[GitHub] [spark] imback82 commented on a change in pull request #28123: [SPARK-31350][SQL] Coalesce bucketed tables for sort merge join if applicable
imback82 commented on a change in pull request #28123:
URL: https://github.com/apache/spark/pull/28123#discussion_r442629743
##
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/CoalesceBucketsInSortMergeJoin.scala
##
@@ -0,0 +1,132 @@
+/*
+ * 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.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.{FileSourceScanExec, FilterExec,
ProjectExec, SparkPlan}
+import org.apache.spark.sql.execution.joins.SortMergeJoinExec
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * This rule coalesces one side of the `SortMergeJoin` if the following
conditions are met:
+ * - Two bucketed tables are joined.
+ * - Join keys match with output partition expressions on their respective
sides.
+ * - The larger bucket number is divisible by the smaller bucket number.
+ * - COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_ENABLED is set to true.
+ * - The ratio of the number of buckets is less than the value set in
+ * COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_MAX_BUCKET_RATIO.
+ */
+case class CoalesceBucketsInSortMergeJoin(conf: SQLConf) extends
Rule[SparkPlan] {
+ private def mayCoalesce(numBuckets1: Int, numBuckets2: Int, conf: SQLConf):
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 <=
conf.getConf(SQLConf.COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_MAX_BUCKET_RATIO)) {
+ Some(small)
+} else {
+ None
+}
+ }
+
+ private def updateNumCoalescedBuckets(plan: SparkPlan, numCoalescedBuckets:
Int): SparkPlan = {
+plan.transformUp {
+ case f: FileSourceScanExec =>
+f.copy(optionalNumCoalescedBuckets = Some(numCoalescedBuckets))
+}
+ }
+
+ def apply(plan: SparkPlan): SparkPlan = {
+if (!conf.getConf(SQLConf.COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_ENABLED)) {
+ return plan
+}
+
+plan transform {
+ case ExtractSortMergeJoinWithBuckets(smj, numLeftBuckets,
numRightBuckets)
+if numLeftBuckets != numRightBuckets =>
+mayCoalesce(numLeftBuckets, numRightBuckets, conf).map {
numCoalescedBuckets =>
+ if (numCoalescedBuckets != numLeftBuckets) {
+smj.copy(left = updateNumCoalescedBuckets(smj.left,
numCoalescedBuckets))
+ } else {
+smj.copy(right = updateNumCoalescedBuckets(smj.right,
numCoalescedBuckets))
+ }
+}.getOrElse(smj)
+ case other => other
+}
+ }
+}
+
+/**
+ * An extractor that extracts `SortMergeJoinExec` where both sides of the join
have the bucketed
+ * tables and are consisted of only the scan operation.
+ */
+object ExtractSortMergeJoinWithBuckets {
+ private def isScanOperation(plan: SparkPlan): Boolean = plan match {
+case f: FilterExec => isScanOperation(f.child)
+case p: ProjectExec => isScanOperation(p.child)
+case _: FileSourceScanExec => true
+case _ => false
+ }
+
+ private def getBucketSpec(plan: SparkPlan): Option[BucketSpec] = {
+plan.collectFirst {
+ case f: FileSourceScanExec if f.relation.bucketSpec.nonEmpty &&
+ f.optionalNumCoalescedBuckets.isEmpty =>
+f.relation.bucketSpec.get
+}
+ }
+
+ /**
+ * The join keys should match with expressions for output partitioning. Note
that
+ * the ordering does not matter because it will be handled in
`EnsureRequirements`.
+ */
+ private def satisfiesOutputPartitioning(
+ keys: Seq[Expression],
+ partitioning: Partitioning): Boolean = {
+partitioning match {
+ case HashPartitioning(exprs, _) if exprs.length == keys.length =>
+exprs.forall(e => keys.exists(_.semanticE
[GitHub] [spark] imback82 commented on a change in pull request #28123: [SPARK-31350][SQL] Coalesce bucketed tables for sort merge join if applicable
imback82 commented on a change in pull request #28123:
URL: https://github.com/apache/spark/pull/28123#discussion_r442629600
##
File path:
sql/core/src/test/scala/org/apache/spark/sql/execution/bucketing/CoalesceBucketsInSortMergeJoinSuite.scala
##
@@ -0,0 +1,194 @@
+/*
+ * 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.expressions.{Attribute,
AttributeReference}
+import org.apache.spark.sql.catalyst.optimizer.BuildLeft
+import org.apache.spark.sql.catalyst.plans.Inner
+import org.apache.spark.sql.execution.{BinaryExecNode, FileSourceScanExec,
SparkPlan}
+import org.apache.spark.sql.execution.datasources.{HadoopFsRelation,
InMemoryFileIndex, PartitionSpec}
+import org.apache.spark.sql.execution.datasources.parquet.ParquetFileFormat
+import org.apache.spark.sql.execution.joins.{BroadcastHashJoinExec,
SortMergeJoinExec}
+import org.apache.spark.sql.internal.SQLConf
+import org.apache.spark.sql.test.{SharedSparkSession, SQLTestUtils}
+import org.apache.spark.sql.types.{IntegerType, StructType}
+
+class CoalesceBucketsInSortMergeJoinSuite extends SQLTestUtils with
SharedSparkSession {
+ case class RelationSetting(
+ cols: Seq[Attribute],
+ numBuckets: Int,
+ expectedCoalescedNumBuckets: Option[Int])
+
+ object RelationSetting {
+def apply(numBuckets: Int, expectedCoalescedNumBuckets: Option[Int]):
RelationSetting = {
+ val cols = Seq(AttributeReference("i", IntegerType)())
+ RelationSetting(cols, numBuckets, expectedCoalescedNumBuckets)
+}
+ }
+
+ case class JoinSetting(
+ leftKeys: Seq[Attribute],
+ rightKeys: Seq[Attribute],
+ leftRelation: RelationSetting,
+ rightRelation: RelationSetting,
+ isSortMergeJoin: Boolean)
+
+ object JoinSetting {
+def apply(l: RelationSetting, r: RelationSetting, isSortMergeJoin:
Boolean): JoinSetting = {
+ JoinSetting(l.cols, r.cols, l, r, isSortMergeJoin)
+}
+ }
+
+ private def newFileSourceScanExec(setting: RelationSetting):
FileSourceScanExec = {
+val relation = HadoopFsRelation(
+ location = new InMemoryFileIndex(spark, Nil, Map.empty, None),
+ partitionSchema = PartitionSpec.emptySpec.partitionColumns,
+ dataSchema = StructType.fromAttributes(setting.cols),
+ bucketSpec = Some(BucketSpec(setting.numBuckets,
setting.cols.map(_.name), Nil)),
+ fileFormat = new ParquetFileFormat(),
+ options = Map.empty)(spark)
+FileSourceScanExec(relation, setting.cols, relation.dataSchema, Nil, None,
None, Nil, None)
+ }
+
+ private def run(setting: JoinSetting): Unit = {
+val swappedSetting = setting.copy(
+ leftKeys = setting.rightKeys,
+ rightKeys = setting.leftKeys,
+ leftRelation = setting.rightRelation,
+ rightRelation = setting.leftRelation)
+
+Seq(setting, swappedSetting).foreach { case s =>
+ val lScan = newFileSourceScanExec(s.leftRelation)
+ val rScan = newFileSourceScanExec(s.rightRelation)
+ val join = if (s.isSortMergeJoin) {
+SortMergeJoinExec(s.leftKeys, s.rightKeys, Inner, None, lScan, rScan)
+ } else {
+BroadcastHashJoinExec(
+ s.leftKeys, s.rightKeys, Inner, BuildLeft, None, lScan, rScan)
+ }
+
+ val plan = CoalesceBucketsInSortMergeJoin(spark.sessionState.conf)(join)
+
+ def verify(expected: Option[Int], subPlan: SparkPlan): Unit = {
+val coalesced = subPlan.collect {
+ case f: FileSourceScanExec if f.optionalNumCoalescedBuckets.nonEmpty
=>
+f.optionalNumCoalescedBuckets.get
+}
+if (expected.isDefined) {
+ assert(coalesced.size == 1 && coalesced(0) == expected.get)
+} else {
+ assert(coalesced.isEmpty)
+}
+ }
+
+ verify(s.leftRelation.expectedCoalescedNumBuckets,
plan.asInstanceOf[BinaryExecNode].left)
+ verify(s.rightRelation.expectedCoalescedNumBuckets,
plan.asInstanceOf[BinaryExecNode].right)
+}
+ }
+
+ test("bucket coalescing - basic") {
+withSQLConf(SQLConf.COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_ENABLED.key ->
"true") {
+ run(JoinS
[GitHub] [spark] imback82 commented on a change in pull request #28123: [SPARK-31350][SQL] Coalesce bucketed tables for sort merge join if applicable
imback82 commented on a change in pull request #28123:
URL: https://github.com/apache/spark/pull/28123#discussion_r442399880
##
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/DataSourceScanExec.scala
##
@@ -356,7 +360,8 @@ case class FileSourceScanExec(
spec.numBuckets
}
metadata + ("SelectedBucketsCount" ->
-s"$numSelectedBuckets out of ${spec.numBuckets}")
+(s"$numSelectedBuckets out of ${spec.numBuckets}" +
+ optionalNumCoalescedBuckets.map { b => s" (Coalesced to
$b)"}.getOrElse("")))
Review comment:
Added.
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
[GitHub] [spark] imback82 commented on a change in pull request #28123: [SPARK-31350][SQL] Coalesce bucketed tables for sort merge join if applicable
imback82 commented on a change in pull request #28123:
URL: https://github.com/apache/spark/pull/28123#discussion_r442400050
##
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/DataSourceScanExec.scala
##
@@ -544,8 +549,19 @@ case class FileSourceScanExec(
filesGroupedToBuckets
}
-val filePartitions = Seq.tabulate(bucketSpec.numBuckets) { bucketId =>
- FilePartition(bucketId, prunedFilesGroupedToBuckets.getOrElse(bucketId,
Array.empty))
+val filePartitions = optionalNumCoalescedBuckets.map { numCoalescedBuckets
=>
+ logInfo(s"Coalescing to ${numCoalescedBuckets} buckets")
+ val coalescedBuckets = prunedFilesGroupedToBuckets.groupBy(_._1 %
numCoalescedBuckets)
+ Seq.tabulate(numCoalescedBuckets) { bucketId =>
+val partitionedFiles = coalescedBuckets.get(bucketId).map {
+ _.values.flatten.toArray
+}.getOrElse(Array.empty)
+FilePartition(bucketId, partitionedFiles)
+ }
+} getOrElse {
Review comment:
Updated.
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
[GitHub] [spark] imback82 commented on a change in pull request #28123: [SPARK-31350][SQL] Coalesce bucketed tables for sort merge join if applicable
imback82 commented on a change in pull request #28123:
URL: https://github.com/apache/spark/pull/28123#discussion_r442365966
##
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/CoalesceBucketsInSortMergeJoin.scala
##
@@ -0,0 +1,132 @@
+/*
+ * 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.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.{FileSourceScanExec, FilterExec,
ProjectExec, SparkPlan}
+import org.apache.spark.sql.execution.joins.SortMergeJoinExec
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * This rule coalesces one side of the `SortMergeJoin` if the following
conditions are met:
+ * - Two bucketed tables are joined.
+ * - Join keys match with output partition expressions on their respective
sides.
+ * - The larger bucket number is divisible by the smaller bucket number.
+ * - COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_ENABLED is set to true.
+ * - The ratio of the number of buckets is less than the value set in
+ * COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_MAX_BUCKET_RATIO.
+ */
+case class CoalesceBucketsInSortMergeJoin(conf: SQLConf) extends
Rule[SparkPlan] {
+ private def mayCoalesce(numBuckets1: Int, numBuckets2: Int, conf: SQLConf):
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 <=
conf.getConf(SQLConf.COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_MAX_BUCKET_RATIO)) {
+ Some(small)
+} else {
+ None
+}
+ }
+
+ private def updateNumCoalescedBuckets(plan: SparkPlan, numCoalescedBuckets:
Int): SparkPlan = {
+plan.transformUp {
+ case f: FileSourceScanExec =>
+f.copy(optionalNumCoalescedBuckets = Some(numCoalescedBuckets))
+}
+ }
+
+ def apply(plan: SparkPlan): SparkPlan = {
+if (!conf.getConf(SQLConf.COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_ENABLED)) {
+ return plan
+}
+
+plan transform {
+ case ExtractSortMergeJoinWithBuckets(smj, numLeftBuckets,
numRightBuckets)
+if numLeftBuckets != numRightBuckets =>
+mayCoalesce(numLeftBuckets, numRightBuckets, conf).map {
numCoalescedBuckets =>
+ if (numCoalescedBuckets != numLeftBuckets) {
+smj.copy(left = updateNumCoalescedBuckets(smj.left,
numCoalescedBuckets))
+ } else {
+smj.copy(right = updateNumCoalescedBuckets(smj.right,
numCoalescedBuckets))
+ }
+}.getOrElse(smj)
+ case other => other
+}
+ }
+}
+
+/**
+ * An extractor that extracts `SortMergeJoinExec` where both sides of the join
have the bucketed
+ * tables and are consisted of only the scan operation.
+ */
+object ExtractSortMergeJoinWithBuckets {
+ private def isScanOperation(plan: SparkPlan): Boolean = plan match {
+case f: FilterExec => isScanOperation(f.child)
+case p: ProjectExec => isScanOperation(p.child)
+case _: FileSourceScanExec => true
+case _ => false
+ }
+
+ private def getBucketSpec(plan: SparkPlan): Option[BucketSpec] = {
+plan.collectFirst {
+ case f: FileSourceScanExec if f.relation.bucketSpec.nonEmpty &&
+ f.optionalNumCoalescedBuckets.isEmpty =>
+f.relation.bucketSpec.get
+}
+ }
+
+ /**
+ * The join keys should match with expressions for output partitioning. Note
that
+ * the ordering does not matter because it will be handled in
`EnsureRequirements`.
Review comment:
If we enforce ordering here, we will miss valid cases. For example, if
we have tables `t1` and `t2` bucketed by columns `i` and `j`, `t1 JOIN t2 on
t1.j = t2.j AND t1.i = t2i` will not be satisfied if the ordering is enforced
here. However, `EnsureReq
[GitHub] [spark] imback82 commented on a change in pull request #28123: [SPARK-31350][SQL] Coalesce bucketed tables for sort merge join if applicable
imback82 commented on a change in pull request #28123:
URL: https://github.com/apache/spark/pull/28123#discussion_r442360453
##
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/CoalesceBucketsInSortMergeJoin.scala
##
@@ -0,0 +1,132 @@
+/*
+ * 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.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.{FileSourceScanExec, FilterExec,
ProjectExec, SparkPlan}
+import org.apache.spark.sql.execution.joins.SortMergeJoinExec
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * This rule coalesces one side of the `SortMergeJoin` if the following
conditions are met:
+ * - Two bucketed tables are joined.
+ * - Join keys match with output partition expressions on their respective
sides.
+ * - The larger bucket number is divisible by the smaller bucket number.
+ * - COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_ENABLED is set to true.
+ * - The ratio of the number of buckets is less than the value set in
+ * COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_MAX_BUCKET_RATIO.
+ */
+case class CoalesceBucketsInSortMergeJoin(conf: SQLConf) extends
Rule[SparkPlan] {
+ private def mayCoalesce(numBuckets1: Int, numBuckets2: Int, conf: SQLConf):
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 <=
conf.getConf(SQLConf.COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_MAX_BUCKET_RATIO)) {
+ Some(small)
+} else {
+ None
+}
+ }
+
+ private def updateNumCoalescedBuckets(plan: SparkPlan, numCoalescedBuckets:
Int): SparkPlan = {
+plan.transformUp {
+ case f: FileSourceScanExec =>
+f.copy(optionalNumCoalescedBuckets = Some(numCoalescedBuckets))
+}
+ }
+
+ def apply(plan: SparkPlan): SparkPlan = {
+if (!conf.getConf(SQLConf.COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_ENABLED)) {
+ return plan
+}
+
+plan transform {
+ case ExtractSortMergeJoinWithBuckets(smj, numLeftBuckets,
numRightBuckets)
+if numLeftBuckets != numRightBuckets =>
+mayCoalesce(numLeftBuckets, numRightBuckets, conf).map {
numCoalescedBuckets =>
+ if (numCoalescedBuckets != numLeftBuckets) {
+smj.copy(left = updateNumCoalescedBuckets(smj.left,
numCoalescedBuckets))
+ } else {
+smj.copy(right = updateNumCoalescedBuckets(smj.right,
numCoalescedBuckets))
+ }
+}.getOrElse(smj)
+ case other => other
+}
+ }
+}
+
+/**
+ * An extractor that extracts `SortMergeJoinExec` where both sides of the join
have the bucketed
+ * tables and are consisted of only the scan operation.
+ */
+object ExtractSortMergeJoinWithBuckets {
+ private def isScanOperation(plan: SparkPlan): Boolean = plan match {
+case f: FilterExec => isScanOperation(f.child)
+case p: ProjectExec => isScanOperation(p.child)
+case _: FileSourceScanExec => true
+case _ => false
+ }
+
+ private def getBucketSpec(plan: SparkPlan): Option[BucketSpec] = {
+plan.collectFirst {
+ case f: FileSourceScanExec if f.relation.bucketSpec.nonEmpty &&
+ f.optionalNumCoalescedBuckets.isEmpty =>
+f.relation.bucketSpec.get
+}
+ }
+
+ /**
+ * The join keys should match with expressions for output partitioning. Note
that
+ * the ordering does not matter because it will be handled in
`EnsureRequirements`.
+ */
+ private def satisfiesOutputPartitioning(
+ keys: Seq[Expression],
+ partitioning: Partitioning): Boolean = {
+partitioning match {
+ case HashPartitioning(exprs, _) if exprs.length == keys.length =>
+exprs.forall(e => keys.exists(_.semanticE
[GitHub] [spark] imback82 commented on a change in pull request #28123: [SPARK-31350][SQL] Coalesce bucketed tables for sort merge join if applicable
imback82 commented on a change in pull request #28123:
URL: https://github.com/apache/spark/pull/28123#discussion_r441023801
##
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/CoalesceBucketsInSortMergeJoin.scala
##
@@ -0,0 +1,112 @@
+/*
+ * 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.rules.Rule
+import org.apache.spark.sql.execution.{FileSourceScanExec, FilterExec,
ProjectExec, SparkPlan}
+import org.apache.spark.sql.execution.joins.SortMergeJoinExec
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * This rule coalesces one side of the `SortMergeJoin` if the following
conditions are met:
+ * - Two bucketed tables are joined.
+ * - The larger bucket number is divisible by the smaller bucket number.
+ * - COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_ENABLED is set to true.
+ * - The ratio of the number of buckets is less than the value set in
+ * COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_MAX_BUCKET_RATIO.
+ */
+case class CoalesceBucketsInSortMergeJoin(conf: SQLConf) extends
Rule[SparkPlan] {
+ private def mayCoalesce(numBuckets1: Int, numBuckets2: Int, conf: SQLConf):
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 <=
conf.getConf(SQLConf.COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_MAX_BUCKET_RATIO)) {
+ Some(small)
+} else {
+ None
+}
+ }
+
+ private def updateNumCoalescedBuckets(plan: SparkPlan, numCoalescedBuckets:
Int): SparkPlan = {
+plan.transformUp {
+ case f: FileSourceScanExec =>
+f.copy(optionalNumCoalescedBuckets = Some(numCoalescedBuckets))
+}
+ }
+
+ def apply(plan: SparkPlan): SparkPlan = {
+if (!conf.getConf(SQLConf.COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_ENABLED)) {
+ return plan
+}
+
+plan transform {
+ case ExtractSortMergeJoinWithBuckets(smj, numLeftBuckets,
numRightBuckets)
+if numLeftBuckets != numRightBuckets =>
+mayCoalesce(numLeftBuckets, numRightBuckets, conf).map {
numCoalescedBuckets =>
+ if (numCoalescedBuckets != numLeftBuckets) {
+smj.copy(left = updateNumCoalescedBuckets(smj.left,
numCoalescedBuckets))
+ } else {
+smj.copy(right = updateNumCoalescedBuckets(smj.right,
numCoalescedBuckets))
+ }
+}.getOrElse(smj)
+ case other => other
+}
+ }
+}
+
+/**
+ * An extractor that extracts `SortMergeJoinExec` where both sides of the join
have the bucketed
+ * tables and are consisted of only the scan operation.
+ */
+object ExtractSortMergeJoinWithBuckets {
+ private def isScanOperation(plan: SparkPlan): Boolean = plan match {
+case f: FilterExec => isScanOperation(f.child)
+case p: ProjectExec => isScanOperation(p.child)
Review comment:
Ah ok, I misunderstood @viirya's question (I thought I had to move the
logic to output "partitioning" in projection). I will make these changes.
Thanks!
##
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/DataSourceScanExec.scala
##
@@ -165,6 +166,7 @@ case class FileSourceScanExec(
requiredSchema: StructType,
partitionFilters: Seq[Expression],
optionalBucketSet: Option[BitSet],
+optionalNumCoalescedBuckets: Option[Int],
Review comment:
Will do.
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: re
[GitHub] [spark] imback82 commented on a change in pull request #28123: [SPARK-31350][SQL] Coalesce bucketed tables for sort merge join if applicable
imback82 commented on a change in pull request #28123:
URL: https://github.com/apache/spark/pull/28123#discussion_r440536834
##
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/CoalesceBucketsInSortMergeJoin.scala
##
@@ -0,0 +1,112 @@
+/*
+ * 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.rules.Rule
+import org.apache.spark.sql.execution.{FileSourceScanExec, FilterExec,
ProjectExec, SparkPlan}
+import org.apache.spark.sql.execution.joins.SortMergeJoinExec
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * This rule coalesces one side of the `SortMergeJoin` if the following
conditions are met:
+ * - Two bucketed tables are joined.
Review comment:
Updated. Thanks.
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
[GitHub] [spark] imback82 commented on a change in pull request #28123: [SPARK-31350][SQL] Coalesce bucketed tables for sort merge join if applicable
imback82 commented on a change in pull request #28123:
URL: https://github.com/apache/spark/pull/28123#discussion_r439706623
##
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/CoalesceBucketsInSortMergeJoin.scala
##
@@ -0,0 +1,112 @@
+/*
+ * 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.rules.Rule
+import org.apache.spark.sql.execution.{FileSourceScanExec, FilterExec,
ProjectExec, SparkPlan}
+import org.apache.spark.sql.execution.joins.SortMergeJoinExec
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * This rule coalesces one side of the `SortMergeJoin` if the following
conditions are met:
+ * - Two bucketed tables are joined.
+ * - The larger bucket number is divisible by the smaller bucket number.
+ * - COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_ENABLED is set to true.
+ * - The ratio of the number of buckets is less than the value set in
+ * COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_MAX_BUCKET_RATIO.
+ */
+case class CoalesceBucketsInSortMergeJoin(conf: SQLConf) extends
Rule[SparkPlan] {
+ private def mayCoalesce(numBuckets1: Int, numBuckets2: Int, conf: SQLConf):
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 <=
conf.getConf(SQLConf.COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_MAX_BUCKET_RATIO)) {
+ Some(small)
+} else {
+ None
+}
+ }
+
+ private def updateNumCoalescedBuckets(plan: SparkPlan, numCoalescedBuckets:
Int): SparkPlan = {
+plan.transformUp {
+ case f: FileSourceScanExec =>
+f.copy(optionalNumCoalescedBuckets = Some(numCoalescedBuckets))
+}
+ }
+
+ def apply(plan: SparkPlan): SparkPlan = {
+if (!conf.getConf(SQLConf.COALESCE_BUCKETS_IN_SORT_MERGE_JOIN_ENABLED)) {
+ return plan
+}
+
+plan transform {
+ case ExtractSortMergeJoinWithBuckets(smj, numLeftBuckets,
numRightBuckets)
+if numLeftBuckets != numRightBuckets =>
+mayCoalesce(numLeftBuckets, numRightBuckets, conf).map {
numCoalescedBuckets =>
+ if (numCoalescedBuckets != numLeftBuckets) {
+smj.copy(left = updateNumCoalescedBuckets(smj.left,
numCoalescedBuckets))
+ } else {
+smj.copy(right = updateNumCoalescedBuckets(smj.right,
numCoalescedBuckets))
+ }
+}.getOrElse(smj)
+ case other => other
+}
+ }
+}
+
+/**
+ * An extractor that extracts `SortMergeJoinExec` where both sides of the join
have the bucketed
+ * tables and are consisted of only the scan operation.
+ */
+object ExtractSortMergeJoinWithBuckets {
+ private def isScanOperation(plan: SparkPlan): Boolean = plan match {
+case f: FilterExec => isScanOperation(f.child)
+case p: ProjectExec => isScanOperation(p.child)
Review comment:
Would this be a cleaner approach? Or handling buckets in one place
(`FileSourceScanExec`) would be cleaner? What do you think @cloud-fan / @maropu?
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
[GitHub] [spark] imback82 commented on a change in pull request #28123: [SPARK-31350][SQL] Coalesce bucketed tables for sort merge join if applicable
imback82 commented on a change in pull request #28123: URL: https://github.com/apache/spark/pull/28123#discussion_r439706576 ## File path: sql/core/src/main/scala/org/apache/spark/sql/execution/DataSourceScanExec.scala ## @@ -165,6 +166,7 @@ case class FileSourceScanExec( requiredSchema: StructType, partitionFilters: Seq[Expression], optionalBucketSet: Option[BitSet], +optionalNumCoalescedBuckets: Option[Int], Review comment: Yes, good idea. @cloud-fan/@maropu Do you want me to include this in this PR or do it as a follow up since this PR is already approved? I am fine with either one. Thanks @viirya! 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
[GitHub] [spark] imback82 commented on a change in pull request #28123: [SPARK-31350][SQL] Coalesce bucketed tables for sort merge join if applicable
imback82 commented on a change in pull request #28123:
URL: https://github.com/apache/spark/pull/28123#discussion_r439706394
##
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/CoalesceBucketsInSortMergeJoin.scala
##
@@ -0,0 +1,112 @@
+/*
+ * 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.rules.Rule
+import org.apache.spark.sql.execution.{FileSourceScanExec, FilterExec,
ProjectExec, SparkPlan}
+import org.apache.spark.sql.execution.joins.SortMergeJoinExec
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * This rule coalesces one side of the `SortMergeJoin` if the following
conditions are met:
+ * - Two bucketed tables are joined.
Review comment:
Since `SortMergeJoinExec` is created only for the equi-join case, I
don't think we don't need to check it in this rule. I can update the PR
description to remove `equality conditions` if it causes a confusion.
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
