HTTP 요구 수준 정보를 이용한 웹 트래픽 모델링 및 분석
- 원문 URL
In this paper, we present characteristics and modeling of web traffic that dominates Internet traffic. Web traffic in existing research has been mainly modeled at packet and application level, but we model the web traffic at HTTP request level to simplify process of modeling with probability distribution through statistical analysis. For this purpose, we used actual traffic data by collecting the web server log and packet trace for the basic traffic information required to model. Item of response size that obviously affects characteristics of whole web traffic has been modeled by pareto distribution only, but we could find that low part of cumulative distribution function of response size that can not be modeled with pareto distribution can be well modeled with exponential distribution. Therefore, response size can be modeled adaptively by combination of pareto distribution and exponential distribution. Also, we present characteristics of web traffic in access network and show whether they follow self-similar process or possion process. As a result of analyzing the characteristics of response size per web server, we can show that characteristics of web traffic are depend on the response size of web server, which heavily influence the web traffic of target network investigated.