A lot of our data sources are becoming fairly large over a few million rows, and we have noticed refresh speeds getting much slower.
I think the solution is to have data appended as of last import rather than replace the whole thing? But I'm curious to know if anyone else has a better solution?