We focus on testing HDFS’s I/O throughput [15] and MapReduce’s computing
ability when we do experiment to test the performance of Hadoop. In our experiment,
we use WordCount program which Hadoop owns to test the performance of Hadoop.
The WordCount program uses the MapReduce Model. We all know that the process
of WordCount use the MapReduce model as its computing framework and the results
will be stored in the disk. The WordCount program can also read the results into
memory from the disk for computing. So we can test both the HDFS’s I/O performance
and the MapReduce computing ability by the WordCount program.
In the experiment, we set the file size of each reading and writing data are GB level
and the file number of the concurrent (the map number during the process of the mapreduce)
is 2,4,6 and 8.