Date of Award

Winter 2000

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Science

Committee Director

C. Michael Overstreet

Committee Director

Kurt J. Maly

Committee Member

Richard E. Nance

Committee Member

Ravi Mukkamala

Committee Member

R. Bowen Loftin

Abstract

Technology advances are providing greater capabilities for most distributed computing environments. However, the advances in capabilities are paralleled by progressively increasing amounts of system complexity. In many instances, this complexity can lead to a lack of understanding regarding bottlenecks in run-time performance of distributed applications. This is especially true in the domain of distributed simulations where a myriad of enabling technologies are used as building blocks to provide large-scale, geographically disperse, dynamic virtual worlds. Persons responsible for the design, configuration, and control of distributed simulations need to understand the impact of decisions made regarding the allocation and use of the logical and physical resources that comprise a distributed simulation environment and how they effect run-time performance. Distributed Interactive Simulation (DIS) and High Level Architecture (HLA) simulation applications historically provide some of the most demanding distributed computing environments in terms of performance, and as such have a justified need for performance information sufficient to support decision-makers trying to improve system behavior.

This research addresses two fundamental questions: (1) Is there an analysis framework suitable for characterizing DIS and HLA simulation performance? and (2) what kind of mechanism can be used to adequately monitor, measure, and collect performance data to support different performance analysis objectives for DIS and HLA simulations? This thesis presents a unified, architectural framework for DIS and HLA simulations, provides details on a performance monitoring system, and shows its effectiveness through a series of use cases that include practical applications of the framework to support real-world U.S. Department of Defense (DoD) programs. The thesis also discusses the robustness of the constructed framework and its applicability to performance analysis of more general distributed computing applications.

DOI

10.25777/pt3w-4242

ISBN

9780493079059

Share

COinS