WebComputing: Design and performance.
Item
-
Title
-
WebComputing: Design and performance.
-
Identifier
-
AAI9969746
-
identifier
-
9969746
-
Creator
-
Ying, Kevin M.
-
Contributor
-
Adviser: David Arnow
-
Date
-
2000
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Computer Science
-
Abstract
-
The advent of the Java programming language, with its support for Web-deliverable applets, has created a new, promising parallel computing platform that we call WebComputing. This platform can break through the conventional boundaries imposed by traditional distributed systems, such as DP, PVM, and MPI. The essential idea is that a master server, or collection thereof, in league with a collection of Web servers coordinates the execution of tasks by applets running in parallel on an ever-changing set of unreliable, heterogeneous client machines. The promise of WebComputing is the potential for achieving an unprecedented degree of parallelism.;WebComputing is in its infancy. It faces many challenging questions that are related to Internet communication delays, unreliable execution environments, and system scalability. Research that can answer the following questions is greatly needed. (1) How does WebComputing communication perform at its current stage? (2) How does communication performance affect the WebComputing architecture design and the programming models? (3) What performance enhancements can be made in terms of scalability and load balancing?;This thesis covers three primary areas of research pertaining to each of these questions. First, it presents a performance study of communication protocols for WebComputing. The results of this study can be used as guidelines for forthcoming WebComputing system designs and their application implementations. Based on these guidelines, the thesis presents a unique WebComputing system design and implementation, called the SWC framework. Using layered architecture, this framework provides a high-level coherent API and a robust system implementation, which enables the system to be executed as threads on SMP machines, as processes on a collection of networked workstations, or as Java applets in a WebComputing context. Furthermore, this thesis describes the design and implementation of a simple, yet effective scalability enhancement of the SWC framework that incorporates multiple levels of system-wide load-balancing schemes. The measurements of the scalability enhanced system show the effectiveness of this design. In addition, the enhanced framework also provides a deployment system using Java servlet technology.;Currently, several projects (including Monte Carlo computation and classical Operational Research projects) are underway using the SWC framework. In addition, two Computer Science courses were taught using this framework and students took it easily.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
