Semester

Fall

Date of Graduation

2005

Document Type

Thesis

Degree Type

MS

College

Statler College of Engineering and Mineral Resources

Department

Lane Department of Computer Science and Electrical Engineering

Committee Chair

Katerina Goseva-Popstojanova.

Abstract

Realistic and formal mathematical description of web-server workload forms a fundamental step in the design of synthetic workload generators, capacity planning and accurate predictions of performance measures. In this thesis we perform detailed empirical analysis of the web workload by analyzing access logs of nine web-servers. Unlike most previous work that focused on request-based workload characterization, we analyze both request and session characteristics. We perform rigorous statistical analysis to determine the self-similarity of web traffic and heavy-tailedness of the distribution of different session parameters. Our analysis shows that web traffic is self-similar and the degree of self-similarity is proportional to the workload intensity. To increase the confidence in our analysis we use several methods for estimating the degree of self-similarity and heavy-tailedness. Additionally we point out specific problems associated with these methods. Finally, we analyze the impact of robots sessions on the heavy-tailedness of the distribution.

Share

COinS