Revisiting the Rellich inequality