Are some/most of the operations O(n)? How practical is this database then?
1. Reference data (countries, car makes and models, zip codes, tax rates, etc)
2. “Human scale” databases - ie stuff that a small group of people are manually curating
O(n) would probably be fine for either of those.