PySpark - Broadcast & Accumulator

For parallel processing, Apache Spark uses shared variables. A copy of shared variable goes on each node of the cluster when the driver sends a task to the executor on the cluster, so that it can be used for performing tasks.

There are two types of shared variables supported by Apache Spark −

  • Broadcast
  • Accumulator

Let us understand them in detail.

Broadcast

Broadcast variables are used to save the copy of data across all nodes. This variable is cached on all the machines and not sent on machines with tasks. The following code block has the details of a Broadcast class for PySpark.

class pyspark.Broadcast (
   sc = None, 
   value = None, 
   pickle_registry = None, 
   path = None
)

The following example shows how to use a Broadcast variable. A Broadcast variable has an attribute called value, which stores the data and is used to return a broadcasted value.

----------------------------------------broadcast.py--------------------------------------
from pyspark import SparkContext 
sc = SparkContext("local", "Broadcast app") 
words_new = sc.broadcast(["scala", "java", "hadoop", "spark", "akka"]) 
data = words_new.value 
print "Stored data -> %s" % (data) 
elem = words_new.value[2] 
print "Printing a particular element in RDD -> %s" % (elem)
----------------------------------------broadcast.py--------------------------------------

Command − The command for a broadcast variable is as follows −

$SPARK_HOME/bin/spark-submit broadcast.py

Output − The output for the following command is given below.

Stored data -> [
   'scala',  
   'java', 
   'hadoop', 
   'spark', 
   'akka'
]
Printing a particular element in RDD -> hadoop

Accumulator

Accumulator variables are used for aggregating the information through associative and commutative operations. For example, you can use an accumulator for a sum operation or counters (in MapReduce). The following code block has the details of an Accumulator class for PySpark.

class pyspark.Accumulator(aid, value, accum_param)

The following example shows how to use an Accumulator variable. An Accumulator variable has an attribute called value that is similar to what a broadcast variable has. It stores the data and is used to return the accumulator's value, but usable only in a driver program.

In this example, an accumulator variable is used by multiple workers and returns an accumulated value.

----------------------------------------accumulator.py------------------------------------
from pyspark import SparkContext 
sc = SparkContext("local", "Accumulator app") 
num = sc.accumulator(10) 
def f(x): 
   global num 
   num+=x 
rdd = sc.parallelize([20,30,40,50]) 
rdd.foreach(f) 
final = num.value 
print "Accumulated value is -> %i" % (final)
----------------------------------------accumulator.py------------------------------------

Command − The command for an accumulator variable is as follows −

$SPARK_HOME/bin/spark-submit accumulator.py

Output − The output for the above command is given below.

Accumulated value is -> 150
Advertisements