The impact of typical applications on network traffic

Over the past few years we have witnessed evolutionary changes to our communication network as the result of new application, computation and communications paradigms. As such, our understanding to the network becomes diminish and it becomes a key challenge to get a thorough understanding for the network applications and their traffic characteristics. While network traffic characterization and typical application analyses, e.g., peer-to-peer (P2P), have been studied extensively due to its importance in network scheduling and throughput, the relation between typical applications and overall traffic still remains elusive. In this paper, we profile network traffic based on the massive data collected in year 2012 with a passive network traffic monitoring system (TMS) placed on a large Internet service provider (ISP) in China. We also provide an insight to the overall traffic distribution and impact of the major traffic contributors on the overall traffic. Our study reveals that the top five applications, i.e., hyper text transport protocol (HTTP), PPStream, HTTPFlash, QQLive and Thunder, contribute more than 60% of the total traffic, and different application has different correlation with the total traffic. In addition, we give a corresponding coefficient function of these five applications to predict the overall network traffic on weekdays and weekends.