Skip to content

[SPARK-16189][SQL] Add ExternalRDD logical plan for input with RDD to have a chance to eliminate serialize/deserialize. #13890

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
ueshin wants to merge 4 commits into from
Closed

[SPARK-16189][SQL] Add ExternalRDD logical plan for input with RDD to have a chance to eliminate serialize/deserialize. #13890

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
036d5fd
Add ExistingRDD logical plan for input with RDD to have a chance to e…
ueshin Jun 14, 2016
62aca29
Revert createDataFrame for Java Beans.
ueshin Jun 24, 2016
e218f5f
Rename ExistingRDD to ExternalRDD.
ueshin Jul 8, 2016
61da040
Uncurry `apply` method.
ueshin Jul 9, 2016
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 3 additions & 9 deletions sql/core/src/main/scala/org/apache/spark/sql/SparkSession.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -237,10 +237,8 @@ class SparkSession private(
@Experimental
def createDataFrame[A <: Product : TypeTag](rdd: RDD[A]): DataFrame = {
SparkSession.setActiveSession(this)
val schema = ScalaReflection.schemaFor[A].dataType.asInstanceOf[StructType]
val attributeSeq = schema.toAttributes
val rowRDD = RDDConversions.productToRowRdd(rdd, schema.map(_.dataType))
Dataset.ofRows(self, LogicalRDD(attributeSeq, rowRDD)(self))
val encoder = Encoders.product[A]
Dataset.ofRows(self, ExistingRDD(rdd)(self)(encoder))
}

/**
Expand Down Expand Up @@ -425,11 +423,7 @@ class SparkSession private(
*/
@Experimental
def createDataset[T : Encoder](data: RDD[T]): Dataset[T] = {
val enc = encoderFor[T]
val attributes = enc.schema.toAttributes
val encoded = data.map(d => enc.toRow(d))
val plan = LogicalRDD(attributes, encoded)(self)
Dataset[T](self, plan)
Dataset[T](self, ExistingRDD(data)(self))
}

/**
Expand Down
72 changes: 64 additions & 8 deletions sql/core/src/main/scala/org/apache/spark/sql/execution/ExistingRDD.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,12 +20,12 @@ package org.apache.spark.sql.execution
import org.apache.commons.lang.StringUtils

import org.apache.spark.rdd.RDD
import org.apache.spark.sql.{AnalysisException, Row, SparkSession, SQLContext}
import org.apache.spark.sql.{AnalysisException, Encoder, Row, SparkSession, SQLContext}
import org.apache.spark.sql.catalyst.{CatalystTypeConverters, InternalRow, TableIdentifier}
import org.apache.spark.sql.catalyst.analysis.MultiInstanceRelation
import org.apache.spark.sql.catalyst.expressions._
import org.apache.spark.sql.catalyst.expressions.codegen.{CodegenContext, ExprCode}
import org.apache.spark.sql.catalyst.plans.logical.{LogicalPlan, Statistics}
import org.apache.spark.sql.catalyst.plans.logical._
import org.apache.spark.sql.catalyst.plans.physical.{HashPartitioning, Partitioning, UnknownPartitioning}
import org.apache.spark.sql.execution.datasources.HadoopFsRelation
import org.apache.spark.sql.execution.datasources.parquet.{ParquetFileFormat => ParquetSource}
Expand Down Expand Up @@ -74,13 +74,71 @@ object RDDConversions {
}
}

private[sql] object ExistingRDD {

def apply[T: Encoder](rdd: RDD[T])(session: SparkSession): LogicalPlan = {

@cloud-fan cloud-fan Jul 8, 2016

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

why curry here?

@ueshin ueshin Jul 8, 2016

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Because I wanted to make the signature similar to the case class constructor.
Should I uncurry?

@cloud-fan cloud-fan Jul 8, 2016

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Actually I'm not sure why we curry the constructor either. Since RDDScan does it, it's ok we follow it. But for this apply method, I don't see the value of doing it. cc @yhuai

@marmbrus marmbrus Jul 8, 2016 •
edited
Loading

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There is probably no reason to use multiple parameter lists here. We sometimes use it for case classes so that arguments that should not effect equality are not included in the generated equals method.

val exisitingRdd = ExistingRDD(CatalystSerde.generateObjAttr[T], rdd)(session)
CatalystSerde.serialize[T](exisitingRdd)
}
}

/** Logical plan node for scanning data from an RDD. */
private[sql] case class ExistingRDD[T](
outputObjAttr: Attribute,
rdd: RDD[T])(session: SparkSession)
extends LeafNode with ObjectProducer with MultiInstanceRelation {

override protected final def otherCopyArgs: Seq[AnyRef] = session :: Nil

override def newInstance(): ExistingRDD.this.type =
ExistingRDD(outputObjAttr.newInstance(), rdd)(session).asInstanceOf[this.type]

override def sameResult(plan: LogicalPlan): Boolean = {
plan.canonicalized match {
case ExistingRDD(_, otherRDD) => rdd.id == otherRDD.id
case _ => false
}
}

override protected def stringArgs: Iterator[Any] = Iterator(output)

@transient override lazy val statistics: Statistics = Statistics(
// TODO: Instead of returning a default value here, find a way to return a meaningful size
// estimate for RDDs. See PR 1238 for more discussions.
sizeInBytes = BigInt(session.sessionState.conf.defaultSizeInBytes)
)
}

/** Physical plan node for scanning data from an RDD. */
private[sql] case class ExistingRDDScanExec[T](

@cloud-fan cloud-fan Jul 6, 2016

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

From the name it's hard to tell what's the difference between this one and RDDScanExec...

@ueshin ueshin Jul 6, 2016

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

How about renaming RDDScanExec to LogicalRDDScanExec ?

@cloud-fan cloud-fan Jul 6, 2016

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't have a good idea here, cc @yhuai

@yhuai yhuai Jul 6, 2016

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LogicalRDDScanExec sounds weird because Logical means that it is a logical node but Exec means that it is physical node.

@cloud-fan cloud-fan Jul 7, 2016

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

how about ExternalRDDScan?

outputObjAttr: Attribute,
rdd: RDD[T]) extends LeafExecNode with ObjectProducerExec {

private[sql] override lazy val metrics = Map(
"numOutputRows" -> SQLMetrics.createMetric(sparkContext, "number of output rows"))

protected override def doExecute(): RDD[InternalRow] = {
val numOutputRows = longMetric("numOutputRows")
val outputDataType = outputObjAttr.dataType
rdd.mapPartitionsInternal { iter =>
val outputObject = ObjectOperator.wrapObjectToRow(outputDataType)
iter.map { value =>
numOutputRows += 1
outputObject(value)
}
}
}

override def simpleString: String = {
s"Scan $nodeName${output.mkString("[", ",", "]")}"
}
}

/** Logical plan node for scanning data from an RDD of InternalRow. */
private[sql] case class LogicalRDD(
output: Seq[Attribute],
rdd: RDD[InternalRow])(session: SparkSession)
extends LogicalPlan with MultiInstanceRelation {

override def children: Seq[LogicalPlan] = Nil
extends LeafNode with MultiInstanceRelation {

override protected final def otherCopyArgs: Seq[AnyRef] = session :: Nil

Expand All @@ -96,16 +154,14 @@ private[sql] case class LogicalRDD(

override protected def stringArgs: Iterator[Any] = Iterator(output)

override def producedAttributes: AttributeSet = outputSet

@transient override lazy val statistics: Statistics = Statistics(
// TODO: Instead of returning a default value here, find a way to return a meaningful size
// estimate for RDDs. See PR 1238 for more discussions.
sizeInBytes = BigInt(session.sessionState.conf.defaultSizeInBytes)
)
}

/** Physical plan node for scanning data from an RDD. */
/** Physical plan node for scanning data from an RDD of InternalRow. */
private[sql] case class RDDScanExec(
output: Seq[Attribute],
rdd: RDD[InternalRow],
Expand Down
8 changes: 6 additions & 2 deletions sql/core/src/main/scala/org/apache/spark/sql/execution/LocalTableScanExec.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,8 +34,12 @@ private[sql] case class LocalTableScanExec(
"numOutputRows" -> SQLMetrics.createMetric(sparkContext, "number of output rows"))

private val unsafeRows: Array[InternalRow] = {
val proj = UnsafeProjection.create(output, output)
rows.map(r => proj(r).copy()).toArray
if (rows.isEmpty) {
Array.empty
} else {
val proj = UnsafeProjection.create(output, output)
rows.map(r => proj(r).copy()).toArray
}
}

private lazy val rdd = sqlContext.sparkContext.parallelize(unsafeRows)
Expand Down
1 change: 1 addition & 0 deletions sql/core/src/main/scala/org/apache/spark/sql/execution/SparkStrategies.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -409,6 +409,7 @@ private[sql] abstract class SparkStrategies extends QueryPlanner[SparkPlan] {
case logical.RepartitionByExpression(expressions, child, nPartitions) =>
exchange.ShuffleExchange(HashPartitioning(
expressions, nPartitions.getOrElse(numPartitions)), planLater(child)) :: Nil
case ExistingRDD(outputObjAttr, rdd) => ExistingRDDScanExec(outputObjAttr, rdd) :: Nil
case LogicalRDD(output, rdd) => RDDScanExec(output, rdd, "ExistingRDD") :: Nil
case BroadcastHint(child) => planLater(child) :: Nil
case _ => Nil
Expand Down
8 changes: 7 additions & 1 deletion sql/core/src/test/scala/org/apache/spark/sql/QueryTest.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,7 +28,7 @@ import org.apache.spark.sql.catalyst.plans._
import org.apache.spark.sql.catalyst.plans.logical._
import org.apache.spark.sql.catalyst.trees.TreeNode
import org.apache.spark.sql.catalyst.util._
import org.apache.spark.sql.execution.LogicalRDD
import org.apache.spark.sql.execution._
import org.apache.spark.sql.execution.aggregate.TypedAggregateExpression
import org.apache.spark.sql.execution.columnar.InMemoryRelation
import org.apache.spark.sql.execution.datasources.LogicalRelation
Expand Down Expand Up @@ -241,6 +241,12 @@ abstract class QueryTest extends PlanTest {
case _: LogicalRelation => return
case p if p.getClass.getSimpleName == "MetastoreRelation" => return
case _: MemoryPlan => return
case p: InMemoryRelation =>
p.child.transform {
case _: ObjectConsumerExec => return
case _: ObjectProducerExec => return
}
p
}.transformAllExpressions {
case a: ImperativeAggregate => return
case _: TypedAggregateExpression => return
Expand Down