Investigating Distribution of Data of HTTP Traffic: An Empirical Study

Hypertext Transfer Protocol is the main protocol used by internet traffic. Therefore, it becomes imperative to understand the statistical characteristics of this protocol. Such understanding is vital for designing better traffic patterns models. This paper seeks to conduct an experiment for the better understanding of the phenomenon. It also endeavors to find ways to improve the performance of internet. The experiment spans almost 34,000 popular websites and collects metadata information from their webpages. This data was then used for studying the distribution patterns. The paper enumerates the various conclusions drawn from the data. One of conclusions was that approximately 5.7% of the data transferred from web server to clients was due to spacer objects. Stale link referencing also accounted for a part of the data traffic. The paper seeks to find the ways to minimize such wastage of bandwidth. This wastage does not occur because of overhead. The simplest method suggested for taking care of this wastage is to make certain additions to the HTML standard. Automating the task of the removal of stale links is also helpful in taking care of wastage of bandwidth. The paper also stresses the need for flexible solutions. It states that internet usage is increasing at lightning speed and therefore, new solutions must be flexible enough to incorporate increased traffic.