Well, without more precision about the problem, I would answer that I see no theoretical concern with having higher variance than mean, especially knowing they are not in the same unit so difficult to compare... Neither with having higher standard deviation than mean, by the way.
If the zero point has no special meaning in your application area then, like Emmanuel, I wouldn't be concerned that the mean is smaller than the standard deviation (an easier comparison). However, if it does have special meaning, e.g. positive values are valid and negative values are invalid, then you might want to apply a log transformation or the like to any variables whose negative values would be invalid.