University: University of Pittsburgh
Name of Sponsoring Professor: Dr.Vladimir I. Zadorozhny
Department: Department of Information Sciences and Telecommunications, DIST

Project: NEBULA

Project Overview

Wide Area Applications utilize a Wide-Area-Network (WAN) infrastructure e.g., the Internet, to connect a federation of hundreds of servers servicing tens of thousands of clients. The use of a public network for data delivery causes deterioration of service reliability (e.g., in the form of broken links) and performance (e.g., increased delivery time due to points of congestion in the network). Therefore, mediators over wide area applications need to provide, in addition to traditional services such as data integration, services that aim at improved reliability and reduced latency.

In this project we aim at constructing profiles, information regarding servers, network, and client performance. With such profiles at hand, mediators can optimize query performance. Also, such information improves Web caching utilization, thus reducing retrieval latency, a common deficiency in contemporary wide area applications.

Our research is motivated, in part, by the evolution of name resolution services, e.g., the Handle system, an emerging IETF/IRTF standard, and the challenges faced in supporting an information-centric infrastructure. In this project, we will undertake a comprehensive study of the changing behavior of digital resources over time and across different applications, when accessed via the dynamic WAN. Our objective is to develop appropriate resource profiles to characterize this behavior. We will then use these profiles to customize service and information delivery to clients, considering both application semantics and the noisy WAN environment.

My work focuses on an extensive simulation study of the network clustering for Wide Area Applications and the impact of performance monitors which contain the latency profiles, on the network.

Brief info on the experiment conducted