Github user tejasapatil commented on a diff in the pull request: https://github.com/apache/spark/pull/16985#discussion_r116163611 --- Diff: sql/core/src/main/scala/org/apache/spark/sql/execution/joins/ReorderJoinPredicates.scala --- @@ -0,0 +1,93 @@ +/* + * 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.joins + +import scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.catalyst.expressions.Expression +import org.apache.spark.sql.catalyst.plans.physical.{HashPartitioning, Partitioning} +import org.apache.spark.sql.catalyst.rules.Rule +import org.apache.spark.sql.execution.SparkPlan + +/** + * When the physical operators are created for JOIN, the ordering of join keys is based on order + * in which the join keys appear in the user query. That might not match with the output + * partitioning of the join node's children (thus leading to extra sort / shuffle being + * introduced). This rule will change the ordering of the join keys to match with the + * partitioning of the join nodes' children. + */ +class ReorderJoinPredicates extends Rule[SparkPlan] { + private def reorderJoinKeys( + leftKeys: Seq[Expression], + rightKeys: Seq[Expression], + leftPartitioning: Partitioning, + rightPartitioning: Partitioning): (Seq[Expression], Seq[Expression]) = { + + def reorder(expectedOrderOfKeys: Seq[Expression], + currentOrderOfKeys: Seq[Expression]): (Seq[Expression], Seq[Expression]) = { + val leftKeysBuffer = ArrayBuffer[Expression]() + val rightKeysBuffer = ArrayBuffer[Expression]() + + expectedOrderOfKeys.foreach(expression => { + val index = currentOrderOfKeys.indexWhere(e => e.semanticEquals(expression)) + leftKeysBuffer.append(leftKeys(index)) + rightKeysBuffer.append(rightKeys(index)) + }) + (leftKeysBuffer, rightKeysBuffer) + } + + if (leftKeys.forall(_.deterministic) && rightKeys.forall(_.deterministic)) { + leftPartitioning match { + case HashPartitioning(leftExpressions, _) + if leftExpressions.length == leftKeys.length && --- End diff -- I don't think that would be right thing to do. If child is partitioned on `a, b, c, d`, its basically means rows are distributed over hash of `a, b, c, d`. Lets say we have two rows with values of `a, b, c, d` as: - row1 : 1,1,1,1 ==> hash(1,1,1,1) = x - row2 : 1,1,2,2 ==> hash(1,1,2,2) = y If the join key `b,a` is reordered as `a,b` and we want to avid shuffle, that would mean that we expect the child to have same values of `a,b` in the same partition. But if you look at row1 and row2 above, even if values of `a` and `b` are the same, there is no guarantee that they would belong to the same partition... as the partition is based on hash of all `a,b,c,d`. If the join keys are a subset of the partitioning, then there needs to be a shuffle to be done. There is only one exception to this (more of a corner case) : https://issues.apache.org/jira/browse/SPARK-18067
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org