First years of Cluster data analysis have started to justify large expectations of community by providing a new level of understanding of space plasma dynamics. We briefly summarize the state of art in two aspects, concerning the tail current sheet (CS) dynamics, and its structure. Both issues are central for understanding of the tail stability (including substorms) and in both cases our view drastically changed from previous picture. With greatly advanced capabilities of measuring gradients and propagation Cluster made clear that previous picture of smooth and almost planar sheets is rarely applied. Real sheet dynamics often includes large-amplitude meso-scale sheet corrugations in which the sheet (1) is strongly deformed and tilted, (2)the perturbation has a kink-like properties (with CS normal rotating in YZ plane) and (3) propagates systematically from central tail part toward its flanks. The formation/propagation mechanisms are under the study although many facts point out to their close relationship to substorm activations and BBF generation. On the other hand, fast current sheet crossings were systematically exploited to investigate the sheet structure. Whereas the thin embedded as well as bifurcated sheet patterns were shown to occur often, the major finding is that, unlike the pre-Cluster era in which the Harris-type model was exclusively used, we are actually in a world of essentially non-Harris and variable plasma distributions. This change requires a big effort in theoretical studies, both findings stimulated a burst of activity in the plasma theory of space current sheets.