Joining Spark RDD's

Hi Friends,

Today I will be demonstrating, how you can perform joins on Spark RDD's. We are going to focus on three basic join operations.

1. Join (Inner)
2.Left Outer Join
3. Right Outer Join

Lets take standard Employee+Department example and create a two RDDs;one holding employee data and another holding department data.

Employee Table :
Eid EName LName
101 Sam Flam
102 Scot Rut
103 Jass Tez 

 val EmpRDD = sc.parallelize(Seq((101,"Sam","Flam"),(102,"Scot","Rut"),(103,"Jas","Tez")))  

Array[(Int, String, String)] = Array((101,Sam,Flam), (102,Scot,Rut), (103,Jas,Tez)) // output


Department Table :
DeptId DepartmentName Eid
D01 Computer 101
D02 Electronic 104
D03 Civil 102 

 val DeptRDD = sc.parallelize(Seq(("D01","Computer",101),("D02","Electronic",104),("D03","Civil",102)))  

Array[(String, String, Int)] = Array((D01,Computer,101), (D02,Electronic,104), (D03,Civil,102)) // output


Now as we know for joining two table, we need a key. So lets format our RDD to have key for performing join operation.


 val EmpWithKeyRDD = EmpRDD.keyBy(x => x._1)  // Forms a new RDD with key as Eid 
  val DetpWithKeyRDD = DeptRDD.keyBy(x => x._3)

Output after Adding Key :

 
Array[(Int, (Int, String, String))] = Array((101,(101,Sam,Flam)), (102,(102,Scot,Rut)), (103,(103,Jas,Tez)))

Array[(Int, (String, String, Int))] = Array((101,(D01,Computer,101)), (104,(D02,Electronic,104)), (102,(D03,Civil,102)))

We are good to go now. Lets perform join operation on two RDD's

 1. Inner Join
val EmpDeptJoin =  EmpWithKeyRDD.join(DetpWithKeyRDD) 

EmpDeptJoin .collect()  // to Check Result

Output :
 Array[(Int, ((Int, String, String), (String, String, Int)))] = Array((101,((101,Sam,Flam),(D01,Computer,101))), (102,((102,Scot,Rut),(D03,Civil,102))))

2.Left Outer Join

val EmpDeptLeftJoin = EmpWithKeyRDD.leftOuterJoin(DetpWithKeyRDD)

EmpDeptLeftJoin.collect()  // to Check Result

Output :

Array[(Int, ((Int, String, String), Option[(String, String, Int)]))] = Array((101,((101,Sam,Flam),Some((D01,Computer,101)))), (102,((102,Scot,Rut),Some((D03,Civil,102)))), (103,((103,Jas,Tez),None)))

3.Left Outer Join


val EmpDeptRightJoin = EmpWithKeyRDD.rightOuterJoin(DetpWithKeyRDD) 

EmpDeptRightJoin .collect()  // to Check Result


Output :
Array[(Int, (Option[(Int, String, String)], (String, String, Int)))] = Array((101,(Some((101,Sam,Flam)),(D01,Computer,101))), (104,(None,(D02,Electronic,104))), (102,(Some((102,Scot,Rut)),(D03,Civil,102))))

 We can perform formatting on data that we received as output from each action. Lets take output from inner join and bring in proper format.

 
val EmpDeptJoinFormatted =  EmpDeptJoin.map(x => (x._1,x._2._1._2,x._2._1._3,x._2._2._1,x._2._2._2))

Output :
Array[(Int, String, String, String, String)] = Array((101,Sam,Flam,D01,Computer), (102,Scot,Rut,D03,Civil))
Explanation of operation performed using map function on each element of RDD.

Looking at diagram , it is clear that how each element is formatted for creating final RDD.

Complete Code


package org.com.td.sparkdemo.spark
import org.apache.spark.SparkConf
import org.apache.spark.SparkContext

object JoinRDD { 
  
val conf = new SparkConf() 
      .setAppName("JoinRDD")
      .setMaster("local")
val sc = new SparkContext(conf) 
  
val EmpRDD = sc.parallelize(Seq((101,"Sam","Flam"),(102,"Scot","Rut"),(103,"Jas","Tez")))   
val DeptRDD = sc.parallelize(Seq(("D01","Computer",101),("D02","Electronic",104),("D03","Civil",102)))
 
val EmpWithKeyRDD  = EmpRDD.keyBy(x => x._1)
val DetpWithKeyRDD = DeptRDD.keyBy(x => x._3)

 
val EmpDeptJoin =  EmpWithKeyRDD.join(DetpWithKeyRDD) 
val EmpDeptLeftJoin = EmpWithKeyRDD.leftOuterJoin(DetpWithKeyRDD)
val EmpDeptRightJoin = EmpWithKeyRDD.rightOuterJoin(DetpWithKeyRDD)  

val EmpDeptJoinFormatted =  EmpDeptJoin.map(x => (x._1,x._2._1._2,x._2._1._3,x._2._2._1,x._2._2._2))

EmpDeptJoinFormatted.saveAsTextFile("/path/to/your/output/directory")
  
}


Enjoy...Happy Coding...!!!!!!

Commentaires

Enregistrer un commentaire

Posts les plus consultés de ce blog

Controlling Parallelism in Spark by controlling the input partitions by controlling the input partitions

Image
It is useful to be able to control the degree of parallelism when using Spark. Spark provides a very convenient method to increase the degree of parallelism which should be adequate in practice. This blog entry of the "Inside Spark" series describes the knobs Spark uses to control the degree of parallelism. Controlling the number of partitions in the " local " mode Spark is designed ground up to enable unit testing. To control the degree of parallelism in the local mode simply utilize the function JavaRDD<String> org.apache.spark.api.java.JavaSparkContext.parallelize(List<String> list, int numSlices) The following snippet of code demonstrates this usage- public class SparkControlPartitionSizeLocally { public static void main ( String [] args ) { JavaSparkContext sc = new JavaSparkContext ( "local" , "localpartitionsizecontrol" ); String input = "four score and sev...
Lire la suite

Spark performance optimization: shuffle tuning

Image
Article directory 1 shuffle tuning 1.1 Summary of tuning 1.2 Overview of ShuffleManager Development 1.3 HashShuffleManager operating principle 1.3.1 Unoptured HashShuffleManager 1.3.2 Optimized HashShuffleManager 1.4 SortShuffleManager operating principle 1.4.1 General operating mechanism 1.4.2 bypass running mechanism 1.5 shuffle related parameters tuning 1.5.1 spark.shuffle.file.buffer 1.5.2 spark.reducer.maxSizeInFlight 1.5.3 spark.shuffle.io.maxRetries 1.5.4 spark.shuffle.io.retryWait 1.5.5 spark.shuffle.memoryFraction 1.5.6 spark.shuffle.manager 1.5.7 spark.shuffle.sort.bypassMergeThreshold 1.5.8 spark.shuffle.consolidateFiles 2 write in the last words Shuffle Summary of tuning Most of the performance of Spark operations is mainly consumed in the shuffle link, because the link contains a large number of disk IO, serialization, network data transmission and othe...
Lire la suite

Spark optimization

Image
One operation and maintenance 1. Master hang up, standby restart is also invalid Master defaults to 512M of memory, when the task in the cluster is particularly high, it will hang, because the master will read each task event log log to generate spark ui, the memory will naturally OOM, you can run the log See that the master of the start through the HA will naturally fail for this reason. solve Increase the Master's memory spark-env.sh , set in the master node spark-env.sh : export SPARK_DAEMON_MEMORY 10g # 根据你的实际情况 Reduce the job information stored in the Master memory spark.ui.retainedJobs 500 # 默认都是1000 spark.ui.retainedStages 500 Hang up or suspend Sometimes we will see the web node in the web ui disappear or in the dead state, the task of running the node will report a variety of lost worker errors, causing the same reasons and the above, worker memory to save a lot of ui The information leads to gc when the heartbeat is lost...
Lire la suite