cloud-fan commented on a change in pull request #29804:
URL: https://github.com/apache/spark/pull/29804#discussion_r492656178
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/internal/SQLConf.scala
##########
@@ -2764,6 +2764,14 @@ object SQLConf {
.booleanConf
.createWithDefault(false)
+ val DYNAMIC_DECIDE_BUCKETING_ENABLED =
+ buildConf("spark.sql.sources.dynamic.decide.bucketing.enabled")
Review comment:
+1
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/internal/SQLConf.scala
##########
@@ -2764,6 +2764,14 @@ object SQLConf {
.booleanConf
.createWithDefault(false)
+ val DYNAMIC_DECIDE_BUCKETING_ENABLED =
+ buildConf("spark.sql.sources.dynamic.decide.bucketing.enabled")
Review comment:
or `dynamicBucketScan`
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/internal/SQLConf.scala
##########
@@ -2764,6 +2764,14 @@ object SQLConf {
.booleanConf
.createWithDefault(false)
+ val DYNAMIC_DECIDE_BUCKETING_ENABLED =
+ buildConf("spark.sql.sources.dynamic.decide.bucketing.enabled")
Review comment:
or `enableDynamicBucketScan`
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/DataSourceScanExec.scala
##########
@@ -165,6 +166,7 @@ case class FileSourceScanExec(
partitionFilters: Seq[Expression],
optionalBucketSet: Option[BitSet],
optionalNumCoalescedBuckets: Option[Int],
+ optionalDynamicDecideBucketing: Option[Boolean],
Review comment:
Why does it need to be `Option[Boolean]`? I'd expect a simple
`disableBucketScan: Boolean = false`
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/PlanBucketing.scala
##########
@@ -0,0 +1,144 @@
+/*
+ * 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.expressions.aggregate.{Partial,
PartialMerge}
+import org.apache.spark.sql.catalyst.plans.physical.{ClusteredDistribution,
HashClusteredDistribution}
+import org.apache.spark.sql.catalyst.rules.Rule
+import org.apache.spark.sql.execution.{FileSourceScanExec, FilterExec,
ProjectExec, SortExec, SparkPlan}
+import org.apache.spark.sql.execution.aggregate.BaseAggregateExec
+import org.apache.spark.sql.execution.exchange.Exchange
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Plans bucketing dynamically based on actual physical query plan.
Review comment:
`bucketing` -> `bucket scan`
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/PlanBucketing.scala
##########
@@ -0,0 +1,144 @@
+/*
+ * 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.expressions.aggregate.{Partial,
PartialMerge}
+import org.apache.spark.sql.catalyst.plans.physical.{ClusteredDistribution,
HashClusteredDistribution}
+import org.apache.spark.sql.catalyst.rules.Rule
+import org.apache.spark.sql.execution.{FileSourceScanExec, FilterExec,
ProjectExec, SortExec, SparkPlan}
+import org.apache.spark.sql.execution.aggregate.BaseAggregateExec
+import org.apache.spark.sql.execution.exchange.Exchange
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Plans bucketing dynamically based on actual physical query plan.
Review comment:
`bucketing` -> `bucket scan`, or `bucketed table scan`, just use a
consistent terminology.
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/PlanBucketing.scala
##########
@@ -0,0 +1,144 @@
+/*
+ * 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.expressions.aggregate.{Partial,
PartialMerge}
+import org.apache.spark.sql.catalyst.plans.physical.{ClusteredDistribution,
HashClusteredDistribution}
+import org.apache.spark.sql.catalyst.rules.Rule
+import org.apache.spark.sql.execution.{FileSourceScanExec, FilterExec,
ProjectExec, SortExec, SparkPlan}
+import org.apache.spark.sql.execution.aggregate.BaseAggregateExec
+import org.apache.spark.sql.execution.exchange.Exchange
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Plans bucketing dynamically based on actual physical query plan.
+ * NOTE: this rule is designed to be applied right after
[[EnsureRequirements]],
+ * where all [[ShuffleExchangeExec]] and [[SortExec]] have been added to plan
properly.
+ *
+ * When BUCKETING_ENABLED and DYNAMIC_DECIDE_BUCKETING_ENABLED are set to
true, go through
+ * query plan to check where bucketed table scan is unnecessary, and disable
bucketed table
+ * scan if needed.
+ *
+ * For all operators which [[hasInterestingPartition]] (i.e., require
[[ClusteredDistribution]]
+ * or [[HashClusteredDistribution]]), check if the sub-plan for operator has
[[Exchange]] and
+ * bucketed table scan (and only allow certain operators in plan, see details
in
+ * [[canDisableBucketedScan]]). If yes, disable the bucketed table scan in the
sub-plan.
+ *
+ * Examples:
+ * (1).join:
+ * SortMergeJoin(t1.i = t2.j)
+ * / \
+ * Sort(i) Sort(j)
+ * / \
+ * Shuffle(i) Scan(t2: i, j)
+ * / (bucketed on column j, enable bucketed scan)
+ * Scan(t1: i, j)
+ * (bucketed on column j, DISABLE bucketed scan)
+ *
+ * (2).aggregate:
+ * HashAggregate(i, ..., Final)
+ * |
+ * Shuffle(i)
+ * |
+ * HashAggregate(i, ..., Partial)
+ * |
+ * Filter
+ * |
+ * Scan(t1: i, j)
+ * (bucketed on column j, DISABLE bucketed scan)
+ *
+ * The idea of [[hasInterestingPartition]] is inspired from "interesting
order" in
+ * the paper "Access Path Selection in a Relational Database Management System"
+ * (http://www.inf.ed.ac.uk/teaching/courses/adbs/AccessPath.pdf).
+ */
+case class PlanBucketing(conf: SQLConf) extends Rule[SparkPlan] {
Review comment:
nit: `DisableUnnecessaryBucketScan`
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/PlanBucketing.scala
##########
@@ -0,0 +1,144 @@
+/*
+ * 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.expressions.aggregate.{Partial,
PartialMerge}
+import org.apache.spark.sql.catalyst.plans.physical.{ClusteredDistribution,
HashClusteredDistribution}
+import org.apache.spark.sql.catalyst.rules.Rule
+import org.apache.spark.sql.execution.{FileSourceScanExec, FilterExec,
ProjectExec, SortExec, SparkPlan}
+import org.apache.spark.sql.execution.aggregate.BaseAggregateExec
+import org.apache.spark.sql.execution.exchange.Exchange
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Plans bucketing dynamically based on actual physical query plan.
+ * NOTE: this rule is designed to be applied right after
[[EnsureRequirements]],
+ * where all [[ShuffleExchangeExec]] and [[SortExec]] have been added to plan
properly.
+ *
+ * When BUCKETING_ENABLED and DYNAMIC_DECIDE_BUCKETING_ENABLED are set to
true, go through
+ * query plan to check where bucketed table scan is unnecessary, and disable
bucketed table
+ * scan if needed.
+ *
+ * For all operators which [[hasInterestingPartition]] (i.e., require
[[ClusteredDistribution]]
+ * or [[HashClusteredDistribution]]), check if the sub-plan for operator has
[[Exchange]] and
+ * bucketed table scan (and only allow certain operators in plan, see details
in
+ * [[canDisableBucketedScan]]). If yes, disable the bucketed table scan in the
sub-plan.
+ *
+ * Examples:
+ * (1).join:
+ * SortMergeJoin(t1.i = t2.j)
+ * / \
+ * Sort(i) Sort(j)
+ * / \
+ * Shuffle(i) Scan(t2: i, j)
+ * / (bucketed on column j, enable bucketed scan)
+ * Scan(t1: i, j)
+ * (bucketed on column j, DISABLE bucketed scan)
+ *
+ * (2).aggregate:
+ * HashAggregate(i, ..., Final)
+ * |
+ * Shuffle(i)
+ * |
+ * HashAggregate(i, ..., Partial)
+ * |
+ * Filter
+ * |
+ * Scan(t1: i, j)
+ * (bucketed on column j, DISABLE bucketed scan)
+ *
+ * The idea of [[hasInterestingPartition]] is inspired from "interesting
order" in
+ * the paper "Access Path Selection in a Relational Database Management System"
+ * (http://www.inf.ed.ac.uk/teaching/courses/adbs/AccessPath.pdf).
+ */
+case class PlanBucketing(conf: SQLConf) extends Rule[SparkPlan] {
+ private def disableBucketWithInterestingPartition(plan: SparkPlan):
SparkPlan = {
+ var hasPlanWithInterestingPartition = false
+
+ val newPlan = plan.transformUp {
+ case p if hasInterestingPartition(p) =>
+ hasPlanWithInterestingPartition = true
Review comment:
This looks tricky and fragile. We shouldn't call `transformUp` and
update a global state. Can we write a recursive method to do bottom-up tree
traverse manually? using `TreeNode.mapChildren.`
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/PlanBucketing.scala
##########
@@ -0,0 +1,144 @@
+/*
+ * 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.expressions.aggregate.{Partial,
PartialMerge}
+import org.apache.spark.sql.catalyst.plans.physical.{ClusteredDistribution,
HashClusteredDistribution}
+import org.apache.spark.sql.catalyst.rules.Rule
+import org.apache.spark.sql.execution.{FileSourceScanExec, FilterExec,
ProjectExec, SortExec, SparkPlan}
+import org.apache.spark.sql.execution.aggregate.BaseAggregateExec
+import org.apache.spark.sql.execution.exchange.Exchange
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Plans bucketing dynamically based on actual physical query plan.
+ * NOTE: this rule is designed to be applied right after
[[EnsureRequirements]],
+ * where all [[ShuffleExchangeExec]] and [[SortExec]] have been added to plan
properly.
+ *
+ * When BUCKETING_ENABLED and DYNAMIC_DECIDE_BUCKETING_ENABLED are set to
true, go through
+ * query plan to check where bucketed table scan is unnecessary, and disable
bucketed table
+ * scan if needed.
+ *
+ * For all operators which [[hasInterestingPartition]] (i.e., require
[[ClusteredDistribution]]
+ * or [[HashClusteredDistribution]]), check if the sub-plan for operator has
[[Exchange]] and
Review comment:
> check if the sub-plan for operator has [[Exchange]]
Can we make it more fine-grained? It's possible that the exchange and the
bucket scan are not in the same lineage:
```
node with interesting partition
|
binary node
/ \
Exchange Bucket Scan
```
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/bucketing/PlanBucketing.scala
##########
@@ -0,0 +1,144 @@
+/*
+ * 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.expressions.aggregate.{Partial,
PartialMerge}
+import org.apache.spark.sql.catalyst.plans.physical.{ClusteredDistribution,
HashClusteredDistribution}
+import org.apache.spark.sql.catalyst.rules.Rule
+import org.apache.spark.sql.execution.{FileSourceScanExec, FilterExec,
ProjectExec, SortExec, SparkPlan}
+import org.apache.spark.sql.execution.aggregate.BaseAggregateExec
+import org.apache.spark.sql.execution.exchange.Exchange
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Plans bucketing dynamically based on actual physical query plan.
+ * NOTE: this rule is designed to be applied right after
[[EnsureRequirements]],
+ * where all [[ShuffleExchangeExec]] and [[SortExec]] have been added to plan
properly.
+ *
+ * When BUCKETING_ENABLED and DYNAMIC_DECIDE_BUCKETING_ENABLED are set to
true, go through
+ * query plan to check where bucketed table scan is unnecessary, and disable
bucketed table
+ * scan if needed.
+ *
+ * For all operators which [[hasInterestingPartition]] (i.e., require
[[ClusteredDistribution]]
+ * or [[HashClusteredDistribution]]), check if the sub-plan for operator has
[[Exchange]] and
Review comment:
The above quey is not valid, I just put it here as an example.
----------------------------------------------------------------
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:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
