wankunde opened a new pull request, #42488:
URL: https://github.com/apache/spark/pull/42488
<!--
Thanks for sending a pull request! Here are some tips for you:
1. If this is your first time, please read our contributor guidelines:
https://spark.apache.org/contributing.html
2. Ensure you have added or run the appropriate tests for your PR:
https://spark.apache.org/developer-tools.html
3. If the PR is unfinished, add '[WIP]' in your PR title, e.g.,
'[WIP][SPARK-XXXX] Your PR title ...'.
4. Be sure to keep the PR description updated to reflect all changes.
5. Please write your PR title to summarize what this PR proposes.
6. If possible, provide a concise example to reproduce the issue for a
faster review.
7. If you want to add a new configuration, please read the guideline first
for naming configurations in
'core/src/main/scala/org/apache/spark/internal/config/ConfigEntry.scala'.
8. If you want to add or modify an error type or message, please read the
guideline first in
'core/src/main/resources/error/README.md'.
-->
### What changes were proposed in this pull request?
In each partition, SortMergeJoin will compute one by one from the streamed
side, so we could respect the streamed side ordering to remove unnecessary sort.
For example, when REQUIRE_ALL_CLUSTER_KEYS_FOR_CO_PARTITION is false:
```sql
SELECT *
FROM (
SELECT t1.*, row_number() over(partition by a order by b) rn
FROM values(1, 1) t1(a, b)
) t1
JOIN values(1) t2(c)
ON a = c
JOIN values(1, 1) t3(d, e)
ON a = d
AND b = e
```
Plan:
```
AdaptiveSparkPlan isFinalPlan=false
+- SortMergeJoin [a#220, b#221], [d#223, e#224], Inner
:- Sort [a#220 ASC NULLS FIRST, b#221 ASC NULLS FIRST], false, 0
: +- SortMergeJoin [a#220], [c#222], Inner
: :- Window [row_number() windowspecdefinition(a#220, b#221 ASC NULLS
FIRST, specifiedwindowframe(RowFrame, unboundedpreceding$(), currentrow$())) AS
rn#218], [a#220], [b#221 ASC NULLS FIRST]
: : +- Sort [a#220 ASC NULLS FIRST, b#221 ASC NULLS FIRST], false, 0
: : +- Exchange hashpartitioning(a#220, 5), ENSURE_REQUIREMENTS,
[plan_id=93]
: : +- LocalTableScan [a#220, b#221]
: +- Sort [c#222 ASC NULLS FIRST], false, 0
: +- Exchange hashpartitioning(c#222, 5), ENSURE_REQUIREMENTS,
[plan_id=98]
: +- LocalTableScan [c#222]
+- Sort [d#223 ASC NULLS FIRST, e#224 ASC NULLS FIRST], false, 0
+- Exchange hashpartitioning(d#223, 5), ENSURE_REQUIREMENTS,
[plan_id=104]
+- LocalTableScan [d#223, e#224]
```
The second **Sort [a#220 ASC NULLS FIRST, b#221 ASC NULLS FIRST], false, 0**
in the plan could be removed
### Why are the changes needed?
Remove unnecessary sort to improve SortMergeJoin performance.
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
Added UT
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
