In this paper, we characterize wide-area network applications that use the TCP transport protocol. We also describe a new way to model the wide-area traffic generated by a stub network. We believe the traffic model presented here will be useful in studying congestion control, routing algorithms, and other resource management schemes for existing and future networks.Our model is based on trace analysis of TCP/IP widearea internetwork traffic. We collected the TCP/IP packet headers of USC, UCB, and Bellcore networks at the point they connect with their respective regional access networks. We then wrote a handful of programs to analyze the traces. Our model characterizes individual TCP conversations by the distributions of: number of bytes transferred, duration, number of packets transferred, packet size, and packet interarrival time.Our trace analysis shows that both interactive and bulk transfer traffic from all sites reflect a large number of short conversations. Similarly, it shows that a very large percentage of traffic is bidirectional, even for bulk transfer. We observed that interactive applications send significantly different amounts of data in each direction of a conversation, and that interarrival times for interactive applications closely follow a constant plus exponential model. Half of the conversations are directed to a handful of networks, but the other half are directed to hundreds of networks. Many of these observations contradict commonly held beliefs regarding wide-area traffic. This research was supported by an equipment grant from the Charles Lee Powell Foundation. Ramón Cáceres was supported by the NSF and DARPA under Cooperative Agreement NCR-8919038 with CNRI, by AT&T Bell Laboratories, Hitachi, a University of California MICRO grant, and ICSI.
Introduction"The key issue in the design or selection of a congestion management scheme is the traffic pattern, and traffic patterns are dependent upon the application [Jain90]." This paper presents conversation level analysis of wide-area TCP traces collected on two campus networks-University of Southern California (USC) and University of California, Berkeley (UCB), and one industrial research site-Bellcore. Most of the analysis was done as part of term projects for graduate courses in performance evaluation and distributed systems at the University of Southern California. Our goal was to collect information that would be useful in evaluating future network designs. Since TCP packets make up roughly 80% of all wide-area network traffic, 1 a model based on TCP traffic is necessary to study network behavior. We restrict our discussion to TCP in this paper. Table 1 summarizes our most important results.When simulating new congestion, flow control, and routing algorithms one needs to model the overall pattern of traffic flowing through the network, from distribution of packet sizes and interarrival times to characteristics such as distribution of host reference patterns and direction of traffic flow. Current practice is to use FTP and TELNET so...