Is there any way to write map only Hadoop jobs

Question

I'm a newbie to Hadoop and I've just started learning Hadoop MapReduce programming. I faced a problem here, at times I only need a map only job because I need map result as the output. So, reduce phase is not required. Can anyone tell me how can I run a map only job in Hadoop?

Thanks in advance!

Answer 1

You can easily set the number of 'reducers' just by changing few parameters in your program.

One way to achieve this is by updating the driver program and set the setNumReduceTasks to the desired value on the job object.

job.setNumReduceTasks(0);

Alternatively, you can set below property as a command line parameter by using -D option

-D mapred.reduce.tasks=0
 

For example you can use it like as mentioned in the below command: 

hadoop jar hadoop-mapreduce-examples.jar wordcount /InputDir /OutputDir