Exploratory Visualization of Distributed Computations

The decreasing costs of computer hardware has made distributed computing
commonplace.  Distributed computing makes some applications feasible, improves
the performance of others, and offers the potential of increased availability
and reliability through redundancy. Distribution also creates problems in
understanding and interacting with these computations.  Multiple threads of
control, non-deterministic message delays, and the potential of partial
failures are all characteristics of distributed systems that hinder
understanding, of designers and users, and pose difficulties for monitoring
tools.  These problems are exacerbated in distributed computations that are
long-lived and large, both in the number of processes and the amount of data
handled.

Visualization can help alleviate these difficulties through interactive
exploratory tools that allow online investigation of a distributed computation
at a variety of levels of abstraction.  In this talk, I describe my research in
developing techniques to support monitoring and steering of distributed
computations.





Delbert Hart is completing his D.Sc. in Computer Science at Washington
University in St. Louis under advisors Dr. Gruia-Catalin Roman and Dr. Eileen
T. Kraemer.  His thesis investigates techniques of visualizing large,
long-running distributed computations.  His current research interests include
distributed systems, visualization, and software agents.

While working on his masters at Washington University Delbert worked on the
Pavane declarative visualization project.  He received his masters from
Washington University in May 1995, and a B.A. in Computer Science from the
State University of New York at Plattsburgh in May 1993.