An Initiative to Transform DOE and Its National Labs, and Set the Standard for R&D Organizations into the 21st Century
I. Executive Summary
DOE2000 will provide tools that qualitatively improve the ability of DOE to accomplish its missions and set the standard for scientific R & D organizations in the 21st century. These tools will eliminate physical distance and organizational structure as limits to collaboration. They will lead to a system-wide culture change in sharing resources and will engender superior ways of doing science, engineering, research collaboration and management. By enabling collective use of resources far in excess of those available in a single facility, significant cost savings will also be realized.
The DOE2000 effort builds on the DOE's strengths in high-performance computing and advanced computational science to enable DOE to perform stockpile stewardship and energy research in the next century. The initiative described here has been started by the Offices of Defense Programs (DP) and Energy Research (ER) and has already been joined by the Office of Energy Efficiency and Renewable Energy (EE). The impetus was a charge from the Under Secretary of Energy to make more formal the already-successful collaboration between DP and ER. The final goal is to expand this collaboration across all offices, divisions and programs of DOE to make a more tightly coupled, integrated, efficient, and flexible Department. This will be accomplished by enabling easy and efficient collaboration and resource sharing Department-wide.
A. Why DOE2000 Is Necessary
The Department faces new challenges in reevaluating its missions and the methods used to achieve them. Science-based stockpile stewardship is central to DOE's mission; advanced, effective approaches to environmental analysis and remediation are required; many DOE facilities provide unique resources to the scientific and engineering communities; DOE is a key participant in both High Performance Computing and Global Change interagency initiatives. All this is needed while anticipated budgets continue to decrease.
Advanced computational simulations will be required to address many of these mission requirements. Full-scale nuclear weapons tests are no longer possible and will be replaced by computer simulation of reliability and accident scenarios; field-scale validation of long-term environmental remediation strategies is prohibitively expensive and politically difficult; the goal of the Global Change initiative is to anticipate the ecological effects of human activity on the global climate. The solution of these problems will require an integrated software framework that leverages the work and expertise of all scientists working on these and similar problems.
These challenges require new, creative solutions in information science and technology and in advanced computational science. The Galvin Report has encouraged the Department to follow the lead of businesses world-wide in achieving new levels of organizational efficiency. While many of the technologies required will be off-the-shelf, DOE must take the lead in those areas in which the Department's needs are unique or in which commercial solutions will not be available in a timely fashion. For example, the Energy Sciences Network has been a pioneer in supplying high-speed connectivity in support of DOE-specific applications. Initial sharing experiments in combining separated supercomputers in a single application, and in remote access to unique instrumentation have been successful in both facility access and personnel collaboration.
B. What DOE2000 Will Accomplish
DOE2000 includes the development of several classes of technology to pursue collaborative, shared-use solutions to the Department's challenges. A high-performance network infrastructure, built on the currently successful ESnet, will deliver data, video and audio, and will support advanced applications such as new ways of viewing experimental and computational results (immersive visualization). Security and authentication are required for classified applications, but are also needed for safe remote use of expensive instrumentation, for collaborations requiring protection of proprietary information, and for preventing disruptions to ongoing Departmental collaborations. Mechanisms to support collaborative work include network-wide file and database facilities, resource locators that work at system-wide scales, reliable "telepresence" software and hardware, shared instrumentation and multiple-supercomputer software, and support for shared, immersive visualization applications. A few of these technologies are available in the marketplace, some others have been prototyped, and others remain to be developed in DOE's laboratories.
However, these technologies must be developed and deployed in close collaboration with critical DOE applications. Direct involvement with these technologies will be through two pilot projects, developed by cooperating programs. The first, the Materials Micro-Characterization Collaboratory, will electronically integrate resources from the DOE national laboratory system and university participants to produce a distributed world-class materials lab. This facility would literally cost billions of dollars to construct at a single location. The second, the Diesel Combustion Collaboratory, overlaps a number of DOE mission-critical capabilities, including stockpile stewardship, global change, and environmental remediation (computational fluid dynamics, CFD, is critical to all of these areas). These pilots will provide the applications pull and will serve to strengthen and insure robustness of the computational and information technologies.
C. DOE2000 Began in Defense Programs and Energy Research
The Offices of DP and ER were partners in beginning the DOE2000 Project design. Both have long and successful records at cooperative programs in the required technologies, including advanced modeling and simulation, computer operating systems, high-performance storage systems, fast networks, and standards for communication. In recognition of these collaborations, DOE management has recommended that a more formal mechanism would be valuable in achieving an integrated program in scientific computing and information infrastructure. Both offices have developed high-performance computing and related infrastructures, driven by needs in basic research, developing new energy sources, and extraordinary computational demands of modern weapons programs. The DP and ER applications use similar approaches in computational physics and chemistry (albeit with different parameters) to solve relevant problems; consequently, advancing the effectiveness of these methods is of interest to both programs. These factors have given DP and ER the impetus to design DOE2000, and pave the way for partners to join in its development. The DOE2000 effort has now been joined by Energy Efficiency, and it can be hoped that it will spread across the Department as the research programs and pilot projects demonstrate its usefulness.
D. Launching DOE2000
In FY97, DOE2000 will launch research and development in the relevant technological areas. Specifically, this will focus on collaborative tools, information surety, tools for complex geometries, advanced numerical techniques, data management tools, and an advanced software framework. This R & D component is essential to achieving the goals of DOE2000 and to establishing the Department as the leader in collaborative computational science and engineering.
E. Additional Partnerships Are Essential
While the technology R & D will begin under DOE2000, we cannot accomplish the culture-changing goals without direct involvement of scientists and engineers, beginning with the two pilots described above. They will demonstrate a new level of achievement on the part of these communities and serve to motivate other communities to become involved. The ultimate success of the DOE2000 initiative requires participation by a substantial number of the DOE communities.
Effective partnerships of two kinds are needed to make DOE2000 successful. First, the applications that use these technologies must be developed. The refinements produced in the stress of production use will perfect the building blocks to the point of general applicability. Applications are also the mechanisms for delivering the results of DOE2000 to the DOE workplaces where they will make the differences promised. The DOE2000 initiative welcomes application pilot-project partners in all DOE Offices, Divisions, and Laboratories.
Second, DOE2000 must have Department management as its full partner in recognition that, while we have the technology roadmap in place, only a Department-wide commitment of resources can make this plan a reality.