The up vector defines one of the axes of the camera coordinate system. The other one being the vector from camera position to target position. The third, which points left or right, is trivially calculated by crossing the two. By choosing an up-vector of (0,1,0) your camera is aligned with world up (+y). By choosing e.g. (0,-1,0) things would be upside down. Without an up vector the camera orientation could not be calculated.
But also read the 3D math intro in the docs and it's hints concerning tranforms and gimbal lock.