HTTP Traffic Model for Web2.0 and Future WebX.0

HTTP Traffic Model for Web2.0 and Future WebX.0

Vladimir Deart (Moscow Technical University of Communications and Informatics, Russia) and Alexander Pilugin (Moscow Technical University of Communications and Informatics, Russia)
DOI: 10.4018/978-1-4666-3902-7.ch004
OnDemand PDF Download:
No Current Special Offers


This article presents a method of estimation for HTTP traffic quality service parameters mean delay and lost packets percentage. This method, based on statistic measurements, includes simulation and analytical modeling. Statistical HTTP traffic models presented earlier take into account typical features of WEB2.0 Internet traffic, which were used for the simulation model. Developed universal simulation models make it possible to research service quality parameters under setting network conditions over a wide range considering Internet development. The presented analytical method based on batch packet arrival model allows an accuracy estimation of mean HTTP-packets delay in Core Router by simple calculations. Objective results of HTTP traffic quality service parameters can be used in QoS standard development for WEB traffic and model QoE standard development.
Chapter Preview

The analytical and simulation models in the article are based on statistic results described in Deart (2009). The measuring complex was presented and based on the open program components and database, which helped for traffic measurements in the provider access network. Deart (2009) showed how these results were processed and how it helped to get the statistical HTTP traffic model.

The developed model is possessed of some advantages in comparison with the model in the work of Shuai (2008). So far as interval approximation are used for a distribution of HTTP response size, HTTP responses are modeled more accuracy. Therefore, it could be shown the lower and tail range of values completely.

The obtained model takes into account the network HTTP traffic WEB2.0 features better by introduction of dependent generation mechanism of TCP sessions, which simulates a user work, and influence on TCP sessions opening of the user browser.

This statistical model is more accurate if it is compared with Mah (1997), Choi (1999), and Padhye (1998) as far as it describes the modern Internet web traffic and ties also amount of HTTP traffic with intensity of TCP sessions. The main parameters of HTTP traffic statistical model in Deart (2009) are opening TCP session intensity, quantity of GET requests in one TCP session, intervals between GET requests, GET request size in bytes and HTTP response size in bytes. Selected parameters are defined unambiguously a HTTP traffic generator according to a session principle and allow to use it in simulations. The modeling network topology is presented in Figure 1.

Figure 1.

Modeling network topology


Complete Chapter List

Search this Book: