imback82 opened a new pull request #28123: [SPARK-31350][SQL] Coalesce bucketed tables for join if applicable URL: https://github.com/apache/spark/pull/28123 <!-- 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'. --> ### What changes were proposed in this pull request? <!-- Please clarify what changes you are proposing. The purpose of this section is to outline the changes and how this PR fixes the issue. If possible, please consider writing useful notes for better and faster reviews in your PR. See the examples below. 1. If you refactor some codes with changing classes, showing the class hierarchy will help reviewers. 2. If you fix some SQL features, you can provide some references of other DBMSes. 3. If there is design documentation, please add the link. 4. If there is a discussion in the mailing list, please add the link. --> When two bucketed tables with different number of buckets are joined, it can introduce a full shuffle: ``` spark.conf.set("spark.sql.autoBroadcastJoinThreshold", "0") val df1 = (0 until 20).map(i => (i % 5, i % 13, i.toString)).toDF("i", "j", "k") val df2 = (0 until 20).map(i => (i % 7, i % 11, i.toString)).toDF("i", "j", "k") df1.write.format("parquet").bucketBy(8, "i").saveAsTable("t1") df2.write.format("parquet").bucketBy(4, "i").saveAsTable("t2") val t1 = spark.table("t1") val t2 = spark.table("t2") val joined = t1.join(t2, t1("i") === t2("i")) joined.explain == Physical Plan == *(5) SortMergeJoin [i#44], [i#50], Inner :- *(2) Sort [i#44 ASC NULLS FIRST], false, 0 : +- Exchange hashpartitioning(i#44, 200), true, [id=#105] : +- *(1) Project [i#44, j#45, k#46] : +- *(1) Filter isnotnull(i#44) : +- *(1) ColumnarToRow : +- FileScan parquet default.t1[i#44,j#45,k#46] Batched: true, DataFilters: [isnotnull(i#44)], Format: Parquet, Location: InMemoryFileIndex[...], PartitionFilters: [], PushedFilters: [IsNotNull(i)], ReadSchema: struct<i:int,j:int,k:string>, SelectedBucketsCount: 8 out of 8 +- *(4) Sort [i#50 ASC NULLS FIRST], false, 0 +- Exchange hashpartitioning(i#50, 200), true, [id=#115] +- *(3) Project [i#50, j#51, k#52] +- *(3) Filter isnotnull(i#50) +- *(3) ColumnarToRow +- FileScan parquet default.t2[i#50,j#51,k#52] Batched: true, DataFilters: [isnotnull(i#50)], Format: Parquet, Location: InMemoryFileIndex[...], PartitionFilters: [], PushedFilters: [IsNotNull(i)], ReadSchema: struct<i:int,j:int,k:string>, SelectedBucketsCount: 4 out of 4 ``` This PR proposes to introduce coalescing buckets when the following conditions are met to eliminate the full shuffle: - Join type is inner join - The larger bucket number is divisible by the smaller bucket number. ### Why are the changes needed? <!-- Please clarify why the changes are needed. For instance, 1. If you propose a new API, clarify the use case for a new API. 2. If you fix a bug, you can clarify why it is a bug. --> Eliminating the full shuffle can benefit for scenarios where two large tables are joined. Especially when the tables are already bucketed but differ in the number of buckets, we could take advantage of it. ### Does this PR introduce any user-facing change? <!-- If yes, please clarify the previous behavior and the change this PR proposes - provide the console output, description and/or an example to show the behavior difference if possible. If no, write 'No'. --> If a config `spark.sql.bucketing.coalesce` is set to `true` and the bucket coalescing conditions are met, a full shuffle can be eliminated: ``` spark.conf.set("spark.sql.autoBroadcastJoinThreshold", "0") spark.conf.set("spark.sql.bucketing.coalesce", "true") val df1 = (0 until 20).map(i => (i % 5, i % 13, i.toString)).toDF("i", "j", "k") val df2 = (0 until 20).map(i => (i % 7, i % 11, i.toString)).toDF("i", "j", "k") df1.write.format("parquet").bucketBy(8, "i").saveAsTable("t1") df2.write.format("parquet").bucketBy(4, "i").saveAsTable("t2") val t1 = spark.table("t1") val t2 = spark.table("t2") val joined = t1.join(t2, t1("i") === t2("i")) joined.explain == Physical Plan == *(3) SortMergeJoin [i#44], [i#50], Inner :- *(1) Sort [i#44 ASC NULLS FIRST], false, 0 : +- *(1) Project [i#44, j#45, k#46] : +- *(1) Filter isnotnull(i#44) : +- *(1) ColumnarToRow : +- FileScan parquet default.t1[i#44,j#45,k#46] Batched: true, DataFilters: [isnotnull(i#44)], Format: Parquet, Location: InMemoryFileIndex[...], PartitionFilters: [], PushedFilters: [IsNotNull(i)], ReadSchema: struct<i:int,j:int,k:string>, SelectedBucketsCount: 8 out of 8 +- *(2) Sort [i#50 ASC NULLS FIRST], false, 0 +- *(2) Project [i#50, j#51, k#52] +- *(2) Filter isnotnull(i#50) +- *(2) ColumnarToRow +- FileScan parquet default.t2[i#50,j#51,k#52] Batched: true, DataFilters: [isnotnull(i#50)], Format: Parquet, Location: InMemoryFileIndex[...], PartitionFilters: [], PushedFilters: [IsNotNull(i)], ReadSchema: struct<i:int,j:int,k:string>, SelectedBucketsCount: 4 out of 4 ``` ### How was this patch tested? <!-- If tests were added, say they were added here. Please make sure to add some test cases that check the changes thoroughly including negative and positive cases if possible. If it was tested in a way different from regular unit tests, please clarify how you tested step by step, ideally copy and paste-able, so that other reviewers can test and check, and descendants can verify in the future. If tests were not added, please describe why they were not added and/or why it was difficult to add. --> Added unit tests
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
