Question
How does Okera scale with varying result sizes (after filtering)?
Answer
We expect cost to scale to be sublinear, so returning twice as many rows should take less than twice as much time. For example, the performance of a join is not impacted by the number of rows resulted from the join, whether it produces 14k rows or 460 million rows.
Note: Please make sure your benchmark query is not writing to the file server as writes slow performance.
Comments
0 comments
Please sign in to leave a comment.