Transcription
National Science Foundation AdvisoryCommittee for Cyberinfrastructure (ACCI)Spring Meeting, April 10 th -11 t h , 2019Meeting MinutesDay 1 (April 10, 2019)ACCI Welcome and Committee IntroductionsGwen Jacobs, Chair, ACCIGwen Jacobs began the meeting, went through the agenda, and allowed current members to introducethemselves. Five new members were welcomed to the Committee, Deborah Dent, Susan Gregurick,Ruth Marinshaw, Ellen Rathje, and Valerie Taylor. Dr. Jacobs also notified the Committee and membersof the public who were present that the proceedings would be audio-captured. The transcript is used tocreate the minutes and resolve questions of fact from the AC members. The transcript is not retained aspart of the record of the meeting and is purged after the minutes were created and approved by the AC.Updates from the Office of Advanced CyberinfrastructureManish Parashar, Office Director, Office of Advanced Cyberinfrastructure (OAC)Manish Parashar gave an overview of activities carried out by OAC. Manish began by pointing out therecent image capture of a black hole by the Event Horizon Telescope, along with the NSF-funded CI thatcontributed to it. After going through an overview of OAC activities, several OAC program directors gaveprogrammatic updates related to Learning and Workforce Development, High-Performance Computing,Software and Data Services, Networking and Cloud, Public Access, and Strategic Engagements. Heconcluded the update by discussing his OAC Vision and Blueprint, the Future of Facilities’ Science and CI,CI and AI, as well as Strategic Partnerships across NSF and toward national priorities.Committee Discussion and Q&A covered Committee interest in translational CI research, partnershipswith commercial cloud services, demographics of Cybertraining proposer domains, and softwareservices.Working Group BreakoutsGwen Jacobs, Chair, ACCIGwen Jacobs led a discussion on the two working groups discussed last Fall (Sustainability andReproducibility) along with a new possible working group (CI Research Innovation). After the discussion,the Committee split into three groups, one for each proposed working group, to discuss their chargesand plans of action.
Update on CISE Directorate ActivitiesJim Kurose, Assistant Director, Computer & Information Sciences & Engineering (CISE) DirectorateJim Kurose gave an overview of recent updates within the directorate. He began by showing thealignment of CISE activities with Administration and Congressional Priorities; the recently signedexecutive order on Artificial Intelligence being called out specifically. He went through a breakdown ofthe CISE budget by programmatic areas, including infrastructure investments and research investmentsacross all CISE divisions. He discussed the overlap of computer science and cyberinfrastructure withseveral of the NSF Big Ideas, focusing on some examples within Multi-messenger Astrophysics andNavigating the New Arctic. Five of the big ideas (Harnessing the Data Revolution (HDR), Future of Workat the Human-Technology Frontier (FW-HTF), Quantum Leap (QL), and Mid-Scale ResearchInfrastructure) have released CISE-focused solicitations so far in FY 2019. Some time was given to theConvergence Accelerators, meant to be more mission driven, use-inspired, transition to practiceactivities. In addition to the Frontera acquisition described earlier in the day, the Exploring Clouds forAcceleration of Science (E-CAS) partnership with Internet2 was briefly described. Two CISE researchinfrastructure solicitations were highlighted: CISE Community Research Infrastructure (NSF 19-512) andCloud Access (NSF 19-510). Several education and workforce activities, including CSforAll and Computingin Undergraduate Education, were described.Panel with Consortia RepresentativesSharon Broude Geva, Coalition for Academic Scientific Computation (CASC)Thomas Cheatham, Campus Research Computing Consortium (CaRCC)John Towns, Practice and Experience in Advanced Research Computing (PEARC)Gwen Jacobs introduced a session to hear from three community consortia representatives. Threequestions were posed to each consortia representative:1) Who are your members?2) What is your scope?3) What is your contribution to our role in provisioning advanced cyberinfrastructure andcyberinfrastructure services?Sharon Broude Geva began with an overview of the mission of CASC, which is to “further thedevelopment of a national infrastructure of advanced scientific computing, communications andeducation resources. CASC will cultivate the funding of, and support for, high performance computingand communications initiatives in support of academic institutions.” CASC’s constituency was describedas “People whose role focuses on strategy, vision, policy, funding, and advocacy of advanced researchcomputing within their institution or organization.”Thomas Cheatham described the Campus Research Computing Consortium (CaRCC). This consortiumcame from the previously NSF funded Advanced Cyberinfrastructure Research & Education Facilitators(ACI-REF) program. System administration professionals, software professionals, CI engineers, campuschampions etc., wanted to collaborate. Currently about 35 universities are involved in CaRCC. The visionof CaRCC is “to advance the frontiers of research at academic institutions by supporting on-campusawareness and facilitation services related to computation for researchers, including inter-institutional
resources and knowledge sharing among research computing professionals, and continuous innovationin research computing (and data) capabilities.” Framework for research computing and data workcategorized into Researcher facing, Systems facing, Software/Data facing, and Sponsor/Stakeholderfacing roles. Current model focuses on what CaRCC “does” vs what CaRCC “is.”John Towns discussed the roles of the Practice and Experience in Research Computing (PEARC)consortium. PEARC mission is “providing a forum owned by the community to foster exchanges aroundthe ‘state of the practice’ in advanced research computing – discussing challenges, opportunities,solutions, and best practices.” Main activity is an annual conference planned by a steering committeewith sessions targeted at the interests of the practitioner community. The annual conference alsoincludes student programs, with 75-100 students involved each year. The annual conference wasoriginally run by XSEDE. Since then there have been two PEARC Conferences, PEARC17 – New Orleans,LA and PEARC18, Pittsburg, PA. This year PEARC19 will be held in Chicago, Ill. PEARC20 will be held inPortland, OR and is now sponsored by ACM SIGHPC.
Day 2 (April 11, 2019)Outstanding Issues from Day 1Gwen Jacobs, Chair, ACCIGwen Jacobs began the second day of the meeting with approval of the Fall meeting minutes. Dates fornext Fall were discussed and tentatively agreed to September 18-19, 2019. Manish Parashar providedthe required annual update on steps OAC has taken to address recommendations in the 2017Committee of Visitors (COV) assessment of OAC activities from 2013-2016.An overview of expectations for ACCI working groups was discussed. Each of the three working groups(reproducibility/replicability, sustainability, and CI research) gave presentations going over the previousday’s discussion.Preparation for visits by NSF Assistant Directors and Office HeadsGwen Jacobs, Chair, ACCIMembers of ACCI prepared questions and logistics for the visit from Assistant Directors and Office HeadsVisit by NSF Assistant Directors and Office HeadsKelli Craig-Henderson, Deputy Assistant Director, Social, Behavioral, and Economic Sciences Directorate(SBE)Stephen Meacham Section Head, Section Head for Integrative Activities, Office of Integrative Activities(OIA)Deborah Lockhart, Deputy Assistant Director Mathematical and Physical Science Directorate (MPS)Sam Howerton, Deputy Office Director, Office of International Science and Engineering (OISE)Jim Kurose, Assistant Director, Computing and Information Science and Engineering Directorate (CISE)Bill Easterling, Assistant Director, Geosciences Directorate (GEO)Karen Marrongelle, Assistant Director, Education and Human Resources Directorate (EHR)Joanne Tornow, Assistant Director, Biological Sciences Directorate (BIO)Dawn Tilbury, Assistant Director, Engineering Directorate (ENG)Douglas Maughan, Office Head, Convergence Accelerator (C-Accel)Kelly Faulkner, Office Head, Office of Polar Programs (OPP) Bill Easterling of the Geosciences (GEO) directorate, discussed the challenges facing theGeosciences, such as how modeling the atmosphere necessitates including ocean, land, andspace weather into some calculations. He noted the Cheyenne machine, a 5.34PFsupercomputer, was recently inaugurated at the National Center for Atmospheric Research(NCAR). He also discussed the Earthcube program, which has many CI focused projects andactivities.Jim Kurose (CISE) highlighted the updated CISE Community Research Infrastructure (CCRI)program.
Dawn Tilbury (ENG) discussed a large need among engineering disciplines for standards, such asdata structures, data storage, and software.Douglas Maughan (C-Accel) discussed the new convergence accelerator at NSF and thetransition of research activities to applications and to market.Samuel Howerton (OISE) highlighted the challenging issue of providing good domestic policy andhow it fits in internationally.Deborah Lockhart (MPS) described the diversity of the MPS directorate, with differentcomputational and cyber needs. Stakeholders range from single investigators to large teams,including large facilities with unique CI needs. As an example, the LHC high luminosity upgrade isa current large facility project with an estimated completion in 2026. This project will come witha huge influx of data. The cost of data transfer to the various centers was noted as a bigconcern. One example of partnering with OAC is the recently awarded Institute for Research andInnovation in Software for High-Energy Physics (IRIS-HEP), but several challenges remain. Otherexamples with large data challenges include the Laser Interferometer Gravitational-WaveObservatory (LIGO) and Large Synaptic Survey Telescope (LSST).Stephen Meacham (OIA) discussed his offices role in coordinating crosscutting efforts, includingMajor Research Instrumentation (MRI) and Science & Technology Centers program, whichinclude cyberinfrastructure. OIA also has a large focus on EPSCoR.Joanne Tornow (BIO) discussed her directorate’s large span from molecular to ecosystemscience. Such research is rich in multi-dimensional data. One major challenge mentioned wasthe integration of heterogeneous datasets. While software is needed in many fields, thequestion is whether it needs to be written from scratch for every case. Large scale recapitulatesthe same problem: software and data developed from scratch. Data integration challenges arenot unique to BIO, so OAC plays a great role in identifying parallel efforts that could be learnedfrom.Kelly Craig-Henderson (SBE) highlighted the broad range of disciplines supported under SBE withdifferent CI needs. One role of the directorate is support of large nationally representativesurveys. Effectively using administrative data and survey data is a large data challenge.Investigators within SBE that rely on data coming from psychological, neural, and behavioralresearch face unique interoperability issues. One program in SBE particularly relevant to CI isRIDIR: Resource Implementation for Data Intensive Research in SBE sciences. SBE also has alarge role in artificial intelligence research. One last point: in addition to the challenge ofworkforce itself, there is a problem of lack of access to high-performance computing by SBEresearchers.Kelly Falkner (OPP; relayed by Patrick Heimbach) highlighted a successful collaboration betweenOAC and OPP on the development of a very high resolution model of Arctic and Antarcticaelevation maps.A common refrain across virtually all the discussants was the challenge of attracting and retaining highlyskilled research staff given the opportunities available in industry.Preparation for visit by NSF Chief Operating OfficerGwen Jacobs, Chair, ACCI
Members of ACCI prepared questions and logistics for the visit from the Chief Operating Officer of NSF,Fleming Crim.Visit by NSF Chief Operating OfficerFleming Crim, Chief Operating Officer, National Science FoundationFleming Crim began by highlighting the recent construction of an image of a black hole by the EventHorizon Telescope; especially the photos of palettes full of hard drives being shipped to data processingfacilities. Fleming was complemented by members of the Committee for the NSF role on thebreakthrough as well as the public press surrounding the discovery.Fleming discussed the NSF budget. The new budget of 8.1B dollars will be the first time over 8B for NSFfunding. Budget discussions started the previous spring/summer. In the fall, NSF submits their budget tothe Office of Management and Budget (OMB). Several iterations ensue over the next few months. Afterall such “passback,” the NSF budget is folded into the President’s budget proposal. Current budgetrequest for the following year is 7.1B and is currently being considered by congress.Of note beyond NSF within the government, Kelvin Droegemeier was recently confirmed to the Office ofScience & Technology Policy (OSTP) and integrating himself into the process. NSF has always workedclosely with OSTP, such as the recent rotation Jim Kurose served over several months.Regarding Convergence Accelerators: the idea is to build public-private partnerships and producedeliverables from research. Two main big ideas for focus of this effort is HDR and Future of Work. NSF isdelighted to welcome Douglas Maughan as Director of the accelerator office.The positioning of OAC within CISE remains strong and critical to NSF as a whole. Any future leader ofCISE require the understanding of this positioning.OAC blueprint document has been getting a lot of feedback.Collaboration with BIO on a CI center of excellence.During large facilities meeting in Austin, recent OAC investments in Stampede 2 and Frontera wereshowcased.DiscussionThe challenge of employing CI professionals was discussed at length. The incentive structures of nationallabs, university applied science labs, and other retention efforts were pointed out by members of theCommittee. Cybertraining initiatives recently put forward by OAC were noted as positive by members ofthe Committee. Sometimes faculty dedicated to cybertraining as not seen as “rewarded” as otherfaculty getting funded for “core” research. Sometimes the faculty doing cybertraining are females orother minorities, and it is important to ensure they are recognized for their contributions.Can NSF fund the transformation of public education? NSF can fund specific research that can transformeducation, though it is hard to support the effort directly. Jim Kurose interjected that NSF can injectsomething in a scalable way. For example, in the NSF CS4All activity resulted in 60,000 students takingthe new CS AP exam. The question is what can be done with a relatively small scale of funding.
Wrap up, next steps, and adjournGwen Jacobs, Chair, ACCIThe logistics of working group activities was discussed. Gwen Jacobs went over options for agendas infuture meetings, including inviting speakers of interest to the Committee. Feedback on the meeting wasrequested, and members praised the agenda structure and materials.The meeting was adjourned at 2pm.APPENDIX: ACCI Member Attendee ListTilak AgerwalaDeborah DentDaniel GoroffPatrick HeimbachGwen JacobsEd LazowskaRich Loft (remote)Ruth MarinshawValerio PascucciKristin Persson (Day 2)Michela TauferValerie Taylor (remote)
National Science Foundation Advisory Committee for Cyberinfrastructure (ACCI) Spring Meeting, April 10th-11th, 2019 Meeting Minutes Day 1
