What you are looking for is a dolly: apply the same translation to both camera location and camera target.