Some remarks on optimal interpolation