[EMAIL PROTECTED] (Robert Dodier) wrote in message news:<[EMAIL PROTECTED]>... > [EMAIL PROTECTED] (Faisal) wrote: > > > when we talk of Bursty traffic over the internet. Then is it correct > > that the bursts which occurs over a period of time resembles all prior > > bursts that why we say that the bursty traffic is self similar > > traffic. Correct me if I am wrong > > In general self-similarity means that if you look at different > scales of measurement, you'll see the same sort of structure. > Sometimes there is exact self-similarity, but in the case of > network traffic, mountains, waves, etc., it's self-similarity > "on the average". > > Specifically self-similar network traffic would mean that if you > look at a burst of traffic, it will have smaller bursts and > calm intervals within it, and if you look at a smaller burst > it will have milli-bursts and micro-bursts, and so on, ad infinitum. > > A real network signal cannot have such structure at all scales; > usually there is some range of scales over which self-similarity > more or less works, and at the shortest and longest scales, > the self-similar model isn't really a good description. > > B. Mandelbrot's book "Fractal Geometry of Nature" can help you > get a feel for the intuitive ideas. > > For what it's worth, > Robert Dodier Dear Robert,
Thanks for the detailed reply. It was really helpful. There is one more question. Please correct if I am wrong "we use long tailed distribution on bursty traffic because the average of bursts occuring is infinite" Thanks Faisal . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
