spark-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From raggy <raghav0110...@gmail.com>
Subject Different Sorting RDD methods in Apache Spark
Date Tue, 09 Jun 2015 06:30:23 GMT
For a research project, I tried sorting the elements in an RDD. I did this in
two different approaches.

In the first method, I applied a mapPartitions() function on the RDD, so
that it would sort the contents of the RDD, and provide a result RDD that
contains the sorted list as the only record in the RDD. Then, I applied a
reduce function which basically merges sorted lists.

I ran these experiments on an EC2 cluster containing 30 nodes. I set it up
using the spark ec2 script. The data file was stored in HDFS.

In the second approach I used the sortBy method in Spark.

I performed these operation on the US census data(100MB) found here

A single lines looks like this

9, Not in universe, 0, 0, Children, 0, Not in universe, Never married, Not
in universe or children, Not in universe, White, All other, Female, Not in
universe, Not in universe, Children or Armed Forces, 0, 0, 0, Nonfiler, Not
in universe, Not in universe, Child <18 never marr not in subfamily, Child
under 18 never married, 1758.14, Nonmover, Nonmover, Nonmover, Yes, Not in
universe, 0, Both parents present, United-States, United-States,
United-States, Native- Born in the United States, 0, Not in universe, 0, 0,
94, - 50000.
I sorted based on the 25th value in the CSV. In this line that is 1758.14.

I noticed that sortBy performs worse than the other method. Is this the
expected scenario? If it is, why wouldn't the mapPartitions() and reduce()
be the default sorting approach?

Here is my implementation

public static void sortBy(JavaSparkContext sc){
        JavaRDD<String> rdd = sc.textFile("/data.txt",32);
        long start = System.currentTimeMillis();
        rdd.sortBy(new Function<String, Double>(){

            @Override
                public Double call(String v1) throws Exception {
                      // TODO Auto-generated method stub
                  String [] arr = v1.split(",");
                  return Double.parseDouble(arr[24]);   
                }
        }, true, 9).collect();
        long end = System.currentTimeMillis();
        System.out.println("SortBy: " + (end - start));
  }

public static void sortList(JavaSparkContext sc){
        JavaRDD<String> rdd = sc.textFile("/data.txt",32); //parallelize(l,
8);
        long start = System.currentTimeMillis();
        JavaRDD<LinkedList&lt;Tuple2&lt;Double, String>>> rdd3 =
rdd.mapPartitions(new FlatMapFunction<Iterator&lt;String>,
LinkedList<Tuple2&lt;Double, String>>>(){

        @Override
        public Iterable<LinkedList&lt;Tuple2&lt;Double, String>>>
call(Iterator<String> t)
            throws Exception {
          // TODO Auto-generated method stub
          LinkedList<Tuple2&lt;Double, String>> lines = new
LinkedList<Tuple2&lt;Double, String>>();
          while(t.hasNext()){       
            String s = t.next();
            String arr1[] = s.split(",");
            Tuple2<Double, String> t1 = new Tuple2<Double,
String>(Double.parseDouble(arr1[24]),s);
            lines.add(t1);
          }
          Collections.sort(lines, new IncomeComparator());
          LinkedList<LinkedList&lt;Tuple2&lt;Double, String>>> list =
new
LinkedList<LinkedList&lt;Tuple2&lt;Double, String>>>();
          list.add(lines);
          return list;
        }



--
View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/Different-Sorting-RDD-methods-in-Apache-Spark-tp23214.html
Sent from the Apache Spark User List mailing list archive at Nabble.com.

---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscribe@spark.apache.org
For additional commands, e-mail: user-help@spark.apache.org


Mime
View raw message