*> *> Another way of looking at this is to count the number of flows which are *> active before and after a certain point in time. Do you mean, the number N(T) of flows with duration > T? Yes, that distribution function makes sense. Bob Braden