The Role of Regional Organizations in Improving Access to the National 1 Computational Infrastructure A REPORT TO THE NATIONAL SCIENCE FOUNDATION



Gregory E. Monaco2 Donald F. (Rick) McMullen Gwendolyn Huntoon Jennifer Leasure David Swanson Henry Neeman Joni Blake Kate Adams FINAL REPORT3 2016

1 This work was funded by National Science Foundation award #1543655 to Kansas State University. 2 Gregory E. Monaco, Ph.D. Department of Psychological Sciences, Kansas State University, Manhattan, KS, and Great Plains Network ([email protected]). 3 This replaces the Draft Report from March 2016.

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

Table of Contents Executive Summary .................................................................................................................................. 3 Introduction ................................................................................................................................................. 4 Project Goals and Approach .................................................................................................................. 5 White Papers ............................................................................................................................................... 6 Submissions ............................................................................................................................................................ 6 Significant Observations by White Paper Authors ................................................................................. 7 Conference .................................................................................................................................................. 14 Attendance ........................................................................................................................................................... 15 Significant Contributions from the Conference .................................................................................... 15 Recommendations .................................................................................................................................. 20 Core Recommendations .................................................................................................................................. 20 Recommendations for Actions Within Specific Communities ....................................................... 22 Conclusions ................................................................................................................................................ 25 Appendix I: Members of the Conference Organizing Committee ...................................... 26 Appendix II: White Paper Submission Guidelines .................................................................... 27 Appendix III: List of Submitted White Papers ........................................................................... 29 Appendix IV: Affiliations of Authors Submitting White Papers ......................................... 31 Appendix V: Conference Agenda ...................................................................................................... 33 Appendix VI: Conference Participants .......................................................................................... 34



2

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

Executive Summary

The Role of Regional Organizations in Improving Access to the National Computational Infrastructure conference was held in Kansas City, Missouri, on October 22 and 23, 2015. A total of 36 white papers were submitted in advance of the meeting, and 39 individuals were in attendance. The majority of the participants were from academic institutions, and many also represented a state, regional or national organization with significant interest in improving access to the national computational infrastructure. The expressed conference goal was to develop a set of recommendations to the National Science Foundation and to the vested community, at large, to assist regional organizations to leverage their work for the benefit of the research community. The implicit goal was to understand what actions, if any, are needed to achieve a radical shift across a diverse set of organizations that will result in improved coordination of, access to and utilization of the national computational infrastructure.

Organization of This Report

Contributions to the discussion by white paper authors and conference attendees are grouped into key thematic areas and presented. Two sets of recommendations are made, core recommendations and recommendations for actions within specific communities.

Recommendations Core recommendations are a set of recommendations that cut across communities, agencies and projects and are consistent with a vision to effect cultural change and make computational and data resources pervasive, connected, shared, distributed and dynamic. Community recommendations are directed towards specific slices of the research and education community, including funding agencies. These recommendations include 1) the adoption of specific practices among existing and emerging communities of expertise that will result in improved coordination, crosscommunication and expanded use of all available resources and 2) the identification of priorities for funding agencies that are likely to influence future dynamics for cyberinfrastructure uptake. There are many cyberinfrastructure based and related organizations from campus to national that enable the use of advanced cyberinfrastructure in research. Now is an excellent time to harness their collective energies and focus them on moving computational and data-intensive science in the United States from an early Internet level to a more ubiquitous level. By carefully considering and choosing to implement the recommendations from this report, it may be possible to accomplish broad scale change across the cyberinfrastructure landscape in support of future computational and dataintensive science in the United States and beyond.

3

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

Introduction

The activities associated with providing the United States’ scientific and research communities with access to the cyberinfrastructure tools that meet their computation and data needs are made through the coordination of resources across multiple levels: local (e.g., campus), regional, national and international. This coordination involves the joint activity of a large number of people across many organizations (Figure 1).

Figure 1: Configuration of Advanced Cyberinfrastructure Resources in the United States. Regional CI includes virtual communities that are neither local nor national.



While useful for enabling scientific discoveries, there are inherent limitations to the current configuration of advanced cyberinfrastructure resources: • Centers of expertise tend to become islands of expertise. As organizations become more mature, they also become more adept at their core business and more specialized. • Communication across islands is limited. Specialization often leads to personnel communicating most frequently with personnel at other organizations that provide similar or closely related services. • Lines of communication between users and providers that were once necessary to establishing advanced services go unused, become less efficient and disappear. State and regional networks, for example, and their operating organizations were created largely to connect researchers on campuses to nationally funded, shared research computing infrastructure. With time network personnel communicate less directly with researchers and more directly with campus level network personnel.



4

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

• •

•

The middle, regional layer, although essential, is often neglected in the bigger picture of national CI. New projects arise not only to deliver new services, but also as workarounds to limitations to the status quo. Efforts such as XSEDE campus bridging, Advance Cyberinfrastructure - Research and Education Facilitators (ACI-REF) and Open Science Grid (OSG) are aimed at creating a more consistent campus-to-national environment. There is no way of knowing whether potential users are being adequately served. There are thousands of scientists at smaller academic institutions who are trained to do computational and data-intensive science, who would like to and are qualified to train their students in these skills, but who do not have access to the appropriate technology.

The purpose of this project was to gain input from relevant organizations and to develop a set of recommendations to reinvigorate the state of advanced cyberinfrastructure and to lay the groundwork for a vibrant, healthy national computational cyberinfrastructure that brings together all the relevant players and is flexible enough to accommodate new developments.

Project Goals and Approach Because of commonalities of purpose and mission among campuses, regional research and education networking organizations, and national computational facilities, as described above, we proposed to facilitate a discussion among representatives from these seemingly diverse organizations so that they might learn from one another and better coordinate their efforts. The ultimate goal was to initiate a process for charting a more coherent course of action that will contribute to a stronger computational and cyberinfrastructure landscape by • Clarifying the roles of campus, regional, and national resources and services in the overall computational ecosystem; • Identifying scalable operating models for effective and dynamic delivery of computational resources and services to geographically distributed researchers; • Identifying collaboration models that reduce barriers across different environments while preserving efficiency and innovation; • Creating effective learning and workforce development programs to ensure next generation capability in the most advanced computational science for all disciplines. We began by assembling a Conference Organizing Committee (COC) comprised of knowledgeable representatives with interests in the project goals and ties to multiple levels of the CI and computational landscape. The members of the COC are listed in Appendix I.



5

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

To start a discussion on the topics at hand we used the following approach: Step 1: Solicit white papers from interested parties and organizations. Step 2: Review the white papers and invite authors to the conference. Step 3: Convene for a face-to-face meeting in October, 2015, in Kansas City, MO, to review white papers, discuss significant topics and draft recommendations.

White Papers

We believed that soliciting white papers on the topics of interest prior to the meeting would prime the community to think about the project goals in advance of the meeting and would help to assure adequate representation from many sectors of the cyberinfrastructure community. The call for submission of white papers is in Appendix II. Members of the Conference Organizing Committee circulated it via email lists to a broad community of interest. Since we also wanted to hear from representatives of small colleges that are not affiliated with mainstream CI groups, but may have faculty with interest in computational research and/or education, we requested names of individuals who might be able to contribute to the discussion and subsequently requested papers directly from those individuals. Authors of white papers were asked to address one or more of the following questions: A. What should be the role of regional organizations in providing broader support to computational researchers? B. How can these organizations expand researcher engagement to underrepresented communities? C. How can these organizations sustain timely and relevant education and outreach efforts to computational researchers over the long term? D. How can efforts across these organizations be effectively coordinated? E. What is an effective collaboration environment and structure? F. How can these organizations promote and coordinate sharing of hardware, software and expertise across campus, state, and regional boundaries?

Submissions Papers were submitted from universities, both small and large, scientific research collaborations, regional organizations, both state and multi-state, and national organizations. In all, 36 white papers were received representing at least 98 authors and 80 organizations (Appendix IV). In some cases, the authors were listed as a community (e.g., Champions Leadership Team, OneOklahoma Cyberinfrastructure



6

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

Initiative Membership). In many cases an author represented more than one organization. The perspectives shared in white papers on the problems posed were diverse. In addition to campus, regional and national cyberinfrastructure perspectives, authors also presented perspectives that represent the needs of other broad groups: 1) individuals who do not have a campus affiliation, 2) established research communities and collaborations, 3) educators and 4) technical developers. Submitted white papers are listed in Appendix III. The white papers are publicly available for download at http://tinyurl.com/pd99qg7.

Significant Observations by White Paper Authors White paper authors put a good deal of thought into the major themes for the conference. An overview of these contributions is organized by thematic question area. Direct quotations and paraphrases from the white papers are included to highlight significant contributions. Direct quotations are italicized, whereas paraphrases are not. In both cases white paper authorship is cited. A. What should be the role of regional organizations in providing broader support to computational researchers? A-1 Regional organizations occupy a key strategic position between campuses and national computational resources. These organizations have a significant role to play in assisting potential users of computational resources at the local level to learn about and use national resources. (McMullen et al.; Von Oehsen & Hauser; Wang) • Regional organizations are geographically closer to colleges and universities, are trusted within the region, and are more easily accessible to the community of users that they serve. (Von Oehsen & Hauser) • A regional organization potentially offers economies of scale - running one larger regional system can be managed more efficiently at one place with user support distributed over the partners of the regional system. (Von Oehsen & Hauser) • The national High Performance Computing (HPC) centers…conduct excellent community engagement, but…can’t visit every campus, attend every conference and workshop, or reach out to the many people that would most benefit from knowledge about and access to advanced digital resources. (Lathrop & Champions Leadership Team) A-2 State and regional networks occupy a unique role in the national cyberinfrastructure due to their stability, the value that they provide member campuses, and their potential to accelerate utilization of the national computational infrastructure. Monaco et al. further note that

7

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

• • •

•

The majority are self-funded by user communities investing in the infrastructure, resources, and services that regional networks provide. They also foster and enable collaborations at a number of levels, across institutional, disciplinary, state and regional boundaries. [M]any…provide a wide range of other services to their members, including data centers, remote storage, identity management, high performance computing, and consulting. Consulting services they provide include analysis, design, configuration, and troubleshooting. They are trusted by their members and are natural conveners across their geographic expanse.

A-3 A number of state and regional HPC centers have demonstrated leadership and successful models for providing at least state-wide support and in some cases regional support. A sampling of notable examples include Ohio Supercomputer Center’s statewide efforts, the OneOklahoma Cyberinfrastructure Initiative’s impact in their state (including access to HPC resources and an annual Symposium), the Great Plains Network efforts to serve their region, efforts by ACI-REF members to share information and their on-going efforts to develop models for resource sharing, and the Regional Champions (e.g. Harvard University) efforts to reach out to Campus Champions within their geographic region. (Lathrop & Champions Leadership Team) A-4 New approaches, models and opportunities for regional collaboration are emerging in California (Smarr et al.), Oklahoma (OneOklahoma Cyberinfrastructure Initiative Membership), Oregon (Barber et al.). The XSEDE Campus Champions project has added a regional component (Campus Champion Leadership Team). Strande points out the potential to enable regional partnerships at the university system level. B. How can these organizations expand researcher engagement to under-represented communities? B-1 Communities that may be considered to be under-represented with respect to knowledge of, expertise in, and access to computational and data resources include, but are not limited to, • Students and faculty at small, under-resourced colleges (Blaustein; McMullen et al.; Rampp). • Researchers who are unaware of how to use computational tools even when they are in a well-resourced environment (Jha et al.). • Researchers in areas of science where breakthroughs in methodology and/or technology lead to increased need for computational analysis. The field of transcriptomics has recently been given a huge boost from the use of NGS-high throughput sequencing (RNAseq) technologies to sequence RNA samples. From 2005 to 2014, the capacity of a single sequencing run increases up to an astounding 1000x. Consequently, the demand of computing resources for downstream data analysis also increases exponentially. (Wang & Siegfried)



8

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

Individual citizen-scientists who might make use of computational resources given training and access (G.D. Brown). • Women and minority scholars (Leake). There is a substantial reservoir of potential future computational scientists who are currently enrolled in small colleges throughout the United States with professors who wish to introduce them to high performance computing, but lack the resources to do so. (Blaustein) •

B-2 Engaging these communities more effectively promises to lead to substantial increases in the demand for computational resources. This issue has implications for both provisioning for future demands for computational resources as well as for the future health and well being of United States science. (Blaustein; Wang & Siegfried; Jha et al.; Leake; McMullen et al.) B-3 However, as Lance points out, engaging these researchers is a challenge. Traditional expectations of researchers in under-represented communities and traditional approaches to trying to engage researchers in these communities are likely to fail. • … the route to obtaining access to outside HPC resources is not obvious. (Blaustein) • …researchers are often unaware of how the computational research world is evolving... (Rampp) • Despite data- and compute-intensive scientific needs, the polar science community does not yet take full advantage of the High-Performance and Distributed Cyberinfrastructure (HPDC) resources available to it, in part because the polar science community lacks awareness of these resources and in part because it lacks the training required to use them. (Jha et al.) • Small colleges do not naturally feel like equal players even when a shared resource lands within our neighborhood, for example at an R1 in our state. (Rampp) B-4 National computational resources (e.g., XSEDE) may not be adequate nor the most appropriate to satisfy both existing and emerging needs. University resources and regional resources can satisfy the expanding needs of many researchers and must be considered a significant provider of computational cycles. (Blaustein; Swanson) B-5 Regional organizations, particularly regional networks, are well-positioned in terms of policy, history of collaboration, understanding of the technical requirements of modern computational research, and contacts with the research



9

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

community, including research centers, across the region, to assist underrepresented groups. (DuRousseau; McMullen et al.; Monaco et al.; Rampp) B-6 Rotman and Zurawski observe that ESnet’s Science Engagement Team has observed the following challenges to researcher engagement: • Lack of networking expertise within the staff of a collaboration or facility - including user services groups, systems administrators, scientists, researchers, and administrators • Pervasive local performance issues, coupled with lack of resources to fully address them • Negative perceptions about data transfer (“too complicated”, “I tried it and it didn’t work”) • Cultural resistance to change (“we’ve always shipped disks, and this meets our needs”) • Fear among scientists of spending more time solving IT, issues than on their research C. How can these organizations sustain timely and relevant education and outreach efforts? C-1 Timely education and outreach efforts are increasingly important for several reasons: (1) The potentially dramatic growth of the user base, (2) rapid change associated with improvements in advanced CI and computational technology, (3) turnover in trained and effective support staff who are also highly sought after in business and industry. (Engel; Blaustein; McMullen et al.) Given the fast-paced changes of available tools and technologies, the connection between teaching and research is crucial. (Engel) C-2 Additionally, Jolly and colleagues in their studies of factors that successfully influence adoption of advanced CI, have found that trained, knowledgeable and available staff are key to boosting CI adoption. This finding is echoed by Hill. Several participants in the study expressed that the ability to request personal assistance from CI staff is very useful and boosts CI adoption. Personal assistance in this study means help from support staff in answering any questions regarding the content and technical support in using CI. Participants also expressed that a quick turnaround time for computational experiments is the key to their success; speed of results generation increases when they use CI. The initial data highlights that CI is a valuable tool in solving complex problems and increases user productivity. Study participants mentioned that their motivation to use CI came from their advisors, professors and collaborators, and such support seems to play an important role in CI adoption. These factors make it compelling for participants to use CI. (Jolly et al.)

10

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

C-3 There are groups with significant training expertise and other groups (regional) that are well positioned to facilitate training and education. • Information exchange between LCI [Linux Clusters Institute] and Regional Organizations represents substantial potential for dissemination of information, recruitment, and development of resources. Further interactions between regional organization membership and the LCI Committee can help shape the instructional content of LCI curriculum to meet these organizations’ needs, provide for the HPC workforce, and inform campus research support staff. (Bright et al.) • UMBC offers coursework in high performance computing through a variety of means, including undergraduate coursework, graduate coursework, a large and formal summer REU program and individual student research experiences. Particularly for the coursework, it may be possible to leverage regional partnerships to better share the work we are already doing and to perhaps use the broader audience to draw in additional partners who are closer to the individual tools (e.g. the tool providers themselves – XSEDE, OSG, Amazon, D-Wave, etc.). (Engel) C-4 Bright et al. and Jennewein point out that regional organizations are well positioned to accelerate education and outreach • By partnering with existing training organizations (e.g., Linux Cluster Institute) to schedule training across the region, • By providing 1) distance learning opportunities for researchers and IT personnel, 2) regional meetings to foster collaboration and provide access to national expertise, and 3) aggregated demand for collaboration and infrastructure regional organizations can increase CI literacy and availability at USD and throughout the region. (Jennewein) C-5 Hampton has created four categories of users, based on hundreds of encounters, with methods for satisfying associated needs of these users: • The Freshman: This user is typically new to computing using a shared resource, and may have never even used a remote computer. Once we determine the problem is relevant for our systems, the focus is on initial training. Handouts of common commands, pointers to simple examples, and practicing short job submissions is usually enough to give them confidence to start their own tests. • The Inquirer: This user knows what they do not know and often seeks us out, eager to learn. A typical case here is when someone comes in with a calculation that takes days or even weeks on a local resource. We spend some face-to-face time analyzing their simulations and may even introduce them to parallel programming or other techniques for speeding up their results. • The DIYer: This user usually has a technical background and wants to only run on and administer their own machines. We often begin with a discussion promoting the benefits of letting us manage their systems while they have more



11

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

•

time to focus on their work. In many cases, we also find that they have been resource limited. The Expert: This user is happily submitting jobs and using shared resources. The opportunity to work with them arises when we analyze their work and note that the submissions are not as efficient as they could be. We will sit down and explain debuggers and profilers and show them how to eke out even more performance.

D. How can efforts across these organizations be effectively coordinated? D-1 Cyberinfrastructure-focused organizations, from campus to regional to national level, appear to recognize the benefits that are likely to accrue from closer coordination of efforts. (Bottum & Atkins; McMullen et al.; Monaco et al.) D-2 As Blake notes, coordination is not easy: The challenge lies in developing a comprehensive strategy to coordinate all of the expertise and resources and to develop best practices, consistent identity management protocols, and standards for data formatting and sharing. D-3 Brown et al. suggest that, [f]rom the standpoint of campus CI staff, one of the best things that can be done to aid the coordination and effectiveness of staff efforts at the local and regional levels is to automate those things that can be automated – and the adoption of common standards at the regional or national level makes it possible to automate a larger fraction of the total work to be done (leaving valuable human time to deal with the specific concerns and needs of local researchers). E. What is an effective collaboration environment and structure? E-1 When considering the topic of an effective collaboration environment, the intent was to consider this with respect to the myriad groups that might work together to improve access to the national computational infrastructure. However, there is a second concern that has to do with an effective environment for potential end users who would like to engage with these resources. E-2 Hampton points out that, in his work at Notre Dame, [w]e provide a mixture of static references including wikis, how-to guides, and FAQs. In order to keep these up-to-date, users are encouraged to modify and add new information. We also promote face-to-face events such as weekly group training, individual sessions, as well as special training events. F. How can these organizations promote and coordinate sharing of hardware, software and expertise across campus, state, and regional boundaries? F-1 Effective economies of effort and scale can be accomplished through incentivizing strong partnerships among organizations with an incentive to share. (Bottum & Atkins; McMullen et al.; Monaco et al.)



12

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

•

•

•

Regional organizations, working in conjunction with campuses and projects at the national scale, have the opportunity to be a catalyst for promoting shared expertise and activities across research. (Bottum & Atkins) Campus level research computing facilities cannot compete with the scale of national resources, but they are an integral part of taking research from an idea to running a computation on 1 or 100,000 cores…Even if the research computation does not reach national scale resource requirements the local aspect is important (flexibility and personalized attention) to ensure researcher success and productivity…collaborating at the campus and regional levels is an important aspect in building the services that support both individuals and national scale computational problems. (Middlekoop) With a more dedicated effort, a regional network organization could foster sharing and leveling of CI and domain computing expertise across its members, and build knowledge-sharing organizations with member participation. (McMullen et al.)

F-2 Building a culture of sharing must be deliberate. As Livny points out, it takes time, requires a sustained commitment, must involve engaged users and depends on technologies that can translate the power of sharing into impact on science…Like any other social process, a network of shared resources is built one node at the time when a satisfied user convinced a reluctant owner to join. Friends bring friends and success stories help generate motivation and secure funding. Engaged users with reallife applications are invaluable as they guide research and development efforts, provide a laboratory for experimentation and evaluation of technologies that facilitate sharing and generate enthusiasm throughout the campus. F-3 Standardization of resources is likely to be a key to effectively sharing computing and data resources. (Brown et al.; Livny; Swanson) • The gaps between local and regional or national CI and the education and training needed are lessened by the adoption of a more uniform infrastructure that serves these multiple scales. This CI should enable users to learn a single interface that may be used on a laptop, on a local cluster, and at regional and national resources. (Swanson) • Standardizing resources at a regional level as much as is practical—that is, resources that conform sufficiently to a particular set of standards that are clearly documented and consistent—is one strategy for making them more accessible to researchers on the local level. These standards should focus on providing the same packages and configurations for the same purposes across the network, while the choice of standard packages should reflect common usage among the research populations those packages are intended to support. Standardizing has other benefits in addition to user accessibility: it allows the organization to streamline its effort to keep packages up to date and makes it easier to support a user regardless of what resource they are utilizing.



13

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

Another strategy to support researchers—one that depends on some level of regional standardization—is to make configuration tools and packages available that allow local system administrators to build or modify a local resource so that it looks and behaves in a way that is consistent with a regional resource if used for a particular task. (Brown et al.) F-4 Wurthwein cautions that sharing must not be imposed from the outside: Local as well as regional organizations need to be able to make autonomous decisions. They opt in based on their abilities and constraints. Not all institutions will contribute equally. And the driving motivation must not be bartering but rather donating to the extent of their abilities and comfort zone to achieve a shared goal. Any attempt of establishing a barter economy will make it impossible to serve communities at all scales. This is true for access to human as well as physical resources.

Conference

White paper authors or their representatives were invited to a conference in Kansas City on October 22 and 23, 2015. The conference agenda is presented in Appendix V. The overall format for the conference was (1) an introduction, (2) panel presentations, (3) working session discussion groups, (4) end of day and end of conference wrap-ups. Panel sessions corresponded to the conference themes: A. Role of regional organizations in providing broader support to computational researchers (State & Multi-state) B. Role of regional organizations in expanding researcher engagement to underrepresented communities C. Role of regional organizations in sustaining timely and relevant Education & Outreach D. Coordination of Efforts across organizations E. Effective collaboration environments & structure F. Role of regional organizations in promoting and coordinating sharing of resources across campus, state, and regional boundaries Panels were lead by members of the conference organizing committee who summarized the major points of the white paper contributions toward the theme. Panel members represented submitted white papers that presented challenging issues related to the theme. Following the panel presentations on Day 1, conference participants were invited to further discuss issues and likely solutions in breakout sessions corresponding to the conference themes. These breakout sessions were lead by the corresponding panel leader. Collective discussions to bring the new findings to the attention of entire group were held before adjourning the conference on Days 1 and 2.



14

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

Attendance There were 39 participants. In addition to the participants who represented white papers, there were additional participants from small colleges in Nebraska who provided additional representation from hard to reach communities.

Significant Contributions from the Conference Panel Discussion Summary What organizations may be productively considered regional in nature to further advance utilization of the computational infrastructure? • Research and Education Networks, state and multi-state • Statewide Collaborations • State High Performance Computing Centers • Projects specifically targeting uptake of advanced computational research • Campus High Performance Computing Centers that serve a broader set of campuses • Regional Big Data Hubs • Regional and State Academic Library Associations • Specific research projects that span multiple campuses What constitutes the National Computational Infrastructure? • XSEDE as both a collection of computational resources and technical support for using them (centers, campus champions and training/documentation) • Regional Computational Entities • Campus High Performance/High Throughput computational centers • State, regional and national research and education networks What are the shared values across these organizations? • Importance of people • Importance of research • Importance of High Performance/High Throughput Computing for solving complex problems • Importance of educating the next generation workforce of scientists and cyberinfrastructure professionals • Importance of expanding the user base What is the shared vision for a national cyberinfrastructure? • Career Friendly: workforce development, retention, and career paths are a primary need and concern, more so than physical resources. • Pervasive: researchers have access to needed data and computing regardless of location. • Easy to use: Computation in science becomes as easy as sending email or completing a file transfer • Connected: Seamless connectivity between people and technology at all scales, both vertically and horizontally, that facilitates and promotes collaboration between necessarily diverse organizations.

15

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF • •

•

Shared: organizations my share resources, human as well as technology, with each other and the nation at large Distributed: both human and technology, while retaining local autonomy, with a shared goal of maximizing benefits to all. Distributed physical resources are an important catalyst to promote distributed and highly scalable workforce development. Dynamic: the scale of various organizations and varied physical resource centers are equipped with resources to meet needed goals (e.g., engagement and training of researchers is most effective locally)

How do these organizations currently support the computational infrastructure? • Providing mechanisms for effective horizontal (peer) collaborations and support, • Bridging the huge scale from a single researcher laboratory to peta-scale national resources, • Informing and equiping the workforce that staffs the aggregate national CI, • Providing larger scale resources than are routinely available at the campus level, • Actively engaging researchers, • Providing outreach training to campuses, regional networks and other recipients of regional Campus CI awards • Representing region as part of national projects such as the XSEDE campus champion project, • Stimulating regional collaborations and common projects with other, nontraditional CI organizations such as regional academic library associations and their membership, • Working across campus with other, non-traditional CI departments (campus library working with campus HPC center), • Holding workshops and training the trainer events, • Making campus visits to smaller institutions to promote aspects of advanced CI, data science and computation, • National to regional collaboration in support of advanced CI training on smaller campuses in a region. Breakout Discussion Summary Breakout discussions were organized around significant topics raised during the panel sessions. Note takers were assigned to each discussion group. A summary of major findings is presented, below. A more complete set of notes are available on request. Accessing the Computational Infrastructure For computation to be pervasive and reach a new level of users, then resources must be readily available to train new computational scientists.



16

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

• •

There is a growing expectation that the undergraduate experience will include exposure to computational research methods and solutions. Industry has growing demand for these skills, increasing the need to expand cyberinfrastructure engagement beyond a small set of highly advanced and specialized sciences.

From certain perspectives, accessing the computational infrastructure is limited to a set of privileged users. These users have • Access to a human support network to help guide them to the resources they need and to support them to use the resources, • Administrators who understand the importance of cyberinfrastructure and are committed to providing what their faculty need to be successful in research and scholarship as well as teaching. Limitations to access can be found at a broad range of research intensive, extensive and other institutions, including campuses that have • no resources, • resources that have been exceeded, and/or • resources that are inappropriate for one’s particular problem. Barriers that contribute to limiting broader adoption of advanced CI resource include • Minimal awareness among potential consumers of the CI resources available; • Lack adequate training and understanding of how to use them; and, • The ‘sharing culture’ of CI facilities may be perceived as threatening to protecting the intellectual property of a potential consumer. There are populations of students and faculty at smaller colleges and universities who want to be intellectually engaged by becoming a part of larger projects. • Engaging these individuals in the larger computational fabric is critical for training of future generations. • Faculty at smaller colleges who trained at resource intensive universities and then move to smaller colleges may still wish to be part of the larger landscape. They are intellectually isolated, not necessarily rewarded, yet anxious to contribute. Many of the discussion participants, familiar with the recent report4 on the value of cyberinfrastructure and computation to EPSCoR jurisdictions, noted that 4 Hill, Paul. (2012). EPSCoR 2030: A report to the National Science Foundation. Retrieved from http://www.nsf.gov/od/oia/programs/epscor/2030%20Report.pdf

17

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

• • • •

• •

The development of robust CI is critical to ensuring that EPSCoR jurisdictions remain competitive, The costs of deploying enhanced cyberinfrastructure can be higher in EPSCoR jurisdictions since they are often rural and sparsely populated, The report’s findings for EPSCoR jurisdictions that are characteristically rural also apply to rural areas of non-EPSCoR jurisdictions, In a state that does not meet EPSCoR eligibility criteria, there are often institutions that closely resemble EPSCoR institutions but are not included, either by choice or neglect, from statewide collaborations, The EPSCoR model of state-wide coordination and support could be applied to multi-state, region-wide efforts, Regional organizations can aggregate the voices of rural institutions to ensure collaborations serve the entire region.

Because of geographic remoteness and sparse populations, faculty, staff and students located in rural regions find it difficult to access CI resources, expertise and training. There is no readily available central repository of resources and expertise for states and regions. Non-Ph.D. degree granting institutions, regardless of size, population served, reputation or financial/human resources are under-represented in computational instrument use. Geographical isolation leaves many rural researchers unaware of rapidly evolving research-computing capabilities. For many of these underserved environments, an aversion to change and a strong culture of self-reliance make adopting advanced digital resources more challenging.

Regional Cyberinfrastructure Cyberinfrastructure, including both networking and computational infrastructure, can be usefully conceptualized as a geography of resources where one or more regional organizations serve as a cyberinfrastructure hub and occupy a key strategic position between campuses and national computational resources. • Institutions across a wide range of scales and missions may be effectively aggregated at a regional level to promote utilization of the computational resources by an ever-expanding user base. • Regional organizations are in a strong position to assist underrepresented institutions in expanding the adoption of computational cyberinfrastructure. • Regional organizations often serve as the connector, including developing personal relationships to support and sustain initiatives. • The proximity to their campus members and familiarity with regional priorities and interests, allow regional organizations to provide focus on challenges and opportunities characteristic of the region as well as to promote new capabilities and resources external to individual campuses.



18

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

• •

•

Regional scale efforts are replicable and can function as a force multiplier for efforts that occur at other levels. With respect to computational and data-intensive science and research, regional organizations have shared values, a shared vision and are poised to move the national science agenda forward through more effective education, outreach, coordination and sharing of resources. Coordination across multiple regional organizations will be useful to develop tactics to maximize adoption of successful activities, minimize duplication of effort and gain a better understanding of what works.

Promoting CI Awareness and Literacy Promoting awareness of and literacy about advanced cyberinfrastructure among researchers, graduate students and IT staff is an important objective in the overall effort to build national cyberinfrastructure. At the campus level outreach and training to the research community is difficult, and even more so at the regional level. Participants in this discussion section confirmed outcomes from other efforts (e.g., Internet2’s Broadening the Reach project) that outreach from “CI aware” campuses to other campuses, particularly smaller campuses and those in more remote and rural environments, is highly desirable but models for successful regional coordination of outreach activities are few. National CI organizations provide their communities of users with information and resources related to their range of services and interests. A key role for a regional CI organization is to organize and provide access to the following resources across their membership, building on existing resources and services, but tuning these to specific regional requirements: • Training (online and in-person) and educational materials and services; • Shared CI infrastructure, tools and services; • Community information sharing and professional development events. Developing a Culture of Bridging the Gap The development of a core community of successful CI consumers within underrepresented communities is critical for expanding the use of computational cyberinfrastructure into a diverse set of users that includes both rural based and smaller organizations. Developing a set of professionals who can speak the language of their peers and bridge between CI providers and CI consumers is essential for successful outreach on a regional level and to build collaborations with researchers and students at smaller institutions. Resource Sharing XSEDE and Open Science Grid (OSG) represent two points along the sharing spectrum. There are increasing efforts to merge the sharing efforts of these respective communities. The OSG is a viable opportunity for a campus to share tangible resources with the broader scientific community. The High Energy Physics



19

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

community is an example that when the incentive for broad sharing is sufficient, the difficulties in sharing physical resources, as well as data, can be overcome. Clouds provide new models for broad sharing of resources, but, to date, are weaker in enabling the distributed sharing of local, autonomous resources between organizations.

Recommendations

Based on needs and opportunities expressed in papers and at the conference, the following recommendations are made. These recommendations are consistent with a vision to effect cultural change and make computational and data resources pervasive, connected, shared, distributed and dynamic.

Core Recommendations The following are over-arching recommendations that cut across communities, agencies and projects. 1. Foster a robust, layered approach to national computational and data-intensive science. Efforts have been focused on strengthening campus and national levels. It is now time to focus on the middle, regional, tier. This middle tier will at least be responsible for standing up shared regional resources where it makes sense, facilitating education and sharing of resources across the region, serving as an accessible point of contact and liaison among campuses and researchers in the region, forging partnerships between regions, and increasing the reach and impact of national efforts such as XSEDE campus champions. Robust regionally-based cyberinfrastructure efforts will support a stronger national infrastructure, creating a layered approach as opposed to a two tiered approach. 2. Engage related communities in support of computational science. As communities become interested in advanced cyberinfrastructure, there is an opportunity to provide an introduction to the field and to encourage member involvement in CIrelated activities. The research library community is an excellent example: Members are natural partners with interest and expertise in lifecycle data management, research tools, research resources, referral and education. The challenge is to identify these communities, to welcome their members and to provide outreach and education. 3. Engage social scientists (psychologists, economists, sociologists) to focus on increasing uptake of advanced CI. The twin issues of expanding access and improving access is a people problem and amenable to the study of those factors that influence behavior at the individual and organizational levels. White paper contributions (e.g., Jolly, et al.) have demonstrated that there are meaningful things to learn. Social science methodology and rigor can help us form meaningful



20

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

hypotheses about variables that influence adoption of advanced CI and help to institute meaningful (valid) metrics. 4. Use technology to improve access to and utilization of computational infrastructure. Just as new smartphone applications such as UBER and Airbnb have led to more seamless sharing of rides and rooms, the use of technology can lead to increased sharing of computational cycles and storage. This must be seamless and work much as the Globus application does, where anyone can put a Globus end point on their system. 5. Break down barriers across the advanced cyberinfrastructure ecosystem. The future of U.S. science depends on a robust U.S. cyberinfrastructure ecosystem. That ecosystem relies, in turn, on advanced computational and data services running on a robust network. It is important that all research and education CI providers understand and acknowledge their role with respect to one another and acknowledge the interdependence of the parts. This means that we create a culture where we are all parts of the same enterprise and where the parts communicate with one another. 6. Incentivize training of the next generation of computational scientists. The availability of advanced cyberinfrastructure to faculty and students at smaller colleges should no longer be a barrier to participation and use. The primary barriers that remain are related to outreach, training and support. Approaches to lowering this barrier to the use of national and regional CI include: • Incentivize training at smaller colleges and universities to include computational science in undergraduate discipline-specific curricula (e.g., chemistry, physics, biology, agriculture), using CI resources available at the national and regional levels. • Partner with existing proven computational and CI training groups, particularly those that cross boundaries (e.g., HPC University, Software and Data Carpentry to create HPC carpentry) to engage in an organized way with smaller colleges and universities. • Support the active inclusion of smaller schools in support networks like XSEDE Campus Champions and the ACI-REF programs. 7. Incentivize training the next generation of CI professionals. The opportunity exists to create new career paths for cyberinfrastructure professionals who will help to improve existing systems, broaden their reach and help to discover new uses. The creation of careers like Cyberinfrastructure Engineer and Data Science Research Specialist are examples. In addition to well-defined job descriptions and expectations, there needs to be • Encouragement and credit for interdepartmental and multi-disciplinary work;



21

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

• •

Incentives for domain scientists to gain expertise in relevant technologies that help advance their fields; Certification programs on proficiency in emerging technologies.

8. Promote a culture of sharing and coordination of existing resources by • Encouraging discussion, arbitration and development of voluntary standards within a region or a discipline that promote sharing and utilization of resources beyond campus boundaries; • Encouraging competition in sharing; • Measuring sharing at campus and regional levels—how much utilization is shared beyond campus members.

Recommendations for Actions Within Specific Communities

This section is concerned with recommendations to existing and emerging communities. Regional networks are well positioned to help expand access to other components of the CI ecosystem. They are self-sustaining and have tentacles into the smaller colleges and universities. While they were originally created to improve access to the national computational infrastructure and improve science, their missions have become more focused on provision of reliable Internet services. A shift is needed to help assure that these organizations play a valued role in the scientific enterprise. We recommend considering that they consider more broadly adopting the following practices: • Adding personnel dedicated to computational education and outreach efforts. These are professionals who can communicate about both the science and the cyberinfrastructure. Their role is to enable researchers to more effectively and efficiently reach their research goals. • Actively seeking out and communicating with researchers at smaller institutions who have unique data storage and/or computational needs. Regional network personnel are in a unique position to assist these researchers, who often have limited time and limited local resources, to navigate the CI terrain. • Identifying and cataloging computational and data resources across their region. Even within campuses, computational and other network reachable resources are often unknown to researchers and campus IT departments. An inventory of resources would be invaluable, not only for researchers, but for regional network organizations in order to provide more resources while keeping costs down. • Facilitating the sharing of computational and data resources across their region. Not all computational resources are appropriate for all research problems and not all researchers have ready access to computational resources. Sharing of resources across campus boundaries will expand



22

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

•

•

•

•

•

availability of the most appropriate resources to researchers who need those resources. Expanding educational and meeting activities to include topics of interest to researchers with computational and data storage needs. Regional networks are geographic centers for cyberinfrastructure and can convene researchers from larger and smaller campuses to learn from one another, to problem solve and to obtain campus, state, regional and national updates. Coordinating efforts with one another (e.g., via the Quilt). State and regional networks have a long history of working together and assisting one another. The Quilt, the member organization of regional networks, currently has an initiative to strengthen researcher engagement and increasing coordination efforts within the research and education networking community. More closely coordinating with national activities like XSEDE Campus Champions, Open Science Grid, and ACI-REF. Regional networks can serve as a communication conduit regarding campus and regional infrastructure and requirements while sharing information about national computation resources with their campus communities. Coordinating efforts with national research and education networks such as Internet2 and ESNet that have similar goals to support academic research for the benefit of furthering scientific exploration. Coordinating end-to-end performance troubleshooting. When problems occur, it is important to rapidly discover things where went awry and get them back on track. Actively implement and promote performance measurement tools such as perfSONAR.

Emerging and Established Projects to Fill Gaps have been developed to work around many of the inherent difficulties with engagement, education. These projects include ACI-REF, XSEDE Campus Champions, XSEDE Campus Bridging and research communities that are virtual in nature. • Recruit participation of other, established communities in meaningful roles to assist in advisory, development, promotion and crossfertilization of communities. These projects can benefit from the capabilities, infrastructure and expertise of regional organizations. • Extend the reach and broaden the impact of project activities by coordinating with other regional (e.g., regional/state network) organizations when working with a campus also supported by them. Through partnerships with regional organizations, these projects can promote their services more broadly (e.g., to smaller campuses, community and tribal colleges). • Increase the audience for training opportunities. Regional CI organizations can help to provide training for multiple institutions in place of a one-time event at a single campus.



23

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

•

Improve the sustainability of support and training activities by involving regional organizations in the provisioning and delivery of these activities, so that regional organizations can provide the services in the future to campuses within their reach. Many regional organizations (e.g., state and regional networks) are self-sustaining and can help to continue valuable services when the project’s lifetime has ended or when key personnel depart the project.

Funding agencies play a vital role in providing direction to the national cyberinfrastructure community in identifying priorities and changing dynamics. • Support acquisition of regional cyberinfrastructure resources to serve a geographic region where the performing regional organization shows commitment to some or all of the following: Operation of the CI resource (including funding the system administration staff, operations cost of the resource and network connectivity); user support, outreach, user training and education; integration of the system with national CI (e.g. level two XSEDE provider and OSG partner, and other regional CI); low barrier of entry for regional researchers to increase and broaden participation; integration of the instrument with high-performance networking and advanced data transfer and data sharing capabilities. Note: Many university consortia that might compete to acquire and operate regional CI resources on behalf of multiple members are dissuaded from doing so when they are administered by a member university. The consortium must, first, compete on a member campus to acquire a limited submission slot in response to infrastructure solicitations (e.g., NSF’s Major Research Instrumentation competition). At one time, consortia were exempt from limitations and did not compete for slots on their own campus. This made it easier for funding agencies to award funds for resources to regional consortia (with bylaws & clearly executed membership agreements) on behalf of their entire membership. • Fund regional centers of CI excellence that are committed to the principles of sharing, education and outreach. Regional centers of CI excellence would assist with sharing resources at multiple levels, with engaging underserved groups and with increasing educational opportunities in the region. These centers would also provide a liaison function between campuses and national infrastructure. Successful models are starting to emerge. However, rather than cloning existing models it is essential that regional CI centers are unique to and meet the needs of a geographic area. • Develop a policy to require a resource-sharing plan akin to a data management plan. • Create a forum for sharing technical expertise developed within university systems on behalf of their campuses and the national community.



24

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

•

Invest in workforce development. For example, National Science Foundation awards to fund the creation of and initial support for Campus Cyberinfrastructure Engineers has lead to increased awareness for this area of specialization across the U.S. Since cross-over between domain knowledge and technical expertise is to be encouraged, new career paths would be open to domain scientists who have acquired sufficient expertise in a given technology to become a critical component of the current cyberinfrastructure workforce.

Conclusions

There are many cyberinfrastructure organizations in the space between campuses and nationally shared cyberinfrastructure facilities that enable use of advanced cyberinfrastructure in research. Now is the time to harness their collective energies and focus them on innovating CI infrastructure and expertise and sharing those solutions on an intra- and inter-regional basis. The recommendations provided in this report are aimed at awakening the potential of the “sleeping middle” of regional network/CI organizations to develop and enrich the national CI ecology through the following broad actions: • Recognize the potential for regional network and CI organizations to be fully integrated into national CI development plans. • Support the development of sustainable models for regional and interregional resource sharing using regional networks as the foundation. • Evaluate and propagate successful collaborative intra-regional programs for education, training, and professional development in CI, computational science and data science skills. • Develop approaches for leveraging regional CI organizations to increase institutional participation in, and uptake of national-scale programs. By carefully considering the information presented, here, and choosing to implement the recommendations from this report, it may be possible to accomplish broad scale change across the cyberinfrastructure landscape in support of future computational and data-intensive science in the United States and beyond.



25

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

Appendix I: Members of the Conference Organizing Committee

Stan Ahalt, RENCI Joni Blake, Greater Western Library Alliance Jim Bottum, Clemson/ACI-REF Gary Crane, SURA Wendy Huntoon, KINBER Tim Lance, NYSERNet Jen Leasure, The Quilt Rick McMullen, Internet2 Gregory E. Monaco, Great Plains Network & Kansas State University Henry Neeman, XSEDE Campus Champions/University of Oklahoma Paul Schopis, OARnet David Swanson, OSG/University of Nebraska-Lincoln John Towns, XSEDE/University of Illinois Taieb Znati, University of Pittsburgh







26

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

Appendix II: White Paper Submission Guidelines

The following was circulated to multiple email lists to encourage white paper submission in advance of the workshop Background Funding agencies like the National Science Foundation (NSF) have made significant investments in the national computational infrastructure to help campus-level researchers answer challenging scientific questions. A diverse set of organizations has emerged to support access to this infrastructure. Nonetheless, gaps remain. The Process We have been funded by the National Science Foundation to develop a set of community-supported recommendations for filling the gaps and for more effectively coordinating education and training on the use of advanced cyberinfrastructure from campus to state to regional to national levels. We intend to use the following process: 1. Solicit white papers from interested parties and organizations. 2. Select among the papers and send conference invitations to the selected group of authors. 3. Hold the meeting in October in Kansas City, MO. 4. Develop a final report and a set of recommendations to funding agencies and the community at large.

Who Should Submit a White Paper We would like to encourage the broadest audience possible to consider submitting white papers to the Conference Organizing Committee. In particular, we would like to hear from ● State and regional networking and research facilitation organizations ● Other organizations that are interested in data and/or computation ● Minority serving institutions ● Researchers ● Campus research support staff ● Organizations interested in workforce development in support of access to computational infrastructure Questions to Address We would like to hear about (a) strategies for engaging researchers to use shared computational infrastructure (e.g., XSEDE, OSG), including shared campus infrastructure, (b) needs that are not currently being met, and (c) successful local and regional models that could be replicated or applied on a broader scale. We would also like white papers to address a subset of the following strategic questions: 1. What should be the role of regional organizations in providing broader support to computational researchers? 2. How can these organizations expand researcher engagement to underrepresented communities?



27

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

3. How can these organizations sustain timely and relevant education and outreach efforts to computational researchers over the long term? 4. How can efforts across these organizations be effectively coordinated? 5. What is an effective collaboration environment and structure? 6. How can these organizations promote and coordinate sharing of hardware, software and expertise across campus, state, and regional boundaries? Due Date September 10, 2015 Paper Format Each paper should be 1.5 to 2 pages in length and single-spaced. Please include the author name and affiliation. These papers will be published at the conference website. Paper Submission Papers may be uploaded at https://kstatedce.wufoo.com/forms/improvingaccess-to-computational-infrastructure/ Conference Organizing Committee • Stan Ahalt, RENCI • Joni Blake, GWLA • Jim Bottum, Clemson/ACI-REF • Gary Crane, SURA • Wendy Huntoon, KINBER • Tim Lance, NYSERNet • Jen Leasure, The Quilt • Rick McMullen, Internet2 • Greg Monaco, GPN/Kansas State University • Henry Neeman, XSEDE Campus Champions/University of Oklahoma • Paul Schopis, OARnet • David Swanson, OSG/University of Nebraska-Lincoln • John Towns, XSEDE/University of Illinois • Taieb Znati, University of Pittsburgh



28

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

Appendix III: List of Submitted White Papers

White papers are available at http://tinyurl.com/pd99qg7 Barber, D., Dolan, J, Reinalda, J., Wolbers, S. Garrick, W., Menken, S., Potter, K., & Corbató, S. Improving Access to the National Computational Infrastructure: the Oregon Perspective Blake, J. M. A Holistic Approach for National, Regional, and Campus Computing Ecosystems Blaustein, G. S. A personal perspective on high-performance computing at a small liberal arts college Bottum, J., & Atkins, D. ACI-REF: Accelerating Campus, Regional, and National Research Through An Advanced Cyberinfrastructure Facilitator Network Bright, N., Knepper, R., Neeman, H., Scott, E., Sherman, A., Smith, P., & Towns, J. The Linux Clusters Institute: Workforce Development and Resources for Campus Research Support Staff Brown, G.D. Individuals: An Untapped Resource, How We Can Encourage and Support Them and Use the Results Brown, J., Coulter, E., Ferguson, J., Hallock, B., Knepper, R., Lifka, D., Pierce, M., Navarro, J.P., Scott, J.R., Smallen, S., & Stewart, C. Technology Contributions toward Regional Coordination of Cyberinfrastructure Resources and Expertise through New Approaches by XSEDE Campus Champion Leadership Team (Hunt, S.K.) Regional Campus Champions Champions Leadership Team (Lathrop, S.) Establishing State/Regional CI Centers of Excellence DuRousseau, D. Beyond Network - Regional Organizations as effective workflow integrators Engel, D., & Suess, J. White Paper for the Workshop on the Role of Regional Organizations in Improving Access to the National Computational Infrastructure Furlani, T., Gallo, S. M., Link, M. R., & Stewart, C. A. How can regional organizations sustain timely and relevant education and outreach efforts to computational researchers over the long term? A proposal for adoption of Open XDMoD (XD Metrics on Demand) Hampton, S.S. White Paper from University of Notre Dame Hill, C. Some lessons from multi-university collaborations in Massachusetts Jennewein, D.M. The Role of Regional Organizations in Advancing Cyberinfrastructure at the University of South Dakota Jha, S., Lynch, H., Nabrzyski, J., Rennermalm, A., Wyngaard, J., & Yarmey, L. Polar Science and HighPerformance Distributed Cyberinfrastructure Jolly, S., Grady, M., Clarke, J., & Guru, A. Opportunities and Gaps in Cyberinfrastructure Lance, T. L. Expanding access to the National Computing Infrastructure in the NYSERNet Community Leake, E. Improving Access to U.S. National Computational Infrastructure Livny, M. Sharing Starts at the Campus



29

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

McMullen, R., Moore, J., & Loftus, G. Inter-regional coordination of programs to increase participation in, and effective use of shared cyberinfrastructure in research, research training, and education Middelkoop, T. The Evolving Roll of Campus Scale Research Computing Monaco, G.E., Leasure, J., Huntoon, W., Meehl, M., Garelik, C., & Shah, P. Toward Integrating Regional Networking Organizations into the Computational Infrastructure Fabric Nabrzyski, J., Gesing, S., Pierce, M., Marru, S., Dahan, M. & Wilkins-Diehr, N. Position Paper from the Science Gateways Community Institute OneOklahoma Cyberinfrastructure Initiative Membership (Brunson, D.) The OneOklahoma Cyberinfrastructure Initiative: A Model for Multi-institutional Collaboration Open Science Grid Executive Team (Würthwein, F.) Towards an Open Pervasive National Cyberinfrastructure Rampp, C White Paper from Franklin & Marshall College Rotman, L., & Zurawski, J. Recommendations for Overcoming Technical and Sociological Barriers to Network Adoption Schopf, J.M., Chevalier, S., Lee, A., & Chitwood, M. PerfSONAR Everywhere for Regional Networks Smarr, L., Crittenden, C., DeFanti, T., Papadopoulos, P., Wuerthwein, F., Andriola, T., Fox, L., & Bell, G. The Pacific Research Platform: A West Coast-Centered Regional Big Data Freeway System Strande, S., & Andriola, T. The Role of University Systems in Improving Access to and Use of Campus-level and National Research Cyberinfrastructure Swanson, D. CAMPUS RESOURCES AS REGIONAL CI Taylor, S., & Deprey, B. CEN’s Role in Improving CT Researchers’ Access to the National Computational Infrastructure von Oehsen, B., & Hauser, T. Funding Regional Advanced Computing Resources to Serve a Broader Community Wang, H., & Siegfried, B. Next Generation Sequencing (NGS) and Campus Computational Infrastructure XSEDE Campus Champion Leadership Team (Brunson, D.) Expanding Roles of Campus Champions



30

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

Appendix IV: Affiliations of Authors Submitting White Papers

Adaptive Computing Enterprises, Inc. Advanced Cyberinfrastructure – Research and Education Facilitation Project (ACIREF) Argonne National Laboratory Benedictine College, Department of Chemistry & Biochemistry Campus Champion Leadership Team Capital Area Advanced Research and Education Network (CAAREN) CENIC Clemson University Connecticut Education Network (CEN) Cornell University EarthCube Research Coordination Network for High-Performance Distributed Computing in the Polar Sciences Energy Sciences Network (ESnet) Franklin & Marshall College Front Range GigaPoP Georgia Tech University Great Plains Network (GPN) Greater Western Library Alliance (GWLA) Holland Computing Center (HCC) of the University of Nebraska (NU) Indiana University (multiple departments) Indiana University Pervasive Technology Institute Internet2 Kansas State University Keystone Initiative for Network Based Education and Research (KINBER) Langston University, Department of Technology Langston University, Dept. of Mathematics Lawrence Berkeley National Lab Linux Clusters Institute Massachusetts Green High Performance Computing Center (MGHPCC) Massachusetts Institute of Technology National Center for Supercomputing Applications NYSERNet Ohio Academic Resources Network (OARnet) Oklahoma Baptist University, Chemistry Oklahoma Innovation Institute Oklahoma Innovation Institute, Tandy Supercomputing Center Oklahoma State University High Performance Computing Center Oklahoma State University (multiple departments) OneNet OneOklahoma Cyberinfrastructure Initiative (OneOCII) Open Science Grid (OSG) Executive Team Oregon Health & Science University



31

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

Oregon State University Pacific Research Platform Pittsburgh Supercomputing Center Portland State University Purdue University San Diego Supercomputer Center Science Gateways Community Institute Shodor Education Foundation, Inc. South Dakota Board of Regents Southeastern Oklahoma State University (multiple departments) Southwestern Oklahoma State University, Department of Business & Computer Science STEM-Trek Nonprofit Texas Advanced Computing Center (TACC), The University of Texas at Austin The George Washington University The Quilt The Samuel Roberts Noble Foundation The University of South Dakota The University of Tennessee, Knoxville The University of Tulsa , Tandy School of Computer Science University of Buffalo University of California Berkeley University of California Office of the President University of California San Diego University of Central Oklahoma, College of Mathematics and Science University of Colorado Boulder University of Florida, Gainesville University of Maryland, Baltimore County University of Missouri University of Missouri, Division of Information Technology University of Nebraska University of Nebraska-Lincoln (multiple departments) University of Notre Dame, Center for Research Computing University of Oklahoma (multiple departments) University of Oklahoma, Cooperative Institute for Mesoscale Meteorological Studies University of Oregon University of Wisconsin – Madison, Center for High Throughput Computing XSEDE XSEDE Campus Champions Leadership Team Yale University



32

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF

Appendix V: Conference Agenda

The following agenda was used to organize the conference segments THURSDAY, OCTOBER 22, 2015 8:00 AM Introduction, Welcome, Logistics REVIEW OF WHITE PAPERS I 8:30 AM Panel A Role of regional organizations in providing broader support to computational researchers (State & Multi-State) 9:15 AM Panel B Role of regional organizations in expanding researcher engagement to under-represented communities 10:30 AM Panel C Role of regional organizations in sustaining timely and relevant education and outreach 11:15 AM Panel D Coordination of efforts across organizations Noon Break REVIEW OF WHITE PAPERS II 1:00 PM Panel E Effective collaboration environments and structure 1:45 PM Panel F Role of regional organizations in promoting and coordinating sharing of resources across campus, state and regional boundaries 3:00 PM Work Session 1: Elaboration of needs and meeting those needs 4:30 PM Review of breakout discussions 5:00 PM Adjourn FRIDAY, OCTOBER 23, 2015 8:00 AM Opening Discussion 8:15 AM Work Session 2: Recommendations and action items 10:30 AM Review and Discussion 12:00 PM Adjourn



33

ROLE OF REGIONALS IN IMPROVING ACCESS: A REPORT TO NSF



Appendix VI: Conference Participants Full Name Kate Adams Daniel Andresen Dustin Atkins Joni Blake Gail Blaustein Gary Brown Dana Brunson Adam Caprez James Cuff Brynn Deprey Patrick Dreher Tessa Durham Brooks Don Engel Jaroslav Flidr Thomas Hauser Robert Henschel Kay Hunt Gwendolyn Huntoon Doug Jennewein Paul Karr Richard Knepper Elizabeth Leake Jennifer Leasure Miron Livny Amit Majumdar Suresh Marru Donald McMullen Timothy Middelkoop Greg Monaco Jarek Nabrzyski Henry Neeman Kristi Potter Jeff Pummill Carrie Rampp David Swanson Barr von Oehsen Dan Voss Frank Wuerthwein Jason Zurawski

Organization* Great Plains Network Kansas State University Clemson University Greater Western Library Alliance Benedictine College Adaptive Computing Oklahoma State University University of Nebraska-Lincoln Harvard University Connecticut Education Network MIT Doane College UMBC George Washington University University of Colorado-Boulder Indiana University Purdue University KINBER The University of South Dakota Wayne State College Indiana University STEM-Trek Nonprofit The Quilt University of Wisconsin-Madison University of California San Diego Indiana University Internet2 University of Missouri Great Plains Network University of Notre Dame University of Oklahoma University of Oregon University of Arkansas Franklin & Marshall College University of Nebraska-Lincoln Clemson University University of Kansas University of California - San Diego ESnet

*This is the Organization Name chosen by participant when registering. In many case, participants represented multiple organizations.

34