Normalization makes more sense to me now. At first, I was worried about not being able to identify the flow being removed because the normalized flow looked different. But, based on the dl_type, I can determine which fields are unimportant and which wildcards should be set.