however, for very large datasets, neither one will scale, which is why https://github.com/authzed/spicedb/issues/207 is being worked on