Rapidminer - importing tables that need to be joined

Question

I have some database tables that I want to use in RapidMiner. Is there going to be a performance difference between writing sql to join the tables or importing the tables separately and performing a join after?

Answer 1

You'd have to measure it in your case since it depends on the amount of data you have and the complexity of the query.

In my experience, as the tables get larger and the joins more complex, I would tend to use SQL because databases give you more tools to tune the queries with indexes and so on. The database itself can be deployed on a separate physical entity so you can take advantage of the processor and memory there.