The paper by Will E. Leland and colleagues demonstrates that Ethernet local area network (LAN) traffic exhibits statistically self-similar behavior, which is not captured by commonly used traffic models. This self-similarity is characterized by the absence of a natural length for "bursts" of traffic, with similar-looking bursts evident at various time scales from milliseconds to hours. The degree of self-similarity, measured by the Hurst parameter, depends on the utilization level of the Ethernet and can be used to quantify the "burstiness" of traffic. The authors support their findings with a rigorous statistical analysis of high-quality Ethernet traffic measurements collected between 1989 and 1992, and discuss the implications for congestion control in high-bandwidth networks. They also present traffic models based on self-similar stochastic processes that are simple, accurate, and realistic for aggregate traffic. The paper highlights the differences between self-similar models and standard packet traffic models, and emphasizes the need for new modeling approaches to understand and manage broadband network traffic.The paper by Will E. Leland and colleagues demonstrates that Ethernet local area network (LAN) traffic exhibits statistically self-similar behavior, which is not captured by commonly used traffic models. This self-similarity is characterized by the absence of a natural length for "bursts" of traffic, with similar-looking bursts evident at various time scales from milliseconds to hours. The degree of self-similarity, measured by the Hurst parameter, depends on the utilization level of the Ethernet and can be used to quantify the "burstiness" of traffic. The authors support their findings with a rigorous statistical analysis of high-quality Ethernet traffic measurements collected between 1989 and 1992, and discuss the implications for congestion control in high-bandwidth networks. They also present traffic models based on self-similar stochastic processes that are simple, accurate, and realistic for aggregate traffic. The paper highlights the differences between self-similar models and standard packet traffic models, and emphasizes the need for new modeling approaches to understand and manage broadband network traffic.