imback82 commented on a change in pull request #28123: URL: https://github.com/apache/spark/pull/28123#discussion_r417052767
########## File path: sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/CoalesceBucketsInJoin.scala ########## @@ -0,0 +1,116 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.spark.sql.execution.bucketing + +import org.apache.spark.sql.catalyst.catalog.BucketSpec +import org.apache.spark.sql.catalyst.plans.Inner +import org.apache.spark.sql.catalyst.plans.logical.{Filter, Join, LogicalPlan, Project} +import org.apache.spark.sql.catalyst.rules.Rule +import org.apache.spark.sql.execution.datasources.{HadoopFsRelation, LogicalRelation} +import org.apache.spark.sql.internal.SQLConf + +/** + * This rule injects a hint if one side of two bucketed tables can be coalesced + * when the two bucketed tables are inner-joined and they differ in the number of buckets. + */ +object CoalesceBucketsInJoin extends Rule[LogicalPlan] { + private def isPlanEligible(plan: LogicalPlan): Boolean = { + def forall(plan: LogicalPlan)(p: LogicalPlan => Boolean): Boolean = { + p(plan) && plan.children.forall(forall(_)(p)) + } + + forall(plan) { + case _: Filter | _: Project | _: LogicalRelation => true + case _ => false + } + } + + private def getBucketSpec(plan: LogicalPlan): Option[BucketSpec] = { + if (isPlanEligible(plan)) { + plan.collectFirst { + case _ @ LogicalRelation(r: HadoopFsRelation, _, _, _) if r.bucketSpec.nonEmpty => + r.bucketSpec.get + } + } else { + None + } + } + + private def mayCoalesce( + numBuckets1: Int, + numBuckets2: Int, + maxNumBucketsDiff: Int): Option[Int] = { + assert(numBuckets1 != numBuckets2) + val (small, large) = (math.min(numBuckets1, numBuckets2), math.max(numBuckets1, numBuckets2)) + // A bucket can be coalesced only if the bigger number of buckets is divisible by the smaller + // number of buckets because bucket id is calculated by modding the total number of buckets. + if ((large % small == 0) && + ((large - small) <= maxNumBucketsDiff)) { + Some(small) + } else { + None + } + } + + private def addCoalesceBuckets(plan: LogicalPlan, numCoalescedBuckets: Int): LogicalPlan = { + plan.transformUp { + case l @ LogicalRelation(_: HadoopFsRelation, _, _, _) => + CoalesceBuckets(numCoalescedBuckets, l) + } + } + + object ExtractJoinWithBuckets { + def unapply(plan: LogicalPlan): Option[(Join, Int, Int)] = { + plan match { + case join: Join => Review comment: It will remove shuffle for full-outer join as well: With the feature off: ``` scala> t1.join(t2, t1("i") === t2("i"), "full_outer").explain == Physical Plan == SortMergeJoin [i#67], [i#73], FullOuter :- *(2) Sort [i#67 ASC NULLS FIRST], false, 0 : +- Exchange hashpartitioning(i#67, 200), true, [id=#418] : +- *(1) ColumnarToRow : +- FileScan parquet default.t1[i#67,j#68,k#69] Batched: true, DataFilters: [], Format: Parquet, Location: InMemoryFileIndex[], PartitionFilters: [], PushedFilters: [], ReadSchema: struct<i:int,j:int,k:string>, SelectedBucketsCount: 8 out of 8 +- *(4) Sort [i#73 ASC NULLS FIRST], false, 0 +- Exchange hashpartitioning(i#73, 200), true, [id=#425] +- *(3) ColumnarToRow +- FileScan parquet default.t2[i#73,j#74,k#75] Batched: true, DataFilters: [], Format: Parquet, Location: InMemoryFileIndex[], PartitionFilters: [], PushedFilters: [], ReadSchema: struct<i:int,j:int,k:string>, SelectedBucketsCount: 4 out of 4 ``` With the feature on: ``` scala> t1.join(t2, t1("i") === t2("i"), "full_outer").explain == Physical Plan == SortMergeJoin [i#67], [i#73], FullOuter :- *(1) Sort [i#67 ASC NULLS FIRST], false, 0 : +- *(1) ColumnarToRow : +- FileScan parquet default.t1[i#67,j#68,k#69] Batched: true, DataFilters: [], Format: Parquet, Location: InMemoryFileIndex[], PartitionFilters: [], PushedFilters: [], ReadSchema: struct<i:int,j:int,k:string>, SelectedBucketsCount: 8 out of 8 +- *(2) Sort [i#73 ASC NULLS FIRST], false, 0 +- *(2) ColumnarToRow +- FileScan parquet default.t2[i#73,j#74,k#75] Batched: true, DataFilters: [], Format: Parquet, Location: InMemoryFileIndex[], PartitionFilters: [], PushedFilters: [], ReadSchema: struct<i:int,j:int,k:string>, SelectedBucketsCount: 4 out of 4 ``` Did I miss something? ########## File path: sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/CoalesceBucketsInJoin.scala ########## @@ -0,0 +1,116 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.spark.sql.execution.bucketing + +import org.apache.spark.sql.catalyst.catalog.BucketSpec +import org.apache.spark.sql.catalyst.plans.Inner +import org.apache.spark.sql.catalyst.plans.logical.{Filter, Join, LogicalPlan, Project} +import org.apache.spark.sql.catalyst.rules.Rule +import org.apache.spark.sql.execution.datasources.{HadoopFsRelation, LogicalRelation} +import org.apache.spark.sql.internal.SQLConf + +/** + * This rule injects a hint if one side of two bucketed tables can be coalesced + * when the two bucketed tables are inner-joined and they differ in the number of buckets. + */ +object CoalesceBucketsInJoin extends Rule[LogicalPlan] { Review comment: Added. ########## File path: sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/CoalseceBuckets.scala ########## @@ -0,0 +1,34 @@ +/* Review comment: good catch! ---------------------------------------------------------------- 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 --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
