There is nonzero overhead for doing so: optimizing for duplicate strings invariably adds cost to handling unique strings. This sounds like something you could

There is nonzero overhead for doing so: optimizing for duplicate strings invariably adds cost to handling unique strings. This sounds like something you could do at the schema and application level.