Dataset — Structured Query with Data Encoder-spark技术分享

Dataset — Structured Query with Data Encoder

Dataset is a strongly-typed data structure in Spark SQL that represents a structured query.

Note	A structured query can be written using SQL or Dataset API.

The following figure shows the relationship between different entities of Spark SQL that all together give the Dataset data structure.

Figure 1. Dataset’s Internals

It is therefore fair to say that Dataset consists of the following three elements:

QueryExecution (with the parsed unanalyzed LogicalPlan of a structured query)
Encoder (of the type of the records for fast serialization and deserialization to and from InternalRow)
SparkSession

When created, Dataset takes such a 3-element tuple with a SparkSession, a QueryExecution and an Encoder.

Dataset is created when:

Dataset.apply (for a LogicalPlan and a SparkSession with the Encoder in a Scala implicit scope)
Dataset.ofRows (for a LogicalPlan and a SparkSession)
Dataset.toDF untyped transformation is used
Dataset.select, Dataset.randomSplit and Dataset.mapPartitions typed transformations are used
KeyValueGroupedDataset.agg operator is used (that requests KeyValueGroupedDataset to aggUntyped)
SparkSession.emptyDataset and SparkSession.range operators are used
CatalogImpl is requested to
makeDataset (when requested to list databases, tables, functions and columns)
Spark Structured Streaming’s MicroBatchExecution is requested to runBatch

Datasets are lazy and structured query operators and expressions are only triggered when an action is invoked.



import org.apache.spark.sql.SparkSession
val spark: SparkSession = ...

scala> val dataset = spark.range(5)
dataset: org.apache.spark.sql.Dataset[Long] = [id: bigint]

// Variant 1: filter operator accepts a Scala function
dataset.filter(n => n % 2 == 0).count

// Variant 2: filter operator accepts a Column-based SQL expression
dataset.filter('value % 2 === 0).count

// Variant 3: filter operator accepts a SQL query
dataset.filter("value % 2 = 0").count

import org.apache.spark.sql.SparkSession

val spark: SparkSession = ...

scala> val dataset = spark.range(5)

dataset: org.apache.spark.sql.Dataset[Long] = [id: bigint]

// Variant 1: filter operator accepts a Scala function

dataset.filter(n => n % 2 == 0).count

// Variant 2: filter operator accepts a Column-based SQL expression

dataset.filter('value % 2 === 0).count

// Variant 3: filter operator accepts a SQL query

dataset.filter("value % 2 = 0").count

The Dataset API offers declarative and type-safe operators that makes for an improved experience for data processing (comparing to DataFrames that were a set of index- or column name-based Rows).

Note

Dataset was first introduced in Apache Spark 1.6.0 as an experimental feature, and has since turned itself into a fully supported API.

As of Spark 2.0.0, DataFrame – the flagship data abstraction of previous versions of Spark SQL – is currently a mere type alias for Dataset[Row]:



type DataFrame = Dataset[Row]

type DataFrame = Dataset[Row]

See package object sql.

Dataset offers convenience of RDDs with the performance optimizations of DataFrames and the strong static type-safety of Scala. The last feature of bringing the strong type-safety to DataFrame makes Dataset so appealing. All the features together give you a more functional programming interface to work with structured data.



scala> spark.range(1).filter('id === 0).explain(true)
== Parsed Logical Plan ==
'Filter ('id = 0)
+- Range (0, 1, splits=8)

== Analyzed Logical Plan ==
id: bigint
Filter (id#51L = cast(0 as bigint))
+- Range (0, 1, splits=8)

== Optimized Logical Plan ==
Filter (id#51L = 0)
+- Range (0, 1, splits=8)

== Physical Plan ==
*Filter (id#51L = 0)
+- *Range (0, 1, splits=8)

scala> spark.range(1).filter(_ == 0).explain(true)
== Parsed Logical Plan ==
'TypedFilter <function1>, class java.lang.Long, [StructField(value,LongType,true)], unresolveddeserializer(newInstance(class java.lang.Long))
+- Range (0, 1, splits=8)

== Analyzed Logical Plan ==
id: bigint
TypedFilter <function1>, class java.lang.Long, [StructField(value,LongType,true)], newInstance(class java.lang.Long)
+- Range (0, 1, splits=8)

== Optimized Logical Plan ==
TypedFilter <function1>, class java.lang.Long, [StructField(value,LongType,true)], newInstance(class java.lang.Long)
+- Range (0, 1, splits=8)

== Physical Plan ==
*Filter <function1>.apply
+- *Range (0, 1, splits=8)

scala> spark.range(1).filter('id === 0).explain(true)

== Parsed Logical Plan ==

'Filter ('id = 0)

+- Range (0, 1, splits=8)

== Analyzed Logical Plan ==

id: bigint

Filter (id#51L = cast(0 as bigint))

+- Range (0, 1, splits=8)

== Optimized Logical Plan ==

Filter (id#51L = 0)

+- Range (0, 1, splits=8)

== Physical Plan ==

*Filter (id#51L = 0)

+- *Range (0, 1, splits=8)

scala> spark.range(1).filter(_ == 0).explain(true)

== Parsed Logical Plan ==

'TypedFilter <function1>, class java.lang.Long, [StructField(value,LongType,true)], unresolveddeserializer(newInstance(class java.lang.Long))

+- Range (0, 1, splits=8)

== Analyzed Logical Plan ==

id: bigint

TypedFilter <function1>, class java.lang.Long, [StructField(value,LongType,true)], newInstance(class java.lang.Long)

+- Range (0, 1, splits=8)

== Optimized Logical Plan ==

TypedFilter <function1>, class java.lang.Long, [StructField(value,LongType,true)], newInstance(class java.lang.Long)

+- Range (0, 1, splits=8)

== Physical Plan ==

*Filter <function1>.apply

+- *Range (0, 1, splits=8)

It is only with Datasets to have syntax and analysis checks at compile time (that was not possible using DataFrame, regular SQL queries or even RDDs).

Using Dataset objects turns DataFrames of Row instances into a DataFrames of case classes with proper names and types (following their equivalents in the case classes). Instead of using indices to access respective fields in a DataFrame and cast it to a type, all this is automatically handled by Datasets and checked by the Scala compiler.

If however a LogicalPlan is used to create a Dataset, the logical plan is first executed (using the current SessionState in the SparkSession) that yields the QueryExecution plan.

A Dataset is Queryable and Serializable, i.e. can be saved to a persistent storage.

Note	SparkSession and QueryExecution are transient attributes of a `Dataset` and therefore do not participate in Dataset serialization. The only firmly-tied feature of a `Dataset` is the Encoder.

You can request the “untyped” view of a Dataset or access the RDD that is generated after executing the query. It is supposed to give you a more pleasant experience while transitioning from the legacy RDD-based or DataFrame-based APIs you may have used in the earlier versions of Spark SQL or encourage migrating from Spark Core’s RDD API to Spark SQL’s Dataset API.

The default storage level for Datasets is MEMORY_AND_DISK because recomputing the in-memory columnar representation of the underlying table is expensive. You can however persist a Dataset.

Note	Spark 2.0 has introduced a new query model called Structured Streaming for continuous incremental execution of structured queries. That made possible to consider Datasets a static and bounded as well as streaming and unbounded data sets with a single unified API for different execution models.

A Dataset is local if it was created from local collections using SparkSession.emptyDataset or SparkSession.createDataset methods and their derivatives like toDF. If so, the queries on the Dataset can be optimized and run locally, i.e. without using Spark executors.

Note	`Dataset` makes sure that the underlying `QueryExecution` is analyzed and checked.

Name Description

boundEnc

ExpressionEncoder

Used when…FIXME

deserializer

Deserializer expression to convert internal rows to objects of type T

Created lazily by requesting the ExpressionEncoder to resolveAndBind

Used when:

Dataset is created (for a logical plan in a given SparkSession)
Dataset.toLocalIterator operator is used (to create a Java Iterator of objects of type T)
Dataset is requested to collect all rows from a spark plan

exprEnc

Implicit ExpressionEncoder

Used when…FIXME

logicalPlan

Analyzed logical plan with all logical commands executed and turned into a LocalRelation.



logicalPlan: LogicalPlan

logicalPlan: LogicalPlan

When initialized, logicalPlan requests the QueryExecution for analyzed logical plan. If the plan is a logical command or a union thereof, logicalPlan executes the QueryExecution (using executeCollect).

planWithBarrier



planWithBarrier: AnalysisBarrier

planWithBarrier: AnalysisBarrier

rdd

(lazily-created) RDD of JVM objects of type T (as converted from rows in Dataset in the internal binary row format).



rdd: RDD[T]

rdd: RDD[T]

Note	`rdd` gives `RDD` with the extra execution step to convert rows from their internal binary row format to JVM objects that will impact the JVM memory as the objects are inside JVM (while were outside before). You should not use `rdd` directly.

Internally, rdd first creates a new logical plan that deserializes the Dataset’s logical plan.



val dataset = spark.range(5).withColumn("group", 'id % 2)
scala> dataset.rdd.toDebugString
res1: String =
(8) MapPartitionsRDD[8] at rdd at <console>:26 [] // <-- extra deserialization step
 |  MapPartitionsRDD[7] at rdd at <console>:26 []
 |  MapPartitionsRDD[6] at rdd at <console>:26 []
 |  MapPartitionsRDD[5] at rdd at <console>:26 []
 |  ParallelCollectionRDD[4] at rdd at <console>:26 []

scala> dataset.queryExecution.toRdd.toDebugString
res2: String =
(8) MapPartitionsRDD[11] at toRdd at <console>:26 []
 |  MapPartitionsRDD[10] at toRdd at <console>:26 []
 |  ParallelCollectionRDD[9] at toRdd at <console>:26 []

val dataset = spark.range(5).withColumn("group", 'id % 2)

scala> dataset.rdd.toDebugString

res1: String =

(8) MapPartitionsRDD[8] at rdd at <console>:26 [] // <-- extra deserialization step

| MapPartitionsRDD[7] at rdd at <console>:26 []

| MapPartitionsRDD[6] at rdd at <console>:26 []

| MapPartitionsRDD[5] at rdd at <console>:26 []

| ParallelCollectionRDD[4] at rdd at <console>:26 []

scala> dataset.queryExecution.toRdd.toDebugString

res2: String =

(8) MapPartitionsRDD[11] at toRdd at <console>:26 []

| MapPartitionsRDD[10] at toRdd at <console>:26 []

| ParallelCollectionRDD[9] at toRdd at <console>:26 []

rdd then requests SessionState to execute the logical plan to get the corresponding RDD of binary rows.

Note	`rdd` uses SparkSession to access `SessionState`.

rdd then requests the Dataset’s ExpressionEncoder for the data type of the rows (using deserializer expression) and maps over them (per partition) to create records of the expected type T.

Note	`rdd` is at the “boundary” between the internal binary row format and the JVM type of the dataset. Avoid the extra deserialization step to lower JVM memory requirements of your Spark application.

sqlContext

Lazily-created SQLContext

Used when…FIXME

Getting Input Files of Relations (in Structured Query) — `inputFiles` Method



inputFiles: Array[String]

inputFiles: Array[String]

inputFiles requests QueryExecution for optimized logical plan and collects the following logical operators:

LogicalRelation with FileRelation (as the BaseRelation)
FileRelation
HiveTableRelation

inputFiles then requests the logical operators for their underlying files:

inputFiles of the FileRelations
locationUri of the HiveTableRelation

`resolve` Internal Method



resolve(colName: String): NamedExpression

resolve(colName: String): NamedExpression

Caution

FIXME

Creating Dataset Instance

Dataset takes the following when created:

SparkSession
QueryExecution
Encoder for the type T of the records

Note	You can also create a `Dataset` using LogicalPlan that is immediately executed using `SessionState`.

Internally, Dataset requests QueryExecution to analyze itself.

Dataset initializes the internal registries and counters.

Is Dataset Local? — `isLocal` Method



isLocal: Boolean

isLocal: Boolean

isLocal flag is enabled (i.e. true) when operators like collect or take could be run locally, i.e. without using executors.

Internally, isLocal checks whether the logical query plan of a Dataset is LocalRelation.

Is Dataset Streaming? — `isStreaming` method



isStreaming: Boolean

isStreaming: Boolean

isStreaming is enabled (i.e. true) when the logical plan is streaming.

Internally, isStreaming takes the Dataset’s logical plan and gives whether the plan is streaming or not.

Queryable

Caution

FIXME

`withNewRDDExecutionId` Internal Method



withNewRDDExecutionId[U](body: => U): U

withNewRDDExecutionId[U](body: => U): U

withNewRDDExecutionId executes the input body action under new execution id.

Caution

FIXME What’s the difference between withNewRDDExecutionId and withNewExecutionId?

Note	`withNewRDDExecutionId` is used when Dataset.foreach and Dataset.foreachPartition actions are used.

Creating DataFrame (For Logical Query Plan and SparkSession) — `ofRows` Internal Factory Method



ofRows(sparkSession: SparkSession, logicalPlan: LogicalPlan): DataFrame

ofRows(sparkSession: SparkSession, logicalPlan: LogicalPlan): DataFrame

Note	`ofRows` is part of `Dataset` Scala object that is marked as a `private[sql]` and so can only be accessed from code in `org.apache.spark.sql` package.

ofRows returns DataFrame (which is the type alias for Dataset[Row]). ofRows uses RowEncoder to convert the schema (based on the input logicalPlan logical plan).

Internally, ofRows prepares the input logicalPlan for execution and creates a Dataset[Row] with the current SparkSession, the QueryExecution and RowEncoder.

Note

ofRows is used when:

DataFrameReader is requested to load data from a data source
Dataset is requested to execute checkpoint, mapPartitionsInR, untyped transformations and set-based typed transformations
RelationalGroupedDataset is requested to create a DataFrame from aggregate expressions, flatMapGroupsInR and flatMapGroupsInPandas
SparkSession is requested to create a DataFrame from a BaseRelation, createDataFrame, internalCreateDataFrame, sql and table
CacheTableCommand, CreateTempViewUsing, InsertIntoDataSourceCommand and SaveIntoDataSourceCommand logical commands are executed (run)
DataSource is requested to writeAndRead (for a CreatableRelationProvider)
FrequentItems is requested to singlePassFreqItems
StatFunctions is requested to crossTabulate and summary
Spark Structured Streaming’s DataStreamReader is requested to load
Spark Structured Streaming’s DataStreamWriter is requested to start
Spark Structured Streaming’s FileStreamSource is requested to getBatch
Spark Structured Streaming’s MemoryStream is requested to toDF

Tracking Multi-Job Structured Query Execution (PySpark) — `withNewExecutionId` Internal Method



withNewExecutionId[U](body: => U): U

withNewExecutionId[U](body: => U): U

withNewExecutionId executes the input body action under new execution id.

Note	`withNewExecutionId` sets a unique execution id so that all Spark jobs belong to the `Dataset` action execution.

Note

withNewExecutionId is used exclusively when Dataset is executing Python-based actions (i.e. collectToPython, collectAsArrowToPython and toPythonIterator) that are not of much interest in this gitbook.

Feel free to contact me at jacek@japila.pl if you think I should re-consider my decision.

Executing Action Under New Execution ID — `withAction` Internal Method



withAction[U](name: String, qe: QueryExecution)(action: SparkPlan => U)

withAction[U](name: String, qe: QueryExecution)(action: SparkPlan => U)

withAction requests QueryExecution for the optimized physical query plan and resets the metrics of every physical operator (in the physical plan).

withAction requests SQLExecution to execute the input action with the executable physical plan (tracked under a new execution id).

In the end, withAction notifies ExecutionListenerManager that the name action has finished successfully or with an exception.

Note	`withAction` uses SparkSession to access ExecutionListenerManager.

Note	`withAction` is used when `Dataset` is requested for the following: Computing the logical plan (and executing a logical command or their `Union`) Dataset operators: collect, count, head and toLocalIterator

Creating Dataset Instance (For LogicalPlan and SparkSession) — `apply` Internal Factory Method



apply[T: Encoder](sparkSession: SparkSession, logicalPlan: LogicalPlan): Dataset[T]

apply[T: Encoder](sparkSession: SparkSession, logicalPlan: LogicalPlan): Dataset[T]

Note	`apply` is part of `Dataset` Scala object that is marked as a `private[sql]` and so can only be accessed from code in `org.apache.spark.sql` package.

apply…FIXME

Note	`apply` is used when: `Dataset` is requested to execute typed transformations and set-based typed transformations Spark Structured Streaming’s `MemoryStream` is requested to `toDS`

Collecting All Rows From Spark Plan — `collectFromPlan` Internal Method



collectFromPlan(plan: SparkPlan): Array[T]

collectFromPlan(plan: SparkPlan): Array[T]

collectFromPlan…FIXME

Note	`collectFromPlan` is used for Dataset.head, Dataset.collect and Dataset.collectAsList operators.

`selectUntyped` Internal Method



selectUntyped(columns: TypedColumn[_, _]*): Dataset[_]

selectUntyped(columns: TypedColumn[_, _]*): Dataset[_]

selectUntyped…FIXME

Note	`selectUntyped` is used exclusively when Dataset.select typed transformation is used.

Helper Method for Typed Transformations — `withTypedPlan` Internal Method



withTypedPlan[U: Encoder](logicalPlan: LogicalPlan): Dataset[U]

withTypedPlan[U: Encoder](logicalPlan: LogicalPlan): Dataset[U]

withTypedPlan…FIXME

Note	`withTypedPlan` is annotated with Scala’s @inline annotation that requests the Scala compiler to try especially hard to inline it.

Note	`withTypedPlan` is used in the `Dataset` typed transformations, i.e. withWatermark, joinWith, hint, as, filter, limit, sample, dropDuplicates, filter, map, repartition, repartitionByRange, coalesce and sort with sortWithinPartitions (through the sortInternal internal method).

Helper Method for Set-Based Typed Transformations — `withSetOperator` Internal Method



withSetOperator[U: Encoder](logicalPlan: LogicalPlan): Dataset[U]

withSetOperator[U: Encoder](logicalPlan: LogicalPlan): Dataset[U]

withSetOperator…FIXME

Note	`withSetOperator` is annotated with Scala’s @inline annotation that requests the Scala compiler to try especially hard to inline it.

Note	`withSetOperator` is used in the `Dataset` typed transformations, i.e. union, unionByName, intersect and except.

`sortInternal` Internal Method



sortInternal(global: Boolean, sortExprs: Seq[Column]): Dataset[T]

sortInternal(global: Boolean, sortExprs: Seq[Column]): Dataset[T]

sortInternal creates a Dataset with Sort unary logical operator (and the logicalPlan as the child logical plan).



val nums = Seq((0, "zero"), (1, "one")).toDF("id", "name")
// Creates a Sort logical operator:
// - descending sort direction for id column (specified explicitly)
// - name column is wrapped with ascending sort direction
val numsSorted = nums.sort('id.desc, 'name)
val logicalPlan = numsSorted.queryExecution.logical
scala> println(logicalPlan.numberedTreeString)
00 'Sort ['id DESC NULLS LAST, 'name ASC NULLS FIRST], true
01 +- Project [_1#11 AS id#14, _2#12 AS name#15]
02    +- LocalRelation [_1#11, _2#12]

val nums = Seq((0, "zero"), (1, "one")).toDF("id", "name")

// Creates a Sort logical operator:

// - descending sort direction for id column (specified explicitly)

// - name column is wrapped with ascending sort direction

val numsSorted = nums.sort('id.desc, 'name)

val logicalPlan = numsSorted.queryExecution.logical

scala> println(logicalPlan.numberedTreeString)

00 'Sort ['id DESC NULLS LAST, 'name ASC NULLS FIRST], true

01 +- Project [_1#11 AS id#14, _2#12 AS name#15]

02 +- LocalRelation [_1#11, _2#12]

Internally, sortInternal firstly builds ordering expressions for the given sortExprs columns, i.e. takes the sortExprs columns and makes sure that they are SortOrder expressions already (and leaves them untouched) or wraps them into SortOrder expressions with Ascending sort direction.

In the end, sortInternal creates a Dataset with Sort unary logical operator (with the ordering expressions, the given global flag, and the logicalPlan as the child logical plan).

Note	`sortInternal` is used for the sort and sortWithinPartitions typed transformations in the Dataset API (with the only change of the `global` flag being enabled and disabled, respectively).

Helper Method for Untyped Transformations and Basic Actions — `withPlan` Internal Method



withPlan(logicalPlan: LogicalPlan): DataFrame

withPlan(logicalPlan: LogicalPlan): DataFrame

withPlan simply uses ofRows internal factory method to create a DataFrame for the input LogicalPlan and the current SparkSession.

Note	`withPlan` is annotated with Scala’s @inline annotation that requests the Scala compiler to try especially hard to inline it.

Note	`withPlan` is used in the `Dataset` untyped transformations (i.e. join, crossJoin and select) and basic actions (i.e. createTempView, createOrReplaceTempView, createGlobalTempView and createOrReplaceGlobalTempView).

Dataset — Structured Query with Data Encoder