Results, related to the choice of optimum temporal scale during approximation of maximal size of turn for fractal brownian motion (FBM), are presented. The use of statistical descriptions of traffic on small temporal segments allowed to present practical recommendations for the calculation of probability of appearance of tail of turn. This approach is considered for a traffic with of long duration dependence (LDD). For a FBM-traffic it is rotined with LDD, that over the offered exponential breaking up of temporal scales brings to the optimum productivity of the examined area of network.
FBM-traffic, long duration dependence, protocols of networks, model, queue