Ok, I'll try to answer this one. Although my current work is in vibration and control, my degrees are in Aero. The bottom line is that a rough surface decreases drag only on cylindrical or spherical surfaces (like a golf ball).
There are two components to drag, one is pressure or form drag, which is caused by differences in pressures normal (perpendicular) to a body (this is the same property that creates lift). The other is skin friction drag which is a result of the tangential forces on the body. As you would imagine, skin friction drag is smaller for a smooth surface. But form drag is much greater than skin friction drag for cylindrical objects.
In an inviscid flow (no viscosity), the pressure on the windward side and leeward side of a cylinder are equal, thus no drag. But in the real world there is viscosity, which cause the boundary layer to separate from the surface of the cylinder very near the transition from windward to leeward sides. When the boundary layer separates, the pressure is lost on the leeward side which creates lots of drag. So the object is to delay the separation of the boundary layer. It turns out that if the boundary layer is turbulent, it will stay attached longer than a laminar (smooth) boundary layer. So the solution is to make the surface rough to create that turbulent boundary layer. The reason airplane (and racecar) surfaces are smooth is that in the case of a streamline body, the skin friction drag is the larger component in the total drag.
The idea of using pinholes in a surface to suck air into the body is again a way to delay boundary layer separation, thus reduce form drag.
There, I hope I didn't confuse everybody.
