Normalizing series identity into a separate metadata table and referencing it by a compact ID reduces time-series storage by about forty-two percent in our experiment. Instead of repeating dimension ...