The Wide Area Network (WAN) traffic exhibits very unusual, irregular behavior. This behavior cannot be model with classical methods (Poisson like processes). At this moment we have an impression that the great complexity of network, combined with "competitive" protocol is somehow responsible for this. However, we do not know which kind of stochastic process (or even class of processes) is the appropriate model. To make the thing worse there are no statistical tools that can "calibrate" this apparent irregularity.

In this talk I plan to introduce the problem, convince the audience that the WAN traffic cannot be obtained by cosmetic modifications of existing models. I will comment on the recent trend that associate WAN traffic with "multifractal" processes as well as random cascades. Finally I will try to point on a connection between WAN traffic, stock market, and several other irregular processes.