Spark RDD、DataFrame和DataSet的区别

来源：互联网发布：牛津大学古典学数据库编辑：程序博客网时间：2024/05/16 05:18

转自：http://blog.csdn.net/wo334499/article/details/51689549

RDD

优点:

编译时类型安全
编译时就能检查出类型错误
面向对象的编程风格
直接通过类名点的方式来操作数据

缺点:

序列化和反序列化的性能开销
无论是集群间的通信, 还是IO操作都需要对对象的结构和数据进行序列化和反序列化.
GC的性能开销
频繁的创建和销毁对象, 势必会增加GC

import org.apache.spark.sql.SQLContextimport org.apache.spark.{SparkConf, SparkContext}object Run {  def main(args: Array[String]) {    val conf = new SparkConf().setAppName("test").setMaster("local")    val sc = new SparkContext(conf)    sc.setLogLevel("WARN")    val sqlContext = new SQLContext(sc)    /**      * id      age      * 1       30      * 2       29      * 3       21      */    case class Person(id: Int, age: Int)    val idAgeRDDPerson = sc.parallelize(Array(Person(1, 30), Person(2, 29), Person(3, 21)))    // 优点1    // idAge.filter(_.age > "") // 编译时报错, int不能跟String比    // 优点2    idAgeRDDPerson.filter(_.age > 25) // 直接操作一个个的person对象  }}1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

DataFrame

DataFrame引入了schema和off-heap

schema : RDD每一行的数据, 结构都是一样的. 这个结构就存储在schema中. Spark通过schame就能够读懂数据, 因此在通信和IO时就只需要序列化和反序列化数据, 而结构的部分就可以省略了.
off-heap : 意味着JVM堆以外的内存, 这些内存直接受操作系统管理（而不是JVM）。Spark能够以二进制的形式序列化数据(不包括结构)到off-heap中, 当要操作数据时, 就直接操作off-heap内存. 由于Spark理解schema, 所以知道该如何操作.

off-heap就像地盘, schema就像地图, Spark有地图又有自己地盘了, 就可以自己说了算了, 不再受JVM的限制, 也就不再收GC的困扰了.

通过schema和off-heap, DataFrame解决了RDD的缺点, 但是却丢了RDD的优点. DataFrame不是类型安全的, API也不是面向对象风格的.

import org.apache.spark.sql.types.{DataTypes, StructField, StructType}import org.apache.spark.sql.{Row, SQLContext}import org.apache.spark.{SparkConf, SparkContext}object Run {  def main(args: Array[String]) {    val conf = new SparkConf().setAppName("test").setMaster("local")    val sc = new SparkContext(conf)    sc.setLogLevel("WARN")    val sqlContext = new SQLContext(sc)    /**      * id      age      * 1       30      * 2       29      * 3       21      */    val idAgeRDDRow = sc.parallelize(Array(Row(1, 30), Row(2, 29), Row(4, 21)))    val schema = StructType(Array(StructField("id", DataTypes.IntegerType), StructField("age", DataTypes.IntegerType)))    val idAgeDF = sqlContext.createDataFrame(idAgeRDDRow, schema)    // API不是面向对象的    idAgeDF.filter(idAgeDF.col("age") > 25)     // 不会报错, DataFrame不是编译时类型安全的    idAgeDF.filter(idAgeDF.col("age") > "")   }}1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

DataSet

DataSet结合了RDD和DataFrame的优点, 并带来的一个新的概念Encoder

当序列化数据时, Encoder产生字节码与off-heap进行交互, 能够达到按需访问数据的效果, 而不用反序列化整个对象. Spark还没有提供自定义Encoder的API, 但是未来会加入.

下面看DataFrame和DataSet在2.0.0-preview中的实现

下面这段代码, 在1.6.x中创建的是DataFrame

<code class="hljs mathematica has-numbering" style="display: block; padding: 0px; color: inherit; box-sizing: border-box; font-family: 'Source Code Pro', monospace;font-size:undefined; white-space: pre; border-radius: 0px; word-wrap: normal; background: transparent;">// 上文DataFrame示例中提取出来的val idAgeRDDRow = sc.parallelize(<span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">Array</span>(<span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">Row</span>(<span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">1</span>, <span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">30</span>), <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">Row</span>(<span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">2</span>, <span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">29</span>), <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">Row</span>(<span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">4</span>, <span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">21</span>)))val schema = StructType(<span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">Array</span>(StructField(<span class="hljs-string" style="color: rgb(0, 136, 0); box-sizing: border-box;">"id"</span>, DataTypes.IntegerType), StructField(<span class="hljs-string" style="color: rgb(0, 136, 0); box-sizing: border-box;">"age"</span>, DataTypes.IntegerType)))val idAgeDF = sqlContext.createDataFrame(idAgeRDDRow, schema)</code><ul class="pre-numbering" style="box-sizing: border-box; position: absolute; width: 50px; top: 0px; left: 0px; margin: 0px; padding: 6px 0px 40px; border-right-width: 1px; border-right-style: solid; border-right-color: rgb(221, 221, 221); list-style: none; text-align: right; background-color: rgb(238, 238, 238);"><li style="box-sizing: border-box; padding: 0px 5px;">1</li><li style="box-sizing: border-box; padding: 0px 5px;">2</li><li style="box-sizing: border-box; padding: 0px 5px;">3</li><li style="box-sizing: border-box; padding: 0px 5px;">4</li><li style="box-sizing: border-box; padding: 0px 5px;">5</li><li style="box-sizing: border-box; padding: 0px 5px;">6</li></ul>

但是同样的代码在2.0.0-preview中, 创建的虽然还叫DataFrame

<code class="hljs python has-numbering" style="display: block; padding: 0px; color: inherit; box-sizing: border-box; font-family: 'Source Code Pro', monospace;font-size:undefined; white-space: pre; border-radius: 0px; word-wrap: normal; background: transparent;">// sqlContext.createDataFrame(idAgeRDDRow, schema) 方法的实现, 返回值依然是DataFrame<span class="hljs-function" style="box-sizing: border-box;"><span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">def</span> <span class="hljs-title" style="box-sizing: border-box;">createDataFrame</span><span class="hljs-params" style="color: rgb(102, 0, 102); box-sizing: border-box;">(rowRDD: RDD[Row], schema: StructType)</span>:</span> DataFrame = {sparkSession.createDataFrame(rowRDD, schema)}</code><ul class="pre-numbering" style="box-sizing: border-box; position: absolute; width: 50px; top: 0px; left: 0px; margin: 0px; padding: 6px 0px 40px; border-right-width: 1px; border-right-style: solid; border-right-color: rgb(221, 221, 221); list-style: none; text-align: right; background-color: rgb(238, 238, 238);"><li style="box-sizing: border-box; padding: 0px 5px;">1</li><li style="box-sizing: border-box; padding: 0px 5px;">2</li><li style="box-sizing: border-box; padding: 0px 5px;">3</li><li style="box-sizing: border-box; padding: 0px 5px;">4</li></ul>

但是其实却是DataSet, 因为DataFrame被声明为Dataset[Row]

<code class="hljs r has-numbering" style="display: block; padding: 0px; color: inherit; box-sizing: border-box; font-family: 'Source Code Pro', monospace;font-size:undefined; white-space: pre; border-radius: 0px; word-wrap: normal; background: transparent;">package object sql {  // <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">...</span>省略了不相关的代码  type DataFrame = Dataset[Row]}</code><ul class="pre-numbering" style="box-sizing: border-box; position: absolute; width: 50px; top: 0px; left: 0px; margin: 0px; padding: 6px 0px 40px; border-right-width: 1px; border-right-style: solid; border-right-color: rgb(221, 221, 221); list-style: none; text-align: right; background-color: rgb(238, 238, 238);"><li style="box-sizing: border-box; padding: 0px 5px;">1</li><li style="box-sizing: border-box; padding: 0px 5px;">2</li><li style="box-sizing: border-box; padding: 0px 5px;">3</li><li style="box-sizing: border-box; padding: 0px 5px;">4</li><li style="box-sizing: border-box; padding: 0px 5px;">5</li></ul>

因此当我们从1.6.x迁移到2.0.0的时候, 无需任何修改就直接用上了DataSet.

下面是一段DataSet的示例代码

<code class="hljs scala has-numbering" style="display: block; padding: 0px; color: inherit; box-sizing: border-box; font-family: 'Source Code Pro', monospace;font-size:undefined; white-space: pre; border-radius: 0px; word-wrap: normal; background: transparent;"><span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">import</span> org.apache.spark.sql.types.{DataTypes, StructField, StructType}<span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">import</span> org.apache.spark.sql.{Row, SQLContext}<span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">import</span> org.apache.spark.{SparkConf, SparkContext}<span class="hljs-class" style="box-sizing: border-box;"><span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">object</span> <span class="hljs-title" style="box-sizing: border-box; color: rgb(102, 0, 102);">Test</span> {</span>  <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">def</span> main(args: Array[String]) {    <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">val</span> conf = <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">new</span> SparkConf().setAppName(<span class="hljs-string" style="color: rgb(0, 136, 0); box-sizing: border-box;">"test"</span>).setMaster(<span class="hljs-string" style="color: rgb(0, 136, 0); box-sizing: border-box;">"local"</span>) <span class="hljs-comment" style="color: rgb(136, 0, 0); box-sizing: border-box;">// 调试的时候一定不要用local[*]</span>    <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">val</span> sc = <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">new</span> SparkContext(conf)    <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">val</span> sqlContext = <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">new</span> SQLContext(sc)    <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">import</span> sqlContext.implicits._    <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">val</span> idAgeRDDRow = sc.parallelize(Array(Row(<span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">1</span>, <span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">30</span>), Row(<span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">2</span>, <span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">29</span>), Row(<span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">4</span>, <span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">21</span>)))    <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">val</span> schema = StructType(Array(StructField(<span class="hljs-string" style="color: rgb(0, 136, 0); box-sizing: border-box;">"id"</span>, DataTypes.IntegerType), StructField(<span class="hljs-string" style="color: rgb(0, 136, 0); box-sizing: border-box;">"age"</span>, DataTypes.IntegerType)))    <span class="hljs-comment" style="color: rgb(136, 0, 0); box-sizing: border-box;">// 在2.0.0-preview中这行代码创建出的DataFrame, 其实是DataSet[Row]</span>    <span class="hljs-keyword" style="color: rgb(0, 0, 136); box-sizing: border-box;">val</span> idAgeDS = sqlContext.createDataFrame(idAgeRDDRow, schema)    <span class="hljs-comment" style="color: rgb(136, 0, 0); box-sizing: border-box;">// 在2.0.0-preview中, 还不支持自定的Encoder, Row类型不行, 自定义的bean也不行</span>    <span class="hljs-comment" style="color: rgb(136, 0, 0); box-sizing: border-box;">// 官方文档也有写通过bean创建Dataset的例子，但是我运行时并不能成功</span>    <span class="hljs-comment" style="color: rgb(136, 0, 0); box-sizing: border-box;">// 所以目前需要用创建DataFrame的方法, 来创建DataSet[Row]</span>    <span class="hljs-comment" style="color: rgb(136, 0, 0); box-sizing: border-box;">// sqlContext.createDataset(idAgeRDDRow)</span>    <span class="hljs-comment" style="color: rgb(136, 0, 0); box-sizing: border-box;">// 目前支持String, Integer, Long等类型直接创建Dataset</span>    Seq(<span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">1</span>, <span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">2</span>, <span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">3</span>).toDS().show()    sqlContext.createDataset(sc.parallelize(Array(<span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">1</span>, <span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">2</span>, <span class="hljs-number" style="color: rgb(0, 102, 102); box-sizing: border-box;">3</span>))).show()  }}</code>

1 0