. 32.
3. Mark Sheehan, Higher Education IT and Cyberinfrastructure: Integrating Technologies for Scholarship (ECAR Research Studies, Volume 3) (Boulder, CO: EDUCAUSE Center for Analysis and Research, June 2008).
4. Battelle Technology Partnership Practice, "Research Reenvisioned for the 21st Century: Expanding the Reach of Scholarship at the University of Massachusetts Boston," Battelle Memorial Institute, 2006, p. 1.

© Anne Agee, Theresa Rowe, Melissa Woo, and David Woods. The text of this article is licensed under the Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 license.