Page Comparison

...

As described earlier, now we load up the Dr Suess story into an RDD that then gets split into another RDD of a single word per element.  I'll explain more in a bit, but let'Let's also cache this RDD into memory to aid our ultimate test later in the blogit to establish a baseline of simply reading through the RDD so that we don't introduce any additional variability in our comparison testing.

scala> val wordsRdd = sc.textFile("GreenEggsAndHam.txt", 3).flatMap(line => line.split(" "))
wordsRdd: org.apache.spark.rdd.RDD[String] = MapPartitionsRDD[2] at flatMap at <console>:27

scala> wordsRdd.persist() //mark it as cached
res0: wordsRdd.type = MapPartitionsRDD[2] at flatMap at <console>:27

scala> sc.setLogLevel("INFO")  //enable timing messages

scala> wordsRdd.take(3)   //trigger cache load
16/05/19 23:08:22 INFO DAGScheduler: Job 0 finished: take at <console>:30, took 0.565019 s
res2: Array[String] = Array(I, am, Daniel)

...

Now that we got an order of magnitude speed improvement, and somewhat consistent response times, we are ready to stand up a test harness to prove that mapPartitions() is faster than map() when the function we are calling produces negative results when call once per record instead of once per partition.  The wrapSingleWord() function is just going to add ">>" before each word in the RDD and slap on "<<" on the backside.  And to really make the point, I've snuck in a bogus method function that burns 10 millis of clock time for each function call which is my simulation of some arbitrary expensive operation.  Just paste in the following functions separately in the shell.

...

scala> wordsRdd.map(word => wrapSingleWord(word)).take(10)
16/05/20 00:18:46 INFO DAGScheduler: Job 7 finished: take at <console>:34, took 0.151721 s
res9: Array[String] = Array(>>I<<, >>am<<, >>Daniel<<, >><<, >>I<<, >>am<<, >>Sam<<, >>Sam<<, >>I<<, >>am<<)

scala> wordsRdd.map(word => wrapSingleWord(word)).take(10)
16/05/20 00:18:52 INFO DAGScheduler: Job 8 finished: take at <console>:34, took 0.146428 s
res10: Array[String] = Array(>>I<<, >>am<<, >>Daniel<<, >><<, >>I<<, >>am<<, >>Sam<<, >>Sam<<, >>I<<, >>am<<)

scala> wordsRdd.map(word => wrapSingleWord(word)).take(10)
16/05/20 00:18:54 INFO DAGScheduler: Job 9 finished: take at <console>:34, took 0.153416 s
res11: Array[String] = Array(>>I<<, >>am<<, >>Daniel<<, >><<, >>I<<, >>am<<, >>Sam<<, >>Sam<<, >>I<<, >>am<<)

In this tightly constrained Sandbox cluster, we are seeing about 0.15 secs to execute this completely.  Now, to get mapPartitions() to work we need another method function that does the same thing, but it has to allow the whole collection of elements to be passed into the function and it needs to return a whole collection on the way back.  The following is just a jazzed up version of the earlier methodfunction.

import java.util._
import scala.collection.JavaConversions._

def wrapMultiWords(words: Iterator[String]) : Iterator[String] = {
 simulateExpensiveObjectCreation()
  val sb = new StringBuilder()
  val wList = new ArrayList[String]()
  while( words.hasNext ) {
    sb.setLength(0)
    wList.add( sb.append(">>").append(words.next()).append("<<").toString() )
  }
  return wList.iterator()
}

...

As you can see, the new RDD is has the same data, but the performance was dramatically better finishing in about a third of the time in this testing scenario.  Obviously, you'll need to do some testing with your own data and the functions that are being used in the mapping transformations, but if you do have any measurable difference in calling the related function, it will surely surface when you have to call it over and over again.  While Spark is not necessarily Hadoop, the need to POC your particular problem and validate your hypothesis is just as important in this space.  With that, you will likely see improvements by moving from map() to mapPartitions() when your related function can process all the elements at once.