About the series
Good morning, I'm delighted to be here with you today to share with you our future plans for cyberinfrastructure. This community, NPACI and the Alliance, of many disciplines working together with users to create cyberinfrastructure is absolutely essential. I'm here to assure you that as we move forward, we will do everything possible to maintain and build your community.
But before I begin, I'd like to recognize the breadth and depth of the outstanding contributions made to science by my highly-respected colleague, and your fearless leader, Fran Berman. Because of these contributions, Fran was recently appointed to the California Council on Science and Technology - a not-for-profit corporation established by the California State Assembly to examine urgent public policy questions relating to science and technology. My warmest congratulations, Fran. Furthermore, I'm also delighted to see in the audience today, someone who has made a real impact at the National Science Foundation, and who I know will make enormous contributions to SDSC in his new role as Executive Director, Alan Blatecky. There is no doubt that SDSC has a world-class leadership team.
A number of you have asked me how I like being at the National Science Foundation. Well, as the head of an academic unit, I sometimes used to feel like a graveyard keeper. Yes, I had a respected position with a lot of people "below" me in the organization - but they didn't seem to listen very much! Now, as head of a smaller organization of people but possessing substantially greater resources, I find I suddenly have a lot more friends around the country than I ever imagined - and they are all willing to listen to me! Actually, both positions have been greatly rewarding, and I am really enjoying my new role at NSF.
As a result of the leadership at SDSC and in many of the NPACI partners, NPACI has played a critical role in the development of our current cyberinfrastructure, and serves as the critical foundation of resources and human capital assets on which our future cyberinfrastructure depends.
Our opportunity is to capitalize on this community and on our investments in the high-end computing infrastructure created through the PACIs and through our recent investments in the Extensible Terascale Facility, to develop a coherent, integrated programmatic portfolio of activities that realize the promise of our cyberinfrastructure vision ..
- To provide an integrated high end system of hardware, software and services that .
- Enables scientists and engineers to work on advanced research and education problems . that otherwise would not be solvable.
The drivers for our cyberinfrastructure vision are many. This slide lists only a few. I know that the staff at the NPACI and the Alliance partner sites play significant roles in many such activities, clearly demonstrating the promise and value of partnerships. Examples include:
- the Network for Earthquake Engineering Simulation
- the Grid Physics Network (otherwise known as GriPhyN) which is funded by ITR to develop a virtual data toolkit essential to progress in high energy physics "data grids"
- and others including those listed here.
The impact of the Storage Resource Broker (SRB) Data Management Middleware developed by NPACI demonstrates the value of the contributions your work has made to projects around the world, including the National Virtual Observatory, the NIH Biomedical Information Research Network, the UK e-Science effort and dozens more. We seek to build upon and increase support for exactly this type of work in the future.
In partnership with you, our partners in academe - industry and governments and organizations around the world, we want to create a portfolio of integrated cyberinfrastructure activities
1). This integrated cyberinfrastructure builds on shared hardware resources including computational engines, high bandwidth communication networks, large data repositories and digital libraries, and sensors and advanced instruments. Some of this hardware is shared broadly across the science and engineering community (e.g. computational engines, high performance networks like Abilene, etc.) and other hardware may be domain specific (e.g. research instruments and sensors). And, of course, these resources are distributed across the nation and - at least - in the case of networks, data stores, and instruments cross international boundaries.
2). On top of this hardware sits a collection of grid middleware tools and services that make it possible for researchers and educators to discover and use shared resources in the "hardware" layer.
3). On top of these basic software tools and services sits additional software that provides development tools and libraries that span scientific domains. As indicated here, we call both the grid services and the development tools and libraries "Shared Cybertools". I think you have chosen to call such activities "Technology Thrusts".
4).And last, but very important in the grand scheme of things, in support of science and engineering research and education - the raison d'etre for all of these investments, are the domain specific cybertools - which you call Application Thrusts.
Talented people are absolutely essential to the creation and smooth operation of a shared high-end cyberinfrastructure. Collectively, you are developing future generations of cyberinfrastructure while providing maintenance and operations support for current systems and applications. It is not as simple has this diagram would indicate, but for present purposes, this is the best I've come up with.
I know many of you will have read the recently released report of the NSF Advisory Committee on Cyberinfrastructure - aka the Atkins Committee. The Atkins Committee indicated that to achieve the promise of this cyberinfrastructure vision, and to enable the development and the continuing evolution of a state-of-the-art operational cyberinfrastructure, a coordinated set of long-term investments must be made:
- first and foremost in fundamental computer science and engineering research that will help create future generations of cyberinfrastructure
- in the establishment and continuous upgrade of a state-of-the-art operational cyberinfrastructure
- in software activities that create shared cybertools at the basic grid and collaborative tools and service layers and
- in domain specific cybertools ..
all of these investments made in support of discovery, learning and innovation across all fields of science and engineering.
This is the backdrop for our cyberinfrastructure planning. Let's return now to our more recent terascale investments. A series of three coordinated investments made as a result of terascale merit review competitions in FY 2000 and 2001 and an upgrade in FY 2002 have resulted in the creation of the Extensible Teracale Facility.
ETF currently integrates:
- the Terascale Computing System at the Pittsburgh Supercomputing Center - and
- the Distributed Terascale Facility resources here at SDSC, and at NCSA, Caltech and Argonne National Lab.
As you know, these ETF partner resources are now integrated through a 40 GB/s dedicated network connection between extensible hubs in Los Angeles and Chicago.
In FY 2003, the ETF will be enhanced through the connection of new partner resources. NSF released a solicitation last week to do this, and proposals are due June 9th. Resources to be integrated into the existing ETF architecture might include:
- archival repositories
- digital libraries
- computational resources
- sensor networks.
the resulting ETF architecture will look like this
Now what about the partnerships - NPACI and the Alliance? Both are currently funded through September this year (i.e. 2003), at an approximate level of $35M each.
From now until the end of FY 2004, we will work with you, our PACI & Alliance partners in a staged transition. We advanced the first stage last week at NSB. This period involves a final 1-year extension of the existing NPACI cooperative agreement - we plan to bring the 1-year extension to the Board for approval at the May meeting. Note that this additional year of support for the NPACI award is consistent with the recommendations made by the Atkins Committee regarding PACI.
The final year cooperative agreement will continue support of the Technology Thrust activities. It will also continue to support the NPACI resource partners. However, we are strongly encouraging interested NPACI resource partners to consider competing to connect with ETF through the terascale extensions activity I described earlier. FY 2004 will mark the last FY of support for the PACI program.
Support for Domain Specific Cybertools, or what you call Applications Thrust will be supported at a reduced level in FY'04 through the NPACI award. However, NSF-wide support for such activities will be increased in FY'04 when the development of Domain-Specific Cybertools will become an emphasis in the ITR program. Expect to see a solicitation this summer. Those of you currently receiving Application Thrust support through NPACI are strongly encouraged to submit to this competition.
Finally, support for Terascale operations, those being provided by the Terascale Computing System at Pittsburgh, and the Distributed Terascale Facility, will also be provided through the end of FY 2004 until ETF is commissioned.
I cannot emphasize enough - we are not walking away from the considerable physical and human capital assets created with the NPACI investments. The past 6 years have created a number of strong groups across the country. As we move forward to what we plan to be greatly increased funding for cyberinfrastructure, we need to develop new structures that can accommodate and deal with the new world of cyberinfrastructure. We will build on the products and outcomes of these investments as we create a coherent, integrated portfolio of cyberinfrastructure activities. In the ETF activity, NCSA and SDSC are joined by additional partners who will provide complimentary computing-communications-information resources and services in support of today's science and engineering research and education community, and future generations of
We don't have a complete plan in place yet because we need to hear a lot more from all segments of the community. In FY 2005 and further into the future, NSF's cyberinfrastructure planning looks like this. Beginning in October 2004, appropriate PACI-funded resources and the sum of the terascale resources - including those at SDSC, NCSA, PSC, Argonne, Caltech and at our new ETF partner sites - will be allocatable to the broad user community. NSF will support the management and operations of such resources and services at our ETF partner sites for 5 years. As deemed appropriate by NSF's merit review process, management and operations support may include support for advanced development that is often integral to the support of cutting-edge cyberinfrastructure resources and services. Remaining to be worked out is how requests for management and operations might be submitted to NSF for consideration. To a great extent, this will be determined by the governance approach adopted for ETF - an approach in which both our current and future ETF partners have a stake, and in which future ETF users also have a stake. We will hold a workshop in May this year and possibly others later, to explore governance scenarios, both for ETF specifically and for cyberinfrastructure generally.
Now I talked earlier about our plans to develop a coordinated portfolio of cyberinfrastructure activities. Based on an evolving "needs assessment" - developed in concert with the science and engineering community of high-end users - we'll develop what will become an evolving plan to maintain a cyberinfrastructure at the frontier - to maintain technological leadership in cyberinfrastructure.
We'll have to balance the merits of upgrading existing cyberinfrastructure resources against adding new resources at existing or at new sites. Those decisions will be made on a case-by-case basis in consultation with our community of stakeholders.
But, depending on the availability of funds, we do anticipate increased NSF support for
- the development of domain-specific cybertools
- shared cybertools and services
- and of course, the rapidly evolving high-end computing-communications-information hardware resources on which the enterprise rests.
Finally, we do expect to sustain and increase support for Education, Outreach and Training. The NPACI has made outstanding contributions here. I want to use the highly successful EOT program as the nucleus for an expanded effort to train future scientists and engineers to utilize cyberinfrastructure. I also want to use it as a model for EOT in other areas of CISE activity.
This chart describes the impact you had in Houston, reaching 3,000 girls at 18 schools in and around the city. We plan to build on your successes here in an expanded EOT effort in support of cyberinfrastructure. One of the points the Atkins committee is very clear on is that we must invest more in the training of future cyberinfrastructure users. The implementation of a shared high-end cyberinfrastructure is likely to result in a true paradigm shift in the conduct of science and engineering research and education. Broad transformation will result - changing the tools, the methods and the sociology of all fields. The Education, Outreach and Training opportunities afforded by cyberinfrastructure are broad and far-reaching.
This is a work in progress - we are still working through some issues - cyberinfrastructure governance and community building being key, we are mapping out the budgetary implications of our plans and the role of NSF investments in a federal-wide context. As some of you may know, there is increased interest in high-end computing, marked by the establishment of a government-wide Revitalization Task Force on High End Computing.
In wrapping up, let me thank you all for the many contributions you have made. I hope you will continue to work with us to exploit the potential of cyberinfrastructure. You have helped to create the potential and we need your continued help to create new levels of capability.
Let me end with a very important point. I understand and value the kind of community you've built. It is what undergirds and enables all that you do. Community is exceedingly difficult to build and very easy to destroy. I will do everything in my power to help you transform the NPACI and Alliance communities into even stronger cyberinfrastructure communities for the future.