Run WordCount with Scala and Spark on HDInsight

Previously we tried to solve the word count problem with a Scala and Spark approach.
The next step is to deploy our solution to HDInsight using spark, hdfs, and scala

We shall provision a Sprak cluster.

screenshot-from-2017-02-22-23-12-22

Since we are going to use HDInsight we can utilize hdfs and therefore use the azure storage.

screenshot-from-2017-02-22-23-12-59

Then we choose our instance types.

screenshot-from-2017-02-22-23-13-21

And we are ready to create the Spark cluster.

screenshot-from-2017-02-22-23-13-55

Our data shall be uploaded to the hdfs file system
To do so we will upload our text files to the azure storage account which is integrated with hdfs.

For more information on managing a storage account with azure cli check the official guide. Any text file will work.

azure storage blob upload mytextfile.txt sparkclusterscala  example/data/mytextfile.txt

Since we use hdfs we shall make some changes to the original script

val text = sc.textFile("wasb:///example/data/mytextfile.txt")
val counts = text.flatMap(line => line.split(" ")).map(word => (word,1)).reduceByKey(_+_)
counts.collect

Then we can upload our scala class to the head node using ssh

scp WordCountscala.scala demon@{your cluster}-ssh.azurehdinsight.net:/home/demo/WordCountscala.scala

Again in order to run the script, things are pretty straightforward.

spark-shell -i WordCountscala.scala

And once the task is done we are presented with the spark prompt. Plus we can now save our results to the hdfs file system.

scala> counts.saveAsTextFile("/wordcount_results")

And do a quick check.

hdfs dfs -ls /wordcount_results/
hdfs dfs -text /wordcount_results/part-00000

WordCount with Sprak and Scala

Apache Spark has taken over the big data world. Spark is implemented with Scala and is well know for its performance.

In the previous blogs we approached the word count problem by using Scala with hadoop and Scala with storm.
On this blog we will utilize Spark for the word count problem.

Submitting spark jobs implemented with Scala is pretty easy and convenient. All we need is to submit our file as our input to the spark command.

First we have to download and setup a spark version locally.

Then will shall download a text file for testing. In my case the script from MGS2 did the work.

Now on to the WordCount script. For local testing we will use a file from our file system.

val text = sc.textFile("mytextfile.txt")
val counts = text.flatMap(line => line.split(" ")).map(word => (word,1)).reduceByKey(_+_)
counts.collect

Next step is to run the script

spark-shell -i WordCountscala.scala

Once finished a Spark command prompt will appear and we are free to do some experiments with the word count results

Welcome to
      ____              __
     / __/__  ___ _____/ /__
    _\ \/ _ \/ _ `/ __/  '_/
   /___/ .__/\_,_/_/ /_/\_\   version 2.1.0
      /_/
         
Using Scala version 2.11.8 (Java HotSpot(TM) 64-Bit Server VM, Java 1.8.0_111)
Type in expressions to have them evaluated.
Type :help for more information.

scala> res0.length
res1: Int = 20159

Thus we detected 20159 different words.

Our next step is to run our job to a spark cluster on HDInsight.