Distributed-Computing: What Does it Compute?

University Of California-Los Angeles, Los Angeles CA

Investigators

Abstract

In this project, the PI, Prof. Eli Gafni, proposes to develop guidelines, adapted from the PI's research in area of Theoretical Distributed-Computing, and apply them as guidelines to the area of Distributed-Systems Design. If successful, building distributed systems can become less of an ad-hoc process. It will push the designer to answer questions like ``what-if'' and ``why-not'' akin, but not as rudimentary, to the success of check-lists in the practice of medicine. The PI will collaborate with practitioners to test and tweak the development of these guidelines, as well as collaborate with colleagues of teach software/system-design to test such guidelines in class. This project will also advance workforce development through the training of graduate students, as well as through education activities for undergraduates and graduate students. The project is based on the realization within the Theoretical Distributed-Computing community that ``one-shot tasks are to distributed-computing what functions are to centralized-computing.'' This has contributed to the PI's research immensely by pushing the PI to ask the right questions. The PI feels, as already demonstrated on small scale, that pushing a designer to identify the core tasks the system solved, will not be unlike pushing a centralized system designer to realize that the system designed solves an NP-complete problem. Such realization pushes a designer to expose hidden undocumented assumptions, and expose more avenues the design might have taken, that need to be justified if they are to be abandoned.