In message <Pine.LNX.4.33.0106071834370.853-100000@uplift.swm.pp.se>, Mikael Ab rahamsson writes:
I attended a seminar where self-similarty people from Ericsson were talking (At IETF INET2001). They had not tested the theory with thousands of TCP connections on high capacity links, but one thing that caught my eye was that for self similarity to occur, congestion need to exist (according to them).
I'd have to see the Ericsson work. And I'm not an expert in self-similarity though I try to keep up with the papers. That said, the work I've seen to date on self-similarity has nothing to do with congestion. It often does have to do with traffic interacting with other traffic. But the queues don't have to be full, or even near full, for interactions to occur. Craig