For datasets and result, I'd try the smallest workable version first and only add complexity if it actually improves the result. That should make the trade-off a lot clearer. The part about "A practical understanding of Datasets starts with knowing what the system does well, where it is...