On the Self-similar Nature of Ethernet Traffic (Extended Version)-英文文献.pdf

IEEWACM TRANSACTIONS ON NETWORKING, VOL. 2, NO. 1,FEBRUARY 1994 1 On the Self-similar Nature of Ethernet Traffic (Extended Version) Will E. Leland,Member, IEEE, Murad S . Taqqu,Member, IEEE, Walter Willinger, and Daniel V. Wilson, Member, IEEE Abstract-We demonstrate that Ethernet LAN traffic is statis- traffic measurements presented in [141.Moreover, we illustrate tically self-similar,that none of the commonly used traffic models some of the most striking differences between self-similar is able to capture this fractal-like behavior, that such behavior models and the standard models for packet traffic currently has serious implications for the design, control, and analysis of high-speed, cell-based networks, and that aggregating streams of considered in the literature. For example, our analysis of the such traffic typically intensifies the self-similarity (“burstiness”) Ethemet data shows that the generally accepted argument for instead of smoothing it. Our conclusions are supported by a the “Poisson-like’’ nature of aggregate traffic, namely, that rigorous statistical analysis of hundredsof millions of high quality aggregate traffic becomes smoother (less bursty) as the number Ethernet traffic measurements collected between 1989 and 1992, of traffic sources increases, has very little to do with reality. coupled with a discussion of the underlying mathematical and statistical properties of self-similarity and their relationship with In fact, using the degree of self-similarity (which typically actual network behavior. We also present traffic models based dep