> I think maybe one reason people shy away from things like Presto (and the above) is the uneven performance guarantees; waiting for an unoptimized Hadoop or Or

> I think maybe one reason people shy away from things like Presto (and the above) is the uneven performance guarantees; waiting for an unoptimized Hadoop or Orcfile query by accident because you joined on something or another is fine for one-offs, but might become costly in prod workflows. Right, so my question is: how is that solved with Hydra? Seems like you'd arrive at the same issue?