This is intriguing; could you give an example of a distribution Postgres struggles with at scale, and an application that produces such a distribution?
It seems that power law distributions over multiple tables aren’t handled very well: https://stackoverflow.com/questions/65861780/is-there-a-way-...