NSF Workshop on
New Challenges and Directions for Systems Research
St. Louis, Missouri
July 31 - August 1, 1997

Scope and Goals

The society has seen a rapidly changing and evolving technology and applications of computer and communication systems. This rapid evolution has largely integrated new applications and technologies into existing conceptual frameworks. This trend will not only continue but will place more pressures on the research and development of more revolutionary systems to meet the needs of future computing and communication requirements. The goal of this workshop is to address the need for a new perspective on computer systems research, and search for new paradigms for research and development of computer systems.

The grand challenge problems for systems research will center around an integrated approach to support new applications. Note that our definition of a grand challenge involves an application that is not supported or not adequately supported by today's systems; "quantum leaps" that are needed in each systems-related area; and what each area needs from other areas to make the quantum leap (and thus solve the grand challenge problem). In other words, each group (or research discipline) will state what technological advances it can make to address the grand challenge problems provided the other groups (or research disciplines) can come up with solutions for some specific problems.

Among the goals of the workshop are to:

1. Identify new applications that are not adequately supported by current systems (nor can they be supported by incremental improvements to current systems).
2. Identify enabling technologies for meeting the grand challenges presented by such applications.
   We would like for you to think of technologies needed or technologies that will drive certain future implementations.
3. Identify integrated research programs for solving the grand challenges.
   The research programs should address algorithmic and constructive parts (including testing and/or validation).
4. Identify cross-cutting themes and needs across the various disciplines within systems research.
   The term cross-cutting seems to have different meanings. We would like the following two to be used in connection with this item.
   • Cross cutting need or common needs (e.g., infrastructure, measurement, testbeds)
   • Cross-cutting themes -- what one research discipline needs from another discipline.
5. Identify potential sources of quantum leap from the state-of-the-art in systems research.
   What directions take us from here to there? These quantum leaps are per discipline. That is for your group, what directions offer potential for quantum leaps?
6. Identify research directions that have a potential for high payoffs.
   These need not be quantum leaps, but directions that present opportunities for significant progress. Again these are per discipline.
7. Identify models or paradigms for integrative research.
   We would like for you to think of alternatives to current model of systems -- the layered view, from applications to kernels.
8. Identify research directions that produce only incremental or diminishing payoffs.
   If we are going to advocate new research directions to be funded by the federal government, we should be willing to suggest places for funding cuts.

To be more concrete, we will give some examples for you to consider. However, it is not our intention to constrain your thinking. Feel free to suggest different questions or different definitions

Some Grand Challenges

1. Ability to guarantee end-to-end QoS for an arbitrary application.
   
   
2. Real-time steering of large scale computational models
   
   
3. Uniform access to global information resources.
   
   

Some New Applications

1. Everywhere in the home computing. Should integrate different interfaces (voice, including voice recognition of all family members, touch pads, writing pads, remote control, and traditional interfaces), networking a home (using power lines?), etc.
   
   
2. Computer assisted collaborative work. While there are some examples of such systems, we feel that this is an application that is not adequately supported.
   
   

Some Enabling Technologies

1. Internet with 10 billion IP addresses
   
   
2. Million processor systems
   
   
3. New I/O interfaces
   
   
4. New architectural paradigms (IRAM, Data Scalar?)
   
   
5. Deep integration of communication with computation
   
   
6. New programming paradigms
   
   

Some Research Programs

1. Integration of non-traditional interfaces to computer (voice, handwriting, etc).
   
   
2. Seamless composition of systems
   
   
3. High level API's for customization
   
   
4. Specification of robustness, security, assurance and other non traditional performance measures
   
   

Some Cross Cutting Needs

1. Infrastructure and testbeds
   
   
2. Benchmarks
   
   
3. Performance measures
   
   
4. Provability (or assurance)
   
   
5. Seamless composition and integration
   
   
6. New paradigms (other than kernels, threads, ISA, interprocedural analyses)
   
   

Overview of Workshop Organization

For the second day, the organizers along with the keynote speakers and interested group leaders will identify new groups based on cross cutting themes and needs. The participants will be assigned to these new groups. A set of questions for discussion by the new groups will be formulated. The new group leaders will present their groups' responses to these set of questions. The workshop will conclude with a summarization of recommendations to NSF based on these responses.

Role of Group Leaders

Return to: Workshop Home Page