Improve CBO estimates for correlated columns#11324
Merged
sopel39 merged 5 commits intotrinodb:masterfrom Mar 11, 2022
Merged
Conversation
Member
Author
|
TPC benchmark results with SF1000 ORC |
sopel39
approved these changes
Mar 4, 2022
Member
|
prefix already approved in #11066 |
raunaqmorarka
force-pushed
the
cbo-correlation
branch
from
d85c0a2 to
0476291
Compare
raunaqmorarka
force-pushed
the
cbo-correlation
branch
2 times, most recently
from
613370d to
eb33a4a
Compare
Member
Is it about correlation in a sense of query outer scope references, or ... ? |
Member
Author
This is about correlation in the data between columns (e.g. nation, city) |
Member
Author
|
Results with latest PR |
sopel39
approved these changes
Mar 10, 2022
Member
sopel39
left a comment
sopel39
left a comment
There was a problem hiding this comment.
small comments. Did tests changed?
sopel39
reviewed
Mar 10, 2022
Currently we assume that there is no correlation between the terms of a filter conjunction. This can result in underestimation as there is often some correlation between columns in real data sets. In particular, predicates inferred on the build side relation through a join with a partitioned table are often correlated with user provided predicates on the build side. Estimation for filter conjunctions now applies an exponential decay to the selectivity of each successive term to reduce chances of under estimation. optimizer.filter-conjunction-independence-factor is added to allow tuning the strength of the independence assumption.
Currently we assume that there is perfect correlation between the clauses of a join and use the most selective clause for driving output row count estimation. This can result in overestimation as it not necessary that columns in join keys are perfectly correlated in real data sets. Estimation for multi clause joins now applies an exponential decay to the selectivity of each successive term to reduce chances of over estimation. optimizer.join-multi-clause-independence-factor is added to allow tuning the strength of the independence assumption.
raunaqmorarka
force-pushed
the
cbo-correlation
branch
from
eb33a4a to
44c51cb
Compare
sopel39
approved these changes
Mar 10, 2022
| List<Symbol> expressionSymbols = expressionUniqueSymbols.get(term); | ||
| int expressionPartitionId; | ||
| if (expressionSymbols.isEmpty()) { | ||
| expressionPartitionId = symbolPartitions.size(); // For expressions with no symbols |
sopel39
reviewed
Mar 11, 2022
Closed
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
Description
Overall goal of the PR is to work towards enabling optimizer.default-filter-factor-enabled by default.
If default-filter-factor is enabled with existing implementation, it improves q18 and q21 on tpch significantly.
However, it also results in regressions on certain benchmark queries (tpcds partitioned q64, tpcds unpartitioned q78).
These changes update the estimation logic of filters and joins to address the problems
with underestimation of filter conjunctions and overestimation of multi-clause joins observed
when default-filter-factor is enabled with existing implementation.
Improvement
Query optimizer
Improves CBO estimates in the presence of hard to estimate terms.
Related issues, pull requests, and links
Picks first (n-1) commits from #11066
Documentation
( ) No documentation is needed.
(x) Sufficient documentation is included in this PR.
( ) Documentation PR is available with #prnumber.
( ) Documentation issue #issuenumber is filed, and can be handled later.
Release notes
( ) No release notes entries required.
(x) Release notes entries required with the following suggested text: