The NSF Advisory Committee on Cyberinfrastructure

1
The NSF Advisory Committee on Cyberinfrastructure

• Kelvin K. Droegemeier
• (kkd_at_ou.edu)
• School of Meteorology and Center for Analysis and
  Prediction of Storms
• University of Oklahoma

2
Charge

• Evaluate the current PACI programs
• Recommend new areas of emphasis for the CISE
  Directorate
• Recommend an Implementation Plan to Enact
  Recommended Changes

3
The History of Planning at NSF
CSE research elsewhere in NSF
Support for an array of small, medium, and large
CISE basic research projects
1995
CISE Directorate
Current REPORT
Computational Science init. Expanded equip.
program.
Hayes Report
Provision of advanced scientific computing
5 Supercomputer Centers, NSFnet,
PACI NCSA NPACI
1984
Terascale Computing Initiatives
Lax -gtCurtis/Bardon Reports
1993 BRPDesktop to Teraflop
4
Cyberinfrastructure?

• Traditional infrastructure infrastructure is
  required for an industrial economy.
  Cyberinfrastructure is required for an
  information economy.
• Cyberinfrastructure refers to an infrastructure
  based upon computer, information and
  communication technology that is required for
  discovery, dissemination, and preservation of
  knowledge
• High-end computers
• Fast, adaptable networks
• Software tools and applications
• Data repositories
• Collaboration tools
• Education and training

5
Cyberinfrastructure
6
Knowledge Frontiers

• Several recent projects provide a glimpse of
  cyberinfrastructure

7
Network for Earthquake Engineering Simulation
Remote Users
Instrumented Structures and Sites
High-Performance Network(s)
Laboratory Equipment
Field Equipment
Curated Data Repository
Leading Edge Computation
Remote Users
Laboratory Equipment
Global Connections
8
Information Sources

• Community-wide web survey
• Widely publicized
• gt700 responses
• Quantitative comparisons with the Hayes Report
• Oral public testimony (3 sessions)
• 62 participants selected from research
  scientists and engineers computer and
  computational scientists center directors
  agency and corporate leaders system
  administrators educators students and young
  scientists technicians and consultants
• Emphasis given to traditionally underrepresented
  groups and the physically challenged
• Written transcripts and A/V materials assembled
• Dozens of prior reports unsolicited emails/calls
• 250 pages of written critique from 60 reviewers
• Hundreds of hours of deliberation and discussion
  among Panel members

9
Major Recommendations

• The NSF should take the lead in charting a
  national course for cyber infrastructure an
  Advanced Cyberinfrastructure Initiative (ACI) to
  create, deploy, and apply CI in ways that
  radically empower all scientific and engineering
  research and allied education
• Must be a large, long-term concerted effort (1B
  new dollars), not a linear extension of existing
  activities
• Must be carefully organized and managed
• Human capital should be considered a co-equal
  with traditional physical infrastructure
• High-end resources available to the US academic
  research community should be second to none
  (doesnt neglect low end or mid-sized resources
  MREFC inadequate)
• Physical infrastructure must be maintained for
  the long haul NSF cant keep thinking of
  resources like PACI as research or as an
  experiment that frequently is re-competed
• Must fill the emerging need for a new IT
  professional
• Give emphasis to underrepresented groups, the
  physically challenged, and remote users
  (digital divide)

10
A 1B Investment of New Funds
11
Fundamental and Applied Research to Advance CI

• 30 Projects at 2M/year (average)
• Single investigators and teams (e.g., SRB)
• Topics
• Human-computer interactions
• Data base systems
• Networks
• Parallel computing
• Architectures
• Security
• Reliability
• Interoperability
• IP

12
A 1B Investment of New Funds
13
Application of IT to Domain Science and
Engineering Research

• 50 Projects at 2M/year (average)
• Long-term (5 years)
• Single investigators and teams
• Domain-specific scientists working with computer
  and computational scientists, mathematicians,
  technologists
• Similar in concept to current ITR

14
A 1B Investment of New Funds
15
Acquisition and Development of CI and Applications

• 20 Software Development Projects at 5M/year
  (average)
• Develop production software, commercialize,
  maintain, upgrade
• 10 CI Software Centers (10M/year average)
• Focus on specific community-wide issues, e.g.,
• Grid computing
• Compilers
• Run time systems
• Visualization
• Programming environments
• Parallel file systems
• Human-computer interfaces

16
A 1B Investment of New Funds
17
Provisioning of Resources

• 5 high-end general-purpose computational centers
  (75M each, or 2x current level per center)
• 75 data repositories (average of 2M/year)
• Mostly disciplinary but interoperable
• 5 data technology development centers (3M/year
  each)
• 10 discipline-specific teams for meta data
  standards, formats, tools at 2M/year each
• Increase digital libraries from 10M to 30M/year
  for 1-3M/annual projects
• 60M/year for high-speed backbones
• 5 application service centers at 2M/year each
  for non-computational applications, cluster
  computing, etc

18
Cyberinfrastructure Diversity
Capability not just capacity technology,
policy, tools. Still need some center-based
leading- edge,super computers.
On-demand supercomputing,not just batch.
19
Is the Pyramid Still Valid?

• Most feel that it is, though broadening needs to
  occur below the peak
• Especially strong sentiment from PACI User
  Advisory Committees
• Some argue for a single, extremely powerful
  cycle shop, with other funding directed toward
  campuses for the purchase of local machines
• Its clear that increasing emphasis must be
  placed on data
• Acquisition (sensing)
• Generation (numerical simulations)
• Cataloging (meta data)
• Analysis/processing (mining)
• Storage and stewardship (including legacy and
  traditionally non-digital information)
• Interoperability (technical and among disciplines)

20
Recommended Organizational Structure Within the
NSF

• An ACI Program Office (APO), managed in a matrix
  fashion overlaid on existing NSF structure, with
  a highly placed, credible leader will be charged
  with
• Serving as a national ACI coordinator
• Managing the ACI budget/priorities
• Linking ACI activities among directorates and
  funding them the ACPO does not fund
  investigators or projects directly, but funds
  directorates, each of which develops its own
  initiatives
• CISE takes a lead role in infrastructure and
  applications identifies common needs among
  directorates
• ACPO director reports to a steering committee of
  ADs, chaired by the CISE AD
• Workshop held May 14-15 AG session on May 16 to
  discuss management models broadly defined

21
http//www.si.umich.edu/cyberrequest
22
Recommended Future of PACI within the ACI

• Two-year extension of current PACI program
  through 2004
• Until 2007 (10 years from start of program),
  PACIs and PSC should be assured of stable
  funding to provide high-end resources and
  associated operations
• Important to retain skilled PACI staff and
  successful collaborations dont want a mass
  exodus!
• 2004 or 2005 ACI funding begins
• Separate peer-reviewed enabling and application
  infrastructure projects EOT
• PACIs can compete for all aspects of the larger
  ACI funding

23
Current Status at NSF

• Solicitation for new Extensible Terascale
  Facilities Partners (proposals were due 9 June
  panel on 26-27 June)
• Funds connection and integration
• Archival repositories
• Digital libraries
• Computational resources
• Sensor networks
• PACI transition period (March 03 September
  04)
• 1-year extension of Cooperative Agreements
• Maintain technology thrusts
• Support resource partners encourage
  participation in ETF program
• Increase support for Domain-Specific Cybertools
  via FY04 ITR focus area reduce funding through
  PACI
• Fund terascale operations through September 04

24
Current Status at NSF

• Many Directorates/Advisory Committees are in the
  process of planning for cyberinfrastructure

25
Current Status at NSF
26
Current Status at NSF

• Within the Geosciences Directorate
• Efforts underway within each division
• Atmospheric Sciences (ATM)
• Ocean Sciences (OCE)
• Earth Sciences (EAR)
• Approach in ATM as you begin your own planning
• Open letter to the community (completed)
• Regional focus group meetings to obtain broad
  input on
• State of the art
• Key issues of broad relevance
• Recommendations to NSF for funding and
  implementation/management
• Tentative locations Washington, DC Atlanta,
  GA Champaign, IL Boulder, CO Seattle, WA San
  Diego, CA
• Final report to be submitted to GEO AD

27
Current Status at NSF

• Cyberinfrastructure initiative likely will be
  more than traditional priority area efforts
  (e.g., Nano, Biocomplexity, Math, ITR)
• Hopefully will be sustained CI is a very
  different animal
• We all have an opportunity to shape its direction
• Your workshop is extremely important and timely
• A couple of suggestions
• Be sure to communicate with and think about other
  disciplines
• Be sure to participate in the management models
  sessions or give input via the web
• Working document on May workshop is in
  preparation
• 23 June 2003 via Access Grid (3 pm EDT)
  (coordinated with Global Grid Forum)

28
http//www.si.umich.edu/cyber