home · mobile · calendar · colloquia · 1999-2000 · 

Colloquium - Barford

Understanding End-to-End Performance of Wide Area Services
Boston University

The performance of wide area services running over the Internet -- like the World Wide Web -- can be highly variable. Users are often frustrated by long delays in accessing Web pages, yet the root causes of delays in the Web are not easy to discover. In this talk I will describe the difficulties in pinpointing delays in the Web, and a project that addresses these difficulties. The Wide Area Web Measurement (WAWM) project is developing a laboratory for studying and analyzing the performance of the Web, using synthetic workload generation and an infrastructure distributed across the Internet. A key component of this project is the accurate generation of representative Web workloads. I will describe our workload generator, Surge, which accurately reproduces seven different Web server access characteristics.

The WAWM infrastructure provides an environment for controlled study of representative Web transactions, but the environment alone is not sufficient for identifying the precise causes of delays. To address this problem, I will describe a method for pinpointing the sources of delay in applications like the Web which uses the technique of critical path analysis. By profiling the critical path of a Web transaction, one can unambiguously assign delays to various system components. I will present the results of critical path analysis for Web transactions under a variety of server and network conditions. The results show that critical path analysis can shed considerable light on the causes of delays, and can expose subtleties in the behavior of the entire end-to-end system.

Hosted by Dirk Grunwald.

Department of Computer Science
University of Colorado Boulder
Boulder, CO 80309-0430 USA
May 5, 2012 (14:13)