Title: NSF Activities in Shared Cyberinfrastructure
1NSF Activities in Shared Cyberinfrastructure
- Kevin Thompson
- Program Director Division of Shared
Cyberinfrastructure - Directorate for Computer and Information Science
- and Engineering
- National Science Foundation
- November 24, 2004
2Outline
- SCI Guiding Principles
- SCI Programs
- Looking to the Future
3NSB-02-190 on SE infrastructure (2002) also the
more specific charge to the blue ribbon panel on
Cyberinfrastructure (2003)
NSF Programsimplement NSB Policy!
4Research InfrastructureNSB-02-190
- Hardware (tools,equipment,instrumentation,
platforms, facilities) - Software (enabling computer systems, libraries,
databases, data analysis interpretation
systems, comm. networks) - Technical support (human or automated, services
to operate and maintain infrastructure) - Special environments installations - to
effectively create, deploy, access, use research
tools (building space)
5Unmet NEEDSper NSB-02-190
- Lack of long-term stable support for wetware
archives prevents rapid advances in post-genomic
discoveries (Bio) - Lack of a large-scale network infrastructure for
next generation of secure network protocols
architectures could be developed tested
(CISE) - Lack of support for new international social
science surveys (SBE) - Lack of synchrotron radiation facilities with
orders-of-magnitude increase in luminosity (MPS)
6Further Drill Down onNSB-02-190
- NSB-02-190 stresses the importance and impact of
Cyberinfrastructure, but treats it succinctly
the NSB linked their 2002 report to the more
extensive (2003) discussion on CI in the Atkins
Report meaning - Focus on implementing recommendations in these
reports, vs. creating new policy (not our role)
7High Visibility
- The nations IT capability is like adrenaline to
SE as well as to the overall economy - Recognized, periodic infusion of large investment
(e.g. ITR via PITAC) - Next step build the most advanced research
cyberinfrastructure while simultaneously
broadening its accessibility feedback (on their
expectations) from the broad SE community
8Guiding Principles for SCI at NSF
- Serve all of science engineering
- Firm and continuing commitment to providing the
most advanced cyberinfrastructure (CI), with
high-end computing (HEC) at the core - Encourage emerging CI while maintaining and
transitioning extant CI - Provide balance in CI equipment
- Strong links to ongoing fundamental research to
create future generations of CI
9Unique Features of SCI
- A core service mission that most divisions at
NSF do not have - Frequent interactions with the entire spectrum of
hierarchy at NSF, including the Office of the
Director, the NSB, directorate leadership and
program officers across all SE disciplines - High visibility of SCIs activities to Congress
and the broad (international) SE community
10The Culture of Deployment
- Within the dominant NSF culture of discovery
research at the frontiers SCI must create and
sustain a subculture of timely delivery
(deployment) of enabling infrastructure, a
challenge of subtle complexity because - The teams building this infrastructure consist of
(e.g. are lead by) SE researchers working in the
spirit of public service to their peer community
(an axiom based on historical patterns of SE
research as the driver for new cyber
technologies) - Need to create a complete team around the SE
leadership (e.g. sustained support for research
staff on deployment scale projects)
11History of NSF CI Investments
12 Integrated CI System meeting the
needs of a community of communities
- Applications
- Environmental Science
- High Energy Physics
- Proteomics/Genomics
-
DevelopmentTools Libraries
Education and Training
Discovery Innovation
Grid Services Middleware
Hardware
13SCI Active Programs
- Centers (SDSC, NCSA, PSC)
- TeraGrid/TCS/DTF/ETF
- IRNC
- NMI
- Other initiatives
- Other programs with no current solicitations (but
with active award activities)
14Extensible Terascale Facility 2003
15(No Transcript)
16TeraGrid Partners
17International Research Network Connections (IRNC)
Program
- IRNC Solicitation - a small number of awards to
provide network connections linking U.S. research
networks with peer networks in other parts of the
world - Follow-on to HPIIS program
- 5M total per year over 5 years
- Production-level state-of-the-art network
services - Activities will include strong monitoring/measurem
ent across all funded connections - Awards
- Not yet official! (left blank until December)
18Other SCI-related Programs
- Cyber Trust (CI)
- Open now (2/5/05)
- goal to improve national cyber security
- Science and Engineering Information Integration
and Informatics (SEIII) - Open now (12/15/04)
- info mgmt and data analysis applied to specific
science domains - Special emphasis on domain-specific and
general-purpose tools for integrating info from
disparate source - Science of Design
- focuses on the scientific study of the design of
software-intensive systems that perform
computing, communications and information
processing
19Historical Programs in SCI
- Strategic Technologies for the Internet (STI)
- Experimental Infrastructure Networks (EIN)
- Complimentary program to NRT
- Characteristics
- Control/management of the infrastructure
end-to-end - End-to-end performance/support w/ dedicated
provisioning - Pre-market technologies, experimental h/w, alpha
software - Significant collaboration vertically and across
sites - Persistence, w/ repeatable network experiments
and/or reconfigurability - Experimental protocols, configurations, and
approaches for high throughput, low latency,
large bursts - 10M spent in 2003 funds
20Active EIN Activities
- Collaborative Research Dynamic Resource
Allocation for GMPLS Optical Networks (DRAGON) - PIs Jerry Sobieski UMd), Herbert Schorr (Tom
Lehman - ISI East), and Bijan Jabbari (GMU) - Collaborative Research End-to-End Provisioned
Optical Network Tested for Large-Scale eScience
Applications (CHEETAH) - PIs Malathi Veeraraghavan (UVA), John Blondin
(NC St.), and Nageswara Rao (ORNL) - Several other awards not directly relevant to
optical networking
21CHEETAH Circuit-switched High-speed End-to-End
Transport ArcHitecture
- Create a network offering end-to-end BW-on-demand
service - Make it a parallel network to existing high-speed
IP networks - Combination of Ethernet and SONET extend to MPLS
and VPLS as part of a DOE project. - Transport protocols
- Fixed-Rate Transport Protocol (FRTP)
- Based on SABUL code modified for dedicated
circuits - Terascale Supernova Initiative (TSI) Applications
- High-throughput file transfers
- Interactive apps like remote viz
- Remote computational steering
- Team University of Virginia, Oakridge Natl.
Labs, North Carolina State Univ., City Univ. of
NY
22Cheetah network
23DRAGON Overview
- Experimental deployment of advanced optical
network in Washington D.C. metro area - Research (and development) of advanced network
services in support of eScience applications - Real-time inter-domain provisioning of
deterministic paths in response to application
requests - Associated security, AAA, scheduling capabilities
- Automatic generation of Application Specific
Topologies based on above capabilities - Collaboration with specific eScience applications
to enhance domain science - eVLBI (electronic Very Long Baseline
Interferometry) - High Definition (HD) format collaboration and
remote visualization - See dragon.east.isi.edu
24DRAGON Status
- Network Research Status
- Inter and Intra-Domain provisioning development
and testing underway - Working on architectures for security, AAA, and
scheduling capabilities - Application Research Status
- eVLBI real-time correlation testing underway.
Working on 1 Gbit/s real-time correlation between
MIT Haystack Correlator, Westford Antenna, and
GGAO Antenna. - Testing underway for provisioning of low latency
paths for native and uncompressed HD formatted
data streams - Demos of both at SC04
- Network Deployment Status
- Year 1 network deployment complete, Year 2
deployment underway - Current Sites NASA GSFC, UMD/MAX, GWU, USC/ISI
East, Eckington, MIT Haystack Observatory - Sites anticipated for Yr2 USNO, NCSA ACCESS
- Anticipated Yr2 Optical Network Peerings NLR,
HOPI, DOD GIGEF (formerly ATDNet-BOSSNet) - Other Possible Optical Network Peerings DOE
UltraScience Net, CHEETAH, OMNINet
25Other SCI Activities
- 2004 ITR Awards
- HPWREN (PI Braun, SDSC) - see hpwren.ucsd.edu
- Improving the Integrity of Domain Name System
(DNS) Monitoring and Protection (PI Kim Claffy,
SDSC) - see http//www.caida.org/projects/dnsitr/ - ITR large projects (started prior to 2004)
- Optiputer
- iVDGL/GriPhyN
- GEON, LEAD, others
- Other CI-heavy Programs
- NEES
26NSF Middleware Initiative (NMI)
- Purpose - To design, develop, deploy and support
a set of reusable, expandable set of middleware
functions and services that benefit applications
in a networked environment - Program encourages open source software
development and development of middleware
standards - Release 6 coming out in December
- NMI fills an important need for software and
services that address the complexity of achieving
interoperability among heterogeneous platforms
and environments, to let end users concentrate on
developing productive applications rather than
having to reinvent core middleware for each
project.
27NMI System Integrators
- Disseminating and Supporting Middleware
Infrastructure Engaging and Expanding Scientific
Grid Communities - PI Randy Butler (NCSA)
- Designing and Building a National Middleware
Infrastructure (NMI-2) - PI Carl Kesselman (USC/ISI)
- An Integrative Testing Framework for Grid
Middleware and Grid Environments - PI Miron Livny (U Wisc)
- Extending Integrated Middleware to Collaborative
Environments in Research and Education - PI Klingenstein (UCAID)
- Instruments and Sensors as Network Services
Instruments as First Class Members of the Grid - PI Rick McMullen (Indiana)
- Collaborative Proposal Middleware for Grid
Portal Development - Pis Pierce (Indiana - Lead), Alameda, Severance,
Thomas, and von Laszewski
28NMI 2004 Solicitation
- Due May 14, 2004
- Two Areas
- Integration and Deployment of middleware with
domain se apps to create production environments
(2-3 years, - Development and prototyping of new functions and
services (2-3 years, 50k-500k/year) - Anticipated Funding 7-14 awards
292004 NMI Awards
- Over 140 proposals
- 13 awards made
- 2 Deployment awards totaling 6M
- 11 Development awards totaling 6M
- 3 awards in Small Grants for Exploratory Research
(SGERs) - Other Awards of note
- See www.nsf-middleware.org
30NMI Deployment Awards
- nanoHub (PI Sebastien Goasguen, Purdue)
- www.nanohub.org
- Deploys middleware to extend/enable NSF Network
for Computational Nanotechnology (NCN) - Condor-G , Teragrid, UWisc grid assets used
- Apps include NEMO and nanoMOS
- 3 year timeframe
31NMI Deployment Awards
- GridChem (PI John Connolly, Kentucky)
- Builds shared grid for computational chemistry
with additional apps for genomics and proteomics - Condor-G , GT3, uberFTP, GridPort, NWS used
- App codes Gaussian, GAMESS, Molpro, NWChem,
ACES2, Q-Chem, Crystal, NBO, Wien2K, MCCCS,
Towhee - Partners UKy, Texas, OSC, LSU, NCSA
- 3 year timeframe
32NMI Development Awards
- Scalable Distributed Information Management
System - PI Michael Dahlin, Texas, 3 years
- Policy Contolled Attribute Framework
- Collaborative, 2 years
- PI Katarzyna Keahey, ANL
- PI Von Welch, UIUC
- Performance Inside Performance Monitoring and
Diagnosis for NMI Software and Applications - PI Jennifer Schopf, ANL, 2 years
33NMI Development Awards
- Middleware for Adaptive Robust Collaborations
across Heterogeneous Environments and Systems - PI Liang Cheng, Lehigh, 3 years
- NMI Development for Public-Resource Computing
- PI David Anderson, UC Berkeley, 2 years
- Mobile-Agent-Based Middleware for Distributed
Job Coordination - PI Munehiro Fukuda, U Washington, 3 years
- Energy-Conserving Middleware with Off-Network
Control Processing in Embedded Sensor Networks - PI Subir Biswas, MSU, 3 years
34NMI Development Awards
- Design, Development, and Deployment of the Web
Services Resource Framework on the .NET Framework
- PI Marty Humphrey, Virginia, 2 years
- The Computational Chemistry Prototyping
Environment - PI Kim Baldridge, SDSC, 2 years
- GridNFS
- PI Peter Honeyman, Michigan, 3 years
35Other NMI Related Funding
- Keeping Condor Flight Worthy
- PI Miron Livny, 3 years
- Core support for Condor group activities
- Towards a Virtual Solar-Terrestrial Observatory
- PI Peter Fox, UCAR, 3 years
- SEIII Program award
- Dependable Grids
- PI Andrew Grimshaw, UVA
- ITR award
36Cyberinfrastructure the future consists of
- Computational engines (supercomputers, clusters,
workstations capability and capacity) - Mass storage (disk drives, tapes, ) and
persistence - Networking (including optical, wireless,
ubiquitous) - Digital libraries/data bases
- Sensors/effectors
- Software (operating systems, middleware, domain
specific tools/platforms for building
applications) - Services (education, training, consulting, user
assistance)
All working together in an integrated fashion.
37SCI Future Deliverables
- substantial investments to provide increasingly
powerful cyberinfrastructure to support demands
of modeling, data analysis interpretation,
management research - testbeds to develop prove experimental
technologies - expand availability of HPC network resources to
the broader research education community - user-friendly software better software
integration for effective utilization of advanced
systems - support for technical (support) staff for
evolving CI
38Looking to the Future
- Science frontiers as the drivers
- Balance capability and capacity
- the Extensible Terascale Facility (ETF)
- A strategy of a balanced and broad CI to serve
all of science and engineering transition from
extant CI to the exciting possibilities of future
CI - A few specifics
- Continued support of ETF
- CI-TEAM solicitation aimed at
educating/training/advancing/mentoring current
and future generations of computational
scientists and engineers - Establishing a CI software center? (must address
Grids Center functions) - Post-ITR program(s) and support for experimental
networking?
39(No Transcript)