The concept of the data mesh makes sense, but I'm not sure what it means in practice? You have one big redshift and a catalog that says "this team owns this dataset", and that team does their own ETL? Likewise teams own kafka topics, etc.?