hvanhovell commented on a change in pull request #26434: [SPARK-29544] [SQL] 
optimize skewed partition based on data size
URL: https://github.com/apache/spark/pull/26434#discussion_r366376061
 
 

 ##########
 File path: 
sql/core/src/main/scala/org/apache/spark/sql/execution/adaptive/OptimizeSkewedJoin.scala
 ##########
 @@ -0,0 +1,293 @@
+/*
+ * 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.adaptive
+
+import scala.collection.mutable
+import scala.collection.mutable.ArrayBuffer
+
+import org.apache.spark.{MapOutputStatistics, MapOutputTrackerMaster, SparkEnv}
+import org.apache.spark.rdd.RDD
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.expressions.Attribute
+import org.apache.spark.sql.catalyst.plans._
+import org.apache.spark.sql.catalyst.plans.physical.{Partitioning, 
UnknownPartitioning}
+import org.apache.spark.sql.catalyst.rules.Rule
+import org.apache.spark.sql.execution._
+import org.apache.spark.sql.execution.exchange.ShuffleExchangeExec
+import org.apache.spark.sql.execution.joins.SortMergeJoinExec
+import org.apache.spark.sql.internal.SQLConf
+
+case class OptimizeSkewedJoin(conf: SQLConf) extends Rule[SparkPlan] {
+
+  private val supportedJoinTypes =
+    Inner :: Cross :: LeftSemi :: LeftAnti :: LeftOuter :: RightOuter :: Nil
+
+  /**
+   * A partition is considered as a skewed partition if its size is larger 
than the median
+   * partition size * spark.sql.adaptive.skewedPartitionFactor and also larger 
than
+   * spark.sql.adaptive.skewedPartitionSizeThreshold.
+   */
+  private def isSkewed(
+      stats: MapOutputStatistics,
+      partitionId: Int,
+      medianSize: Long): Boolean = {
+    val size = stats.bytesByPartitionId(partitionId)
+    size > medianSize * 
conf.getConf(SQLConf.ADAPTIVE_EXECUTION_SKEWED_PARTITION_FACTOR) &&
+      size > 
conf.getConf(SQLConf.ADAPTIVE_EXECUTION_SKEWED_PARTITION_SIZE_THRESHOLD)
+  }
+
+  private def medianSize(stats: MapOutputStatistics): Long = {
+    val numPartitions = stats.bytesByPartitionId.length
+    val bytes = stats.bytesByPartitionId.sorted
+    if (bytes(numPartitions / 2) > 0) bytes(numPartitions / 2) else 1
+  }
+
+  /**
+   * Get the map size of the specific reduce shuffle Id.
+   */
+  private def getMapSizesForReduceId(shuffleId: Int, partitionId: Int): 
Array[Long] = {
+    val mapOutputTracker = 
SparkEnv.get.mapOutputTracker.asInstanceOf[MapOutputTrackerMaster]
+    
mapOutputTracker.shuffleStatuses(shuffleId).mapStatuses.map{_.getSizeForBlock(partitionId)}
+  }
+
+  /**
+   * Split the skewed partition based on the map size and the max split number.
+   */
+  private def getMapStartIndices(stage: ShuffleQueryStageExec, partitionId: 
Int): Array[Int] = {
+    val shuffleId = stage.shuffle.shuffleDependency.shuffleHandle.shuffleId
+    val mapPartitionSizes = getMapSizesForReduceId(shuffleId, partitionId)
+    val maxSplits = math.min(conf.getConf(
+      SQLConf.ADAPTIVE_EXECUTION_SKEWED_PARTITION_MAX_SPLITS), 
mapPartitionSizes.length)
+    val avgPartitionSize = mapPartitionSizes.sum / maxSplits
+    val advisoryPartitionSize = math.max(avgPartitionSize,
+      conf.getConf(SQLConf.ADAPTIVE_EXECUTION_SKEWED_PARTITION_SIZE_THRESHOLD))
+    val partitionIndices = mapPartitionSizes.indices
+    val partitionStartIndices = ArrayBuffer[Int]()
+    var postMapPartitionSize = mapPartitionSizes(0)
+    partitionStartIndices += 0
+    partitionIndices.drop(1).foreach { nextPartitionIndex =>
+      val nextMapPartitionSize = mapPartitionSizes(nextPartitionIndex)
+      if (postMapPartitionSize + nextMapPartitionSize > advisoryPartitionSize) 
{
+        partitionStartIndices += nextPartitionIndex
+        postMapPartitionSize = nextMapPartitionSize
+      } else {
+        postMapPartitionSize += nextMapPartitionSize
+      }
+    }
+
+    if (partitionStartIndices.size > maxSplits) {
+      partitionStartIndices.take(maxSplits).toArray
+    } else partitionStartIndices.toArray
+  }
+
+  private def getStatistics(stage: ShuffleQueryStageExec): MapOutputStatistics 
= {
+    assert(stage.resultOption.isDefined, "ShuffleQueryStageExec should" +
+      " already be ready when executing OptimizeSkewedPartitions rule")
+    stage.resultOption.get.asInstanceOf[MapOutputStatistics]
+  }
+
+  private def supportSplitOnLeftPartition(joinType: JoinType) = {
+    joinType == Inner || joinType == Cross || joinType == LeftSemi ||
+      joinType == LeftAnti || joinType == LeftOuter
+  }
+
+  private def supportSplitOnRightPartition(joinType: JoinType) = {
+    joinType == Inner || joinType == Cross || joinType == RightOuter
+  }
+
+  private def getNumMappers(stage: ShuffleQueryStageExec): Int = {
+    stage.shuffle.shuffleDependency.rdd.partitions.length
+  }
+
+  def handleSkewJoin(plan: SparkPlan): SparkPlan = plan.transformUp {
+    case smj @ SortMergeJoinExec(leftKeys, rightKeys, joinType, condition,
+        s1 @ SortExec(_, _, left: ShuffleQueryStageExec, _),
+        s2 @ SortExec(_, _, right: ShuffleQueryStageExec, _))
+      if supportedJoinTypes.contains(joinType) =>
+      val leftStats = getStatistics(left)
+      val rightStats = getStatistics(right)
+      val numPartitions = leftStats.bytesByPartitionId.length
+
+      val leftMedSize = medianSize(leftStats)
+      val rightMedSize = medianSize(rightStats)
+      val leftSizeInfo = s"median size: $leftMedSize, max size: 
${leftStats.bytesByPartitionId.max}"
+      val rightSizeInfo = s"median size: $rightMedSize," +
+        s" max size: ${rightStats.bytesByPartitionId.max}"
+      logDebug(
+        s"""
+          |Try to optimize skewed join.
+          |Left side partition size: $leftSizeInfo
+          |Right side partition size: $rightSizeInfo
+        """.stripMargin)
+
+      val skewedPartitions = mutable.HashSet[Int]()
+      val subJoins = mutable.ArrayBuffer[SparkPlan]()
+      for (partitionId <- 0 until numPartitions) {
+        val isLeftSkew = isSkewed(leftStats, partitionId, leftMedSize)
+        val isRightSkew = isSkewed(rightStats, partitionId, rightMedSize)
+        val leftMapIdStartIndices = if (isLeftSkew && 
supportSplitOnLeftPartition(joinType)) {
+          getMapStartIndices(left, partitionId)
+        } else {
+          Array(0)
+        }
+        val rightMapIdStartIndices = if (isRightSkew && 
supportSplitOnRightPartition(joinType)) {
+          getMapStartIndices(right, partitionId)
+        } else {
+          Array(0)
+        }
+
+        if (leftMapIdStartIndices.length > 1 || rightMapIdStartIndices.length 
> 1) {
+          skewedPartitions += partitionId
+          for (i <- 0 until leftMapIdStartIndices.length;
+               j <- 0 until rightMapIdStartIndices.length) {
+            val leftEndMapId = if (i == leftMapIdStartIndices.length - 1) {
+              getNumMappers(left)
+            } else {
+              leftMapIdStartIndices(i + 1)
+            }
+            val rightEndMapId = if (j == rightMapIdStartIndices.length - 1) {
+              getNumMappers(right)
+            } else {
+              rightMapIdStartIndices(j + 1)
+            }
+            // TODO: we may can optimize the sort merge join to broad cast 
join after
+            //       obtaining the raw data size of per partition,
+            val leftSkewedReader = SkewedPartitionReaderExec(
+              left, partitionId, leftMapIdStartIndices(i), leftEndMapId)
+            val rightSkewedReader = SkewedPartitionReaderExec(right, 
partitionId,
+              rightMapIdStartIndices(j), rightEndMapId)
+            subJoins += SortMergeJoinExec(leftKeys, rightKeys, joinType, 
condition,
+              s1.copy(child = leftSkewedReader), s2.copy(child = 
rightSkewedReader))
+          }
+        }
+      }
+      logDebug(s"number of skewed partitions is ${skewedPartitions.size}")
+      if (skewedPartitions.nonEmpty) {
+        val optimizedSmj = smj.transformDown {
+          case sort @ SortExec(_, _, shuffleStage: ShuffleQueryStageExec, _) =>
+            sort.copy(child = PartialShuffleReaderExec(shuffleStage, 
skewedPartitions.toSet))
+        }
+        subJoins += optimizedSmj
+        UnionExec(subJoins)
+      } else {
+        smj
+      }
+  }
+
+  override def apply(plan: SparkPlan): SparkPlan = {
+    if (!conf.getConf(SQLConf.ADAPTIVE_EXECUTION_SKEWED_JOIN_ENABLED)) {
+      return plan
+    }
+
+    def collectShuffleStages(plan: SparkPlan): Seq[ShuffleQueryStageExec] = 
plan match {
+      case _: LocalShuffleReaderExec => Nil
+      case _: CoalescedShuffleReaderExec => Nil
+      case stage: ShuffleQueryStageExec => Seq(stage)
+      case _ => plan.children.flatMap(collectShuffleStages)
+    }
+
+    val shuffleStages = collectShuffleStages(plan)
+
+    if (shuffleStages.length == 2) {
+      // Currently we only support handling skewed join for 2 table join.
+      handleSkewJoin(plan)
+    } else {
+      plan
+
+    }
+  }
+}
+
+/**
+ * A wrapper of shuffle query stage, which submits one reduce task to read a 
single
+ * shuffle partition 'partitionIndex' produced by the mappers in range 
[startMapIndex, endMapIndex).
+ * This is used to increase the parallelism when reading skewed partitions.
+ *
+ * @param child It's usually `ShuffleQueryStageExec`, but can be the shuffle 
exchange
+ *              node during canonicalization.
+ * @param partitionIndex The pre shuffle partition index.
+ * @param startMapIndex The start map index.
+ * @param endMapIndex The end map index.
+ */
+case class SkewedPartitionReaderExec(
 
 Review comment:
   Like with the RDDs: In general I would be in favor of creating one reader 
node that can deal with the different kinds of shuffle reads. That avoid a 
sprawl of readers, and it also allows us to create a much simpler plan if we 
can just use 1 reader with a join instead of using a union.

----------------------------------------------------------------
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

Reply via email to