Let's begin by creating a performance baseline for our system. When creating a baseline, you should keep the Hadoop default configuration settings and use the TeraSort benchmark tool, which is a part of the example JAR files provided with the Hadoop distribution package. TeraSort is accepted as an industry standard benchmark to compare the performance of Hadoop. This benchmark tool tries to use the entire Hadoop cluster to sort 1 TB of data as quickly as possible and is divided into three main modules:
TeraGen: This module is used to generate a file of the desired size as an input that usually ranges between 500 GB up to 3 TB. Once the input data is generated by TeraGen, it can be used in all the runs with the same file size.
TeraSort: This module will sort the input file across the Hadoop cluster. TeraSort stresses the Hadoop cluster on both layers, MapReduce and HDFS, in terms of computation, network bandwidth, and I/O response. Once sorted, a single reduced...