neither. So I think in every large-scale app, We have to use this model. Am I right?
It's not about large, but whether it's clustered or not. Then again, the JOIN being talked of there was about the JOIN that the DBMS provides by default. Now of course one DBMS can't automatically join its table with another table that is being managed by another DBMS running on another machine altogether. That's where approaches like map reduce come in. They are aggregations for clustered data.
Обсуждают сегодня