Introduction
High Performance Computing (HPC) is increasingly vital for diverse research fields, supported by the National Science Foundation (NSF) and the Department of Energy (DOE) through advanced national cyberinfrastructure facilities These include the NSF's Track 1 and Track 2 facilities, along with DOE's Leadership Class Facilities, which provide elite users access to cutting-edge computing resources The petascale Computational Science and Engineering applications at these centers facilitate the modeling of complex phenomena that are challenging to measure through traditional methods Access to such tier 1 facilities empowers scientists and engineers to accelerate discoveries, generate new insights, and drive innovation.
National resources offer powerful computing capabilities essential for researchers tackling complex scientific and engineering challenges However, participants in the NSF Workshop agree that most researchers still depend on departmental, campus, or regional computing resources A forthcoming report from the Campus Bridging survey supports this view, indicating that comprehensive surveys of the HPC ecosystem would likely confirm it These local resources are vital for meeting fundamental computational needs and for training the next generation of scientists and engineers, which is crucial for maintaining competitiveness and national security Additionally, some researchers utilize these resources to prepare their software for future use in national facilities.
Many universities are striving to create cost-effective and sustainable second and third tier cyberinfrastructure (CI) facilities that provide high-performance computing (HPC) resources to their research communities However, the recent economic downturn poses significant challenges to the funding, organizational structure, and long-term sustainability of these facilities While first and second tier facilities funded by the NSF and those associated with the DOE are also experiencing budget pressures, this workshop emphasizes the financial challenges faced by unit, institutional, and regional CI facilities as the NSF transitions from the TeraGrid to a new funding model, intensifying competition for resources Identifying viable sustainability models for these CI facilities is now more critical than ever, as their survival is essential for advancing national science and engineering discoveries.
Academic institutions adopt diverse strategies for research computing, with some viewing it as a strategic investment and offering consistent support for substantial research infrastructures like large parallel clusters in designated centers Conversely, other universities treat research computing as a tactical necessity, providing sporadic funding for smaller, informal setups Regardless of their approach, these facilities face challenges in effectively organizing, managing, funding, and optimizing their hardware and personnel.
Industry standard computing solutions provide a low cost of entry into HPC hardware, but there are significant hidden costs, including:
Building renovations, including space, power and cooling
Administrative staff to install, maintain and support computational resources and research users
Additional infrastructure requirements such as disk storage, backup, networks and visualization
Consulting staff that can support optimization and scaling for research codes, as well as assist researchers in discovering and leveraging national resources and funding opportunities.
To maintain our competitive edge in research, it is crucial to sustain and expand our national research computing ecosystem A recent workshop provided a platform for leaders from academic institutions to discuss the requirements, challenges, and solutions related to sustainable cyberinfrastructure This report highlights the key findings and recommendations from the workshop, aimed at fostering awareness and promoting ongoing collaboration The discussions led to a shared understanding of effective organizational, funding, and management models for cyberinfrastructure facilities Additionally, the workshop emphasized the importance of transparency among existing facilities, which is vital for justifying the significant expenses and demonstrating the Return on Investment (ROI) necessary for establishing new facilities.
Workshop Objectives and Participation
The workshop aimed to facilitate an open dialogue among center directors, campus Chief Information Officers, and Research Officers regarding Sustainable Funding and Business Models for Academic Cyberinfrastructure (CI) Facilities The event gathered over 80 experts in high-performance computing (HPC) and cyberinfrastructure from across the nation, alongside industry representatives, to explore innovative funding strategies and sustainable business models.
Dr Jennifer Schopf from the National Science Foundation (NSF) attended a workshop at Cornell University from May 3-5, 2010, which included 32 additional participants who accessed the sessions through WebEx The workshop's participants, both in-person and online, are detailed in "Appendix D" and "Appendix E."
All participants were strongly encouraged to submit position papers covering any or all of the proposed workshop discussion topics, including:
Metrics of success and return on investment.
Please see “Appendix G” on page 51 of this report for links to the 28 workshop position papers as well as
“Appendix H” for other useful papers and publications.
The invited presentations and breakout sessions aimed to encourage active participation, enabling attendees to concentrate on and offer in-depth feedback on various topics For additional resources, Appendix F contains links to all workshop presentations and summary notes from the breakout sessions.
Organizational Models
To effectively compare institutional models for research computing and cyberinfrastructure support, it is essential to first understand the different reporting structures, advisory boards, and interactions with key users Most workshop participants were affiliated with one of four primary organizational structures.
1 A director reporting to the Chief Information Officer (CIO) of the university, as part of overall campus IT mission
2 A director reporting to the Vice Provost/President/Chancellor for Research (VP/CR), as part of the overall campus research mission
3 A director reporting to the Provost/Chancellor as part of the overall campus infrastructure mission
4 A director reporting to one or more Deans of heavily invested colleges, often in conjunction with the CIO or VP/CR, as part of a focused research mission for specific college(s).
Cyberinfrastructure facility directors can be tenured or non-tenure-track faculty members, with some holding adjunct faculty positions that include teaching responsibilities Faculty directors often have personal research needs that align with computational resources, allowing them to advocate effectively for these services within their organization In contrast, non-faculty research staff directors prioritize the service mission of campus CI centers, as they are less burdened by teaching and research obligations, enabling them to focus solely on enhancing service delivery.
Faculty Advisory Committees and other types of oversight boards can be useful to CI directors Faculty Advisory Committees typically perform the following functions:
3 Promote the requirements of researchers
4 Provide input on allocation and policy decisions.
Oversight boards, which often consist of faculty members alongside industry representatives and external colleagues, play a crucial role in providing guidance on various key areas.
Regional Organizational Models
As the demand for computational and data analysis capabilities increases, institutions face challenges in securing funding and hiring qualified staff to manage cyberinfrastructure resources To tackle these issues, many are forming regional partnerships to share costs and expertise, leading to the establishment of shared data centers that can adapt to growing research needs through phased deployments These regional data centers also enable the adoption of green technologies for efficient power and cooling Notable examples of such initiatives include the University of California institutions, the New Jersey Institute for Modeling and Visualization, and the Massachusetts Green HPC Academic Research Computing Facility Additionally, regional models utilizing grid technology, like SURAGrid and the Great Plains Network, offer collaborative research capabilities that allow institutions to provide resources and services on a larger scale than they could independently afford.
Regional facilities serve as a powerful driver for economic, educational, and workforce development, enabling individuals and organizations to strategically concentrate their efforts on a collective set of community resources and services This collaborative approach not only enhances local initiatives but also offers researchers the opportunity to expand their work from campus-level projects to broader regional or national platforms.
Requirements for Resources and Services
Every academic institution has unique research computing needs, making it essential to first understand the specific requirements of researchers, students, and faculty This involves identifying the necessary services and resources deemed strategic by the institution A comprehensive inventory of existing resources and a detailed cost analysis, including hidden costs, are crucial for developing services that meet user needs For instance, supporting a high-performance computing (HPC) system entails not only physical infrastructure but also staffing, office space, and training costs A thorough accounting of these expenses is vital for revealing the total cost of operating cyberinfrastructure resources, which is necessary for creating a sustainable funding model Once user requirements and costs are clearly defined, negotiations for institutional support can commence However, without adequate institutional backing or external funding, maintaining or establishing new resources may prove challenging.
A successful negotiation for resources and support requires a mission statement that aligns with the goals of the institution and resonates with faculty and researchers In the current economic climate, administrative management must evaluate each CI investment by considering its cost, impact, strategic potential, and alignment with institutional objectives Utilizing data and engaging prominent faculty and researchers to endorse this data will enhance the effectiveness of the negotiation process.
The workshop highlighted the diverse activities, resources, and services offered by Cyberinfrastructure (CI) facilities, emphasizing that each institution presents a unique combination of these elements Key services discussed included consulting, which involves providing professional technical support for effectively utilizing cyberinfrastructure resources, facilitating multidisciplinary research, and offering expertise in data analysis and emerging technologies Additionally, the workshop covered computing resources, which vary by mission and funding and are essential for meeting user needs Maintaining up-to-date computing resources and exploring new technologies are crucial, as exemplified by North Carolina State University's Virtual Computer Lab (VCL), which adapts to the evolving requirements of its users.
Data storage services are essential for managing the increasing volume of data generated by scientific instruments and simulations, necessitating high-performance storage solutions and backup strategies As funding agencies like the NSF and NIH begin to require data management plans, institutions must reassess their data storage approaches, affecting dataset creation, format usage, and long-term data curation Networking is crucial for accessing cyberinfrastructure resources and efficiently transferring large datasets for analysis and visualization The growing importance of data visualization tools, ranging from workstations to high-resolution immersive graphics, enhances data analysis capabilities Education and training initiatives are vital for developing a skilled workforce in computational and data-enabled science, delivered through workshops and academic courses Software development efforts aim to improve the usability of cyberinfrastructure resources, ensuring optimal utilization by researchers The rise of virtual organizations highlights the need for infrastructure that facilitates collaboration, while outreach activities aim to support new users and expand the impact of cyberinfrastructure resources Finally, economic development initiatives focus on providing local and regional industries with competitive advantages through information sharing and partnership agreements.
Funding Models and Budget Sustainability
Successful cyberinfrastructure facilities share three key characteristics: a compatible organizational model aligned with institutional missions, a resource portfolio that meets the evolving needs of the research community, and a funding model that matches the scale of their objectives, whether local, regional, or national Workshop participants emphasized the importance of developing sustainable funding models to maintain a skilled technical staff and up-to-date computational infrastructure, especially in light of budget constraints caused by economic downturns.
Participants emphasized the necessity of frequency and clarity in communicating the core principles of the computational infrastructure (CI) community, which may not be evident to administrative personnel in universities or research organizations They highlighted that computational science serves as the third pillar of science, alongside experimental and theoretical approaches, a fact recognized in scientific literature and governmental reports Furthermore, advanced research computing is now an essential tool, comparable to utilities and communication networks, making access to computational resources vital for researchers to maintain their competitive edge in rapidly evolving fields The exponential growth of data generated and required by researchers necessitates robust resources for storage, analysis, and visualization, supported by skilled personnel The demand for computational resources is expanding across various disciplines, including social sciences, economics, and humanities, alongside traditional fields such as astronomy and bioengineering, underscoring the need for scalable solutions to address complex questions Lastly, the increasingly interdisciplinary and collaborative nature of contemporary research demands professional staff proficient in developing CI tools and technologies to effectively convert vast amounts of data into actionable knowledge.
During the workshop, various funding and budget models were discussed, highlighting that there is no universal solution suitable for all situations, nor is there a single model that remains effective over time without adjustments Dr Eric Sills, Director for Advanced Computing at North Carolina State University, emphasized this idea in his position paper.
"Sustainability evokes the feeling of perpetual motion - start it and it sustains itself - but sustainability actually requires nearly continuous ongoing work, adaptation, and adjustment."
To achieve success, organizations must possess a comprehensive understanding of their computational and data analysis needs, a clear mission statement that aligns with these requirements, and a commitment to a sustainable funding model Key elements of an effective funding model include a strong value proposition that efficiently meets researchers' needs, transparency in cost-sharing to build trust among faculty, fairness in resource allocation to ensure equitable access, economies of scale to reduce institutional costs, and clearly defined base funding that distinguishes recoverable costs from core infrastructure expenses By focusing on these qualities, organizations can foster a broad user community, enhance collaboration, and optimize resource utilization, ultimately driving research success.
Understanding Costs and Potential for Recovery
Operating an academic cyberinfrastructure facility involves four major costs, with the primary goal of any sustainable budget model being to cover these expenses One significant cost is staffing, as the required skill level for cyberinfrastructure facility personnel is typically higher than that of staff in other institutional research support facilities While some institutions may manage their CI centers similarly to other facilities, many cases reflect a distinct operational approach due to the specialized expertise needed.
CI staff members possess advanced expertise in computational science, essential for developing, optimizing, and maintaining applications for cutting-edge computational and data analysis resources Their diverse scientific domain knowledge enhances the value of CI facilities, whether local, regional, or national However, recruiting and retaining skilled staff is challenging due to extensive staffing requirements and relatively high salaries, compounded by human resources overheads like vacation and training, making cost recovery difficult Additionally, the infrastructure necessary for CI operations demands significant data center space, power, and cooling, with facilities needing frequent updates every 10 to 15 years due to increasing power and space density Funding for these facilities often relies on indirect sources As research complexity escalates, so do the demands for computational resources, including processors, memory, and network connectivity, necessitating not only initial investment but also ongoing maintenance and timely upgrades based on performance and resource efficiency.
Software resources, including essential tools for managing computational and data operations, are crucial for researchers to optimize their work These tools encompass scheduling software, deployment and monitoring solutions, as well as support for parallel file systems Additionally, researchers benefit from mathematical libraries, parallel programming libraries, specialized applications, compilers, debuggers, and performance tuning tools It's important to consider the costs and trade-offs involved when choosing between commercial, open-source, public domain, or custom software options.
There is no such thing as a free lunch when it comes to software; both commercial and open-source solutions involve costs While commercial software incurs licensing and maintenance fees, open-source and custom applications demand a significant investment of staff time for development and upkeep Consequently, the overall expenses related to staff support, development efforts, and software maintenance must be thoroughly evaluated in relation to the institution's mission and budgetary constraints.
Additional Motivating Factors
Creating a sustainable budget model for local cyberinfrastructure facilities is driven by strategic motivations, emphasizing the importance of adequate funding to meet the current and future needs of local researchers Supporting local research is essential, as faculty and research staff often exhibit varying levels of advanced computing skills, with many users lacking experience and awareness of how advanced computing can enhance their research projects.
Experienced users, including faculty and researchers, often require straightforward resources such as high-throughput and small-to-large scale HPC clusters As disciplines increasingly demand sophisticated simulation tools, global collaborations, and access to expanding data sets, the capability of local resources will also need to grow This presents a significant opportunity to enhance the utilization of computational infrastructure (CI) resources and improve national competitiveness However, to succeed, these researchers need local support, as the effort required for faculty and graduate students to acquire the computational skills necessary to effectively use advanced CI facilities is considerable.
As institutions increasingly commit to computational science and interdisciplinary research through degrees and certifications, access to computational resources for training becomes essential Local institutional cyberinfrastructure (CI) facilities must be well-connected to regional and national resources, providing an "on ramp" for researchers needing greater capacity than what is available on campus Support for researchers unfamiliar with computational complexities is crucial, necessitating local staff and user-friendly interfaces for seamless resource access Moreover, the rise of energy-efficient computing and virtualization technologies promotes centralized resources, which are more efficient than distributed systems, especially when supported by a skilled staff Funding agencies favor sustainable models that demonstrate effective management of advanced CI resources, highlighting the importance of core skilled staff in securing federal funding opportunities.
Common Strategies, Models, Opportunities, and Challenges
Centralization of Resources and Services
Centralization was a common strategy that many of the workshop participants were working towards, in hopes of saving money by providing operational efficiencies and economies of scale and scope. o Benefits
Centralized data centers enhance operational efficiency by optimizing space, power, and cooling while facilitating long-term planning By reducing reliance on less efficient distributed facilities, organizations can lower operating costs and improve the quality of advanced computing through professional systems administration and maintenance Additionally, the sharing of facilities and resources is becoming a key component of many organizations' green initiatives, contributing to sustainability efforts.
A dedicated core staff that supports centralized resources is essential for institutions aiming to attract and retain faculty and CI staff with advanced skills in key areas like parallel computing, scientific applications, visualization, and data storage and analysis.
Economies of scale, scope, and cost sharing in a well-managed CI facility enhance research efficiency and value for funding Condominium clusters and enterprise storage solutions exemplify effective cost-sharing models, allowing diverse levels of participation with corresponding benefits Collaborative funding among researchers not only amplifies utility and bargaining power with Original Equipment Manufacturers (OEMs) and Independent Software Vendors (ISVs) but also fosters interdisciplinary research opportunities.
Enhanced research support through professionally managed CI resources allows faculty and researchers to concentrate on their research rather than managing computing infrastructure The availability of general-purpose computational resources enables them to leverage advanced computing for their research without the need for significant upfront investment in infrastructure and staff However, challenges remain in fully optimizing these resources for their research needs.
Funding a large-scale centralized data center poses significant challenges due to high costs per square foot and the difficulty in securing sponsorships for non-educational facilities While libraries are viewed as essential infrastructure, computational facilities are often seen as mere expenses Additionally, many administrators outside the computational infrastructure community fail to recognize that as knowledge becomes increasingly digitized, the demand for computational and data analysis resources—and their associated costs—will inevitably rise.
Access control in a computational infrastructure (CI) facility is crucial for researchers, requiring careful management of both physical and administrative access Balancing access requirements with the fundamental principles of maintaining a secure and stable production environment presents unique challenges.
Identifying the right faculty and researchers for feedback on sensitive topics like queuing policies and access priorities is crucial for effective strategic oversight and policy decisions Engaging key stakeholders in discussions about the necessary heterogeneous computing resources ensures that the institution's research community is well-served and that their needs are adequately addressed.
University Funding Models
Many institutions recognize the importance of cyberinfrastructure (CI) as a vital component of campus infrastructure, akin to administrative IT, libraries, and networking services Some fully fund CI through their core internal budgets or utilize indirect funding from research grants Additionally, a "Partner's Program" model has emerged, allowing faculties to enhance a centralized resource with base university funding, rather than investing in individual resources This collaborative approach typically involves institutional support combined with cost-sharing from researchers to bridge funding gaps.
Efficiency – Base funding for CI reduces individual department costs by eliminating the need to build and support their own resources and optimizes institutional CI operations and maintenance.
Institutional funding aims to create strategic advantages for faculty, researchers, and students by granting access to essential cyberinfrastructure resources This support enables exploration of new research areas, fostering innovation and breakthroughs that might otherwise remain unattainable Furthermore, graduate students benefit from gaining crucial experience in computational science, a field increasingly vital across various disciplines, including traditional sciences, social sciences, and humanities.
Sustainability – How will institutions develop a business model that enables them to sustain the staff, computational resources and services on an ongoing basis, especially during economic downturns?
Motivation – If resources and services are free to faculty researchers, is there adequate motivation for faculty to compete for grants that support their computational requirements at some level?
External Funding Models
Institutions primarily funded by external sources, like federal grants and industry partnerships, can offer large-scale resources and services that local funding models cannot support A prime example of this is the NSF-funded TeraGrid resource providers, which exemplify the advantages of such funding structures.
To enhance national competitiveness and support world-class research, it is essential to focus on efficiency in federal funding By investing in a limited number of well-managed centers staffed with highly skilled professionals and equipped with extreme scale resources, we can maximize the impact of our funding efforts.
Innovation thrives in computational centers that enhance scale and performance, leading to advancements in software, tools, and related technologies This progress benefits not only the research fields utilizing these resources but also significantly impacts computer science and the wider domain of computational science.
National competitiveness hinges on the success of industrial outreach and collaboration at nationally funded facilities The technologies developed through the quest for extreme scale and performance ultimately enhance industry capabilities, enabling companies to accelerate the development of innovative products and services However, challenges remain in achieving these goals.
During economic downturns, competition for funding becomes more intense due to limited federal and state support, including legislative line items Additionally, securing federal funds for institutional resources is increasingly challenging, as agencies like the NSF are primarily prioritizing large-scale projects.
Sustainability in institutions reliant on externally-funded projects is crucial for maintaining staff expertise beyond immediate funding timelines National-scale centers face ongoing pressure to enhance their resources for better performance A key challenge lies in securing national funding for timely hardware upgrades to remain competitive in a rapidly evolving landscape.
Cost Recovery Models
Research teams' capacity and readiness to invest in centralized CI computational resources or consulting services are crucial in evaluating the transition to a cost recovery model Institutions with higher external or internal funding per faculty member are generally more equipped to adopt these cost recovery strategies compared to those with limited research funding.
Researchers measure productivity through various metrics, including publication count, student mentorship, and the scientific impact of computational discoveries For those with limited funding, justifying direct payments to a central computational infrastructure (CI) facility can be challenging, potentially leading them to forgo computing-intensive research altogether However, if the costs for accessing centralized CI resources are sufficiently low compared to the productivity benefits gained, even modestly funded research groups may find value in selectively utilizing these services or emerging technologies like cloud computing.
Well-funded research teams often operate close to the optimal size for effective mentorship from leadership Instead of simply increasing team size, enhancing the resources available to current members—particularly continuous improvement (CI) resources—can significantly boost productivity This approach makes the value of service fees much more justifiable.
Implementing a cost recovery model in an institution means that access to resources and services is partially or fully funded by a fee-for-service, which must be transparent to highlight the value of a centralized service This transparency helps prevent faculty from creating their own isolated systems Additionally, the cost of utilizing a centralized resource should be less than or equal to the expense of faculty establishing their own resources, necessitating that centralized services remain cost-competitive with graduate student labor while offering superior service Achieving this pricing balance has significant implications for institutional support and subsidies.
There are benefits and challenges in implementing a cost recovery model. o Benefits
Steady-state funding is crucial for faculty researchers, as adequate research budgets enhance their satisfaction and willingness to invest in resources and services A CI facility's cost recovery model thrives on the support of well-funded researchers, leading to improved financial sustainability Institutional funding allows these facilities to adjust their offerings based on demand, ensuring they meet the needs of their users effectively Additionally, this model offers transparency, enabling institutions to assess the impact of their financial support and subsidies on research outcomes.
Positive incentives within a cost recovery model enhance the motivation of faculty and researchers to submit proposals and secure grants, which in turn helps cover the expenses associated with computational resources and services This approach not only benefits the financial standing of researchers but also positively influences the cyberinfrastructure facility and strengthens the institution's overall research portfolio.
Economies of Scale, Scope and Cost Sharing – By contributing research funds toward a well-run
CI facility resources and professional services offer researchers essential support, ensuring that the combined capabilities exceed individual contributions By providing access to skilled staff and advanced computational resources, researchers can effectively manage peak computing demands that would be financially unfeasible on their own However, challenges remain in maximizing these resources.
Cost recovery models for CI facilities rely on researcher demand and the ability to pay for resources and services Transitioning from a fully subsidized model to cost recovery can be challenging, requiring dedicated CI leadership to understand researchers' needs and their willingness to pay It's essential for the CI facility to establish a compelling value proposition for both the institution and its users to facilitate this change.
Concerns arise that a CI facility functioning solely in a service mode may lag in technological advancements, diminishing its value to researchers If the facility fails to evolve, this concern is valid Conversely, a CI facility operating on a cost recovery model is incentivized to meet the demands of researchers, as its survival depends on providing relevant resources and services This motivation ensures the facility remains valuable and avoids obsolescence.
Staffing and Succession Planning
The staffing levels and diversity at a research center are influenced by the types of services offered and the number of computational researchers, both experienced and novice Each technology, such as clusters, storage systems, and security infrastructure, requires specialized expertise Smaller centers may struggle to acquire the necessary skills for all technologies, prompting them to limit their technology and vendor choices to enhance manageability To reduce staffing needs while still utilizing complex high-performance computing (HPC) technologies, centers can opt for commercial products that come with support, including services for cluster installation, storage solutions, and scheduling software.
To ensure 24-hour uptime and highly available services, a center must maintain a sufficiently large on-call staff Without adequate funding for this level of support, users should be aware that significant failures or issues occurring during late nights or weekends may not be addressed until the next business day.
Staff roles can be divided into two main categories: "inward facing," which focuses on managing the center's systems and resources, and "outward facing," which is dedicated to user support and advanced user analysis services Many staff members possess a diverse skill set that enables them to contribute to both types of activities, effectively balancing their efforts between inward and outward facing responsibilities.
The transition of leadership in HPC centers can be disruptive, particularly in smaller facilities where the director's personality significantly influences the center's vision and persistence To mitigate the impact of a director's departure, it is essential to engage staff in operational decisions before the transition, ensure university administration recognizes the center's ongoing mission through regular communication, and establish a sustainable funding model Additionally, fostering "hero" users within the institution can advocate for the center's continuity, while developing faculty expertise in CI technologies and proposal writing can strengthen operations Finally, providing recommendations for a successor can help ensure a smooth transition.
Changes in senior university administration can significantly impact Continuous Improvement (CI) initiatives, as many facility directors rely on support from key officials who view CI as strategic When these positions change, there is no assurance that new leaders will share the same vision for CI Directors must educate university administration on the importance of CI and provide regular updates with success metrics that align with the institution's mission This ensures that CI becomes a fundamental aspect of the university rather than just the focus of individual administrators Additionally, community support for materials that showcase the return on investment (ROI), cost avoidance, and cost savings is essential, with organizations like NSF and CASC playing a crucial role in this effort.
Industry and Vendor Relations
Advanced computing services that leverage optimal performance and economies of scale enhance the relevance of academic CI resource providers to industry and vendors The rapid development and improvement of computing technologies each year make computational science both exciting and challenging Researchers are eager to utilize new technologies for better and faster results, yet they must weigh the investment of time and effort against potential benefits This dynamic compels cyberinfrastructure service providers to stay updated on emerging technologies and swiftly adopt those with promise, despite the time-consuming process of developing prototype test systems, software, and tools.
The profit-driven nature of industry emphasizes the importance of a sustainable recovery model, which fosters careful decision-making in evaluating and implementing appropriate technologies In academia, researchers often inflate their needs until financial contributions are required, at which point they tend to provide more accurate assessments of their computational infrastructure (CI) requirements When researchers have a financial stake in the project, they are more inclined to express realistic needs rather than simply requesting the largest available resources.
Cyberinfrastructure providers present valuable partnership opportunities for industries, thanks to their willingness to experiment and their experienced staff While companies, especially start-ups, are eager to explore new technologies, they often lack the time and resources that academic institutions can dedicate By forming industry partnerships and engaging in technology transfer or licensing agreements, academic providers can capitalize on their intellectual assets to generate additional revenue and remain at the forefront of technological advancements However, it's important to note that not all institutions are equipped to accept corporate funding or gifts due to their primary funding sources, making this opportunity selective.
OEM and ISV partnerships are crucial for a sustainable Continuous Improvement (CI) model, as they provide CI leaders with vital technology roadmaps for strategic planning These relationships also grant early access to emerging hardware and software for testing and evaluation As partnerships develop, vendors adapt their products to meet evolving research needs and enhance funding opportunities Additionally, strategic collaborations often incentivize vendors to offer competitive pricing to academic partners, aiding them in securing grants.
Metrics of Success and Return on Investment (ROI)
Amid current budget constraints, academic institutions must justify technology and staffing expenses, including those related to academic CI facilities IT services are frequently among the first areas targeted for budget cuts To ensure ongoing support for institutional CI, CI directors must establish clear success metrics and consistently communicate the return on investment (ROI) associated with these initiatives.
Participants in the workshop distinguished between quantitative and qualitative metrics of success, emphasizing that the definition of success varies based on the target audience for the metrics Quantitative metrics consist of measurable data with straightforward collection methods, such as system accounting data, consulting logs, and metrics from tools like the University at Buffalo’s “Metrics on Demand.” In contrast, qualitative metrics, while often compelling, are more challenging to support with statistical evidence, with examples including customer satisfaction testimonials and committee reviews.
Workshop participants expressed an interest in developing more compelling quantitative metrics and accounting methods This is a “New Challenge Area” that needs additional attention, discussion and community collaboration.
Quantitative Metrics of Success
Workshop participants identified key quantitative success metrics for academic Cyberinfrastructure (CI) facilities, including Service Metrics that track user accounts, departmental support, and resource utilization, typically measured annually to demonstrate growth "Science Driver" Metrics illustrate the facility's impact on research, highlighting presentations, publications, and multidisciplinary collaboration, alongside educational offerings like courses and workshops Funding Metrics quantify grants and awards linked to CI services, encompassing cost recovery funds, successful grant proposals, and job creation Intellectual Property Metrics reflect the number of patents, copyrights, and industry partnerships fostered by CI access Lastly, Outreach Metrics assess efforts to engage underrepresented groups, emphasizing initiatives like NSF Research Experiences for Undergraduates (REUs) and virtual training programs, which enhance project impact and credibility.
Qualitative Metrics of Success
Workshop participants highlighted key qualitative success metrics, including Economic Development, which emphasizes the role of local, regional, or national resources in enhancing industry competitiveness through access to essential services As research computing gains traction in commercial sectors, demonstrating return on investment (ROI) becomes increasingly challenging Additionally, Researcher Satisfaction is notable, with many researchers expressing that access to CI facilities has significantly boosted their productivity, allowing them to focus more on research and resulting in increased publications While this enthusiasm is vital for securing ongoing institutional support, quantifying it in terms of cost savings remains complex Lastly, Strategic Metrics are essential for illustrating a cyberinfrastructure facility's significance to its institutions, encompassing factors such as faculty and student recruitment, integration with resources like TeraGrid and Open Science Grid, and collaboration on major national cyberinfrastructure initiatives.
New Challenges
Workshop participants emphasized the need for improved data collection methods to develop meaningful success metrics, particularly in two key areas First, cost savings and cost avoidance metrics are crucial for demonstrating the financial benefits of cyberinfrastructure facilities, such as centralized data centers, which can significantly reduce expenses compared to distributed private research clusters However, quantifying these savings remains challenging Second, institutions must recognize the importance of cyberinfrastructure as essential core infrastructure, akin to libraries and specialized research facilities Effective communication of the value of computation and data management to administrators—many of whom lack a computational background—is vital for securing necessary funding and support for these critical resources, ensuring their role in advancing research and education is acknowledged.
Conclusions
This report outlines various strategies and experiences shared by participants of the NSF workshop focused on sustainable funding and business models for academic cyberinfrastructure facilities nationwide It highlights valuable lessons from the report and over twenty position and white papers included in the appendix Rather than endorsing a specific funding model, the report serves as a resource for institutions reassessing their funding approaches or establishing new CI facilities The collaborative spirit of the 100-plus workshop attendees is expected to foster ongoing dialogue and knowledge sharing to enhance computational science across institutions For more information, the Sustainable Research Computing Centers (SRCC) website at http://www.cac.cornell.edu/SRCC offers access to the report, workshop findings, presentations, position papers, and a LinkedIn group for continued discussions.
The workshop highlighted the importance of broadening participation in computational science, emphasizing that its health is vital for national competitiveness The Branscomb Pyramid model illustrates the need for a strong foundation to cultivate the next generation of researchers and a skilled workforce Participants identified increased geographic participation and adequate training as key directions for enhancing national cyberinfrastructure Additionally, the workshop addressed the sustainability of computational science as a crucial element of academic research, calling for collaboration among national funding agencies, educational institutions, and industry Each institution must develop a tailored business model aligned with its mission, supported by strong institutional commitment and diverse funding sources Continued collaboration among organizations like the Coalition for Academic Scientific is essential for fostering growth and innovation in this field.
CASC fosters ongoing discussions and sharing within the community that began during the workshop By supporting computational science across all tiers of US academic institutions, we can enhance collaboration, drive innovation, and improve our global competitiveness, leading to new economic growth opportunities.
outlines strategies for enhancing science and technology education, fostering innovation, and ensuring a competitive edge in the global economy The findings highlight the critical role of investment in research and development to address challenges and seize opportunities for growth.
[2] Members of the 2005 “Rising Above the Gathering Storm” Committee; Prepared by the Presidents of the National Academy of Sciences, National Academy of Engineering, and Institute of Medicine (2010)
Rising above the gathering storm, revisited: rapidly approaching category 5 Retrieved from http://www.nap.edu/catalog.php?record_id999.
The University of California's shared research computing services pilot program, as detailed in a position paper by Lim et al (2010), explores collaborative computing solutions to enhance research capabilities across institutions This initiative aims to optimize resources, improve access to computing power, and foster innovation in research methodologies The paper emphasizes the importance of shared services in addressing the growing computational demands of research projects and highlights the potential benefits of collaboration among universities For more information, the full position paper is available at the provided link.
[4] Joiner, David (2010) The New Jersey Institute for Modeling and Visualization Position paper Retrieved from http://www.cac.cornell.edu/~lifka/Downloads/SRCC/9.joiner_position_paper.pdf
The Massachusetts Green High-Performance Computing Center (GHPCC) is a state-of-the-art academic research facility designed to enhance computing capabilities while prioritizing sustainability This position paper, authored by Rick Adrion, Ken Blank, Chris Hill, Jim Kurose, and Andrei Ruckenstein in 2010, outlines the facility's commitment to green technology and high performance in research computing For more detailed information, the full document can be accessed at the provided link.
The Southeastern Universities Research Association's position paper by Crane, Robinson, and Smith (2010) discusses the importance of enabling and sustaining cyberinfrastructure between campuses The paper emphasizes the need for collaborative efforts to enhance technological capabilities and support research initiatives across educational institutions For further insights, the full document can be accessed at the provided link.
The paper by McMullen and Monaco (2010) discusses the significance of regional cyberinfrastructure in supporting high-performance computing centers on campuses It emphasizes the need for robust infrastructure to enhance computational capabilities and facilitate research The authors highlight how effective regional collaboration can lead to sustainable advancements in high-performance computing, ultimately benefiting academic institutions and their research initiatives For further insights, the full paper can be accessed at the provided link.
[8] Sills, Eric (2010) North Carolina State University model for providing campus high performance computing services Position paper Retrieved from http://www.cac.cornell.edu/~lifka/Downloads/SRCC/15.NCStateHPCModel.pdf
The 2005 report to the President on Computational Science emphasizes the necessity of enhancing America’s competitiveness through advancements in information technology and computational research It highlights the critical role of federal coordination in fostering innovation and collaboration among various sectors The document underscores the importance of investing in research and development to maintain the United States' leadership in technology and science For further details, the report can be accessed at the National Coordination Office for Information Technology Research and Development's website.
The article by Thomas Furlani discusses metrics of success and return on investment, emphasizing the importance of evaluating performance in computational research It highlights that understanding these metrics is crucial for assessing the effectiveness of projects and ensuring sustainable growth The paper, published by the University of Buffalo Center for Computational Research, serves as a valuable resource for researchers looking to improve their investment strategies and measure their success effectively.
[11] Cornell University Center for Advanced Computing Education and Outreach (2010) Retrieved from http://www.cac.cornell.edu/education/train.aspx
National Science Foundation Workshop on Sustainable Funding and Business Models for High Performance Computing Centers
The NSF-sponsored Workshop on Sustainable Funding and Business Models for High Performance Computing Centers is now accepting applications for registration and position papers Interested participants can register by visiting https://mw1.osc.edu/srcc/index.php/Main_Page and following the provided links.
The workshop aims to facilitate an open dialogue among High Performance Computing (HPC) center directors, campus information officers, and research officers regarding sustainable funding and effective business models for research computing centers This collaborative discussion will foster a collective understanding of various organizational, funding, management, and training models that contribute to the long-term financial sustainability of these centers.
The workshop will equip participants to effectively advocate for the establishment of research computing centers, providing them with essential data to support their arguments for sustainability It aims to assist higher education institutions in economically disadvantaged areas by offering successful models for research computing centers that can significantly enhance local economies By fostering the sharing of knowledge across diverse centers, the workshop will promote the creation of similar initiatives and enrich broader learning communities Ultimately, it will ensure that all science and engineering students gain early exposure to advanced computational concepts, thereby strengthening the future workforce.
Seventy-five invited leaders in sustainable funding for HPC centers will attend the event in person, while additional participants will join via WebEx conferencing The meeting aims for broad engagement from the research computing community to ensure diverse stakeholder representation and meaningful participation throughout the event.
The academic research computing community is encouraged to submit position papers, which will fulfill two key objectives: gathering input from the broader community and selecting individuals for on-site workshop participation Each position paper is limited to three pages and must be submitted by March 15, 2010 A review panel will evaluate the submissions to determine the invitees for on-site participation.