opt: improve per-ON condition cost adjustment for lookup join #39016
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Issue cockroachdb#34810 tracks taking into account the internal row count of lookup joins. We currently have a hack in place to always prefer looking indexes that constrain more columns. We encountered a case where this adjustment doesn't work: when the estimated row count is very very small (which happens when there are a lot of conditions), the per-row cost adjustment ends up not making a difference (this is because of limited floating point precision, and the "tolerance" built into `Cost.Less()`). To address this, we also add a constant per-ON condition cost which isn't scaled by the row count. Release note (bug fix): Fixed bug in the optimizer causing a bad index for lookup join in some cases.
|
This change is |
|
awesome that you were able to pick this up! |
justinj
reviewed
Jul 22, 2019
There was a problem hiding this comment.
Reviewable status:
complete! 1 of 0 LGTMs obtained (waiting on @justinj and @rytaft)
|
I don't think #38800 is directly related. If I remove the |
rytaft
approved these changes
Jul 22, 2019
There was a problem hiding this comment.
Reviewed 8 of 8 files at r1.
Reviewable status:
|
bors r+ |
craig bot
pushed a commit
that referenced
this pull request
Jul 22, 2019
39016: opt: improve per-ON condition cost adjustment for lookup join r=RaduBerinde a=RaduBerinde Issue #34810 tracks taking into account the internal row count of lookup joins. We currently have a hack in place to always prefer looking indexes that constrain more columns (see #35587). We encountered a case where this adjustment doesn't work: when the estimated row count is very very small (which happens when there are a lot of conditions), the per-row cost adjustment ends up not making a difference (this is because of limited floating point precision, and the "tolerance" built into `Cost.Less()`). To address this, we also add a constant per-ON condition cost which isn't scaled by the row count. Release note (bug fix): Fixed bug in the optimizer causing a bad index for lookup join in some cases. Co-authored-by: Radu Berinde <[email protected]>
Build succeeded |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Issue #34810 tracks taking into account the internal row count of
lookup joins. We currently have a hack in place to always prefer
looking indexes that constrain more columns (see #35587). We encountered a case
where this adjustment doesn't work: when the estimated row count is
very very small (which happens when there are a lot of conditions),
the per-row cost adjustment ends up not making a difference (this is
because of limited floating point precision, and the "tolerance" built
into
Cost.Less()). To address this, we also add a constant per-ONcondition cost which isn't scaled by the row count.
Release note (bug fix): Fixed bug in the optimizer causing a bad index
for lookup join in some cases.