Description And Analysis Of The Run Time Pattern

Better Essays
Overview: This section describes the interpretation and analysis of the results from both of the experiments. The analysis is based upon the comparison of the theories against the results of the experiment, whether the results support or reject the theory. This section consists of the following analysis: comparison of the run time pattern in three times execution, comparison of the overall run time using an arithmetic mean approach and error analysis.

Run Time Pattern in Three Executions
MySQL (non cluster) had a stable pattern within three executions, while Hadoop did not. According to my investigation, it caused by the following reason: I executed MySQL (non cluster) through SSH window, whereas Hadoop was executed through Ambari client browser. The connection through SSH did not use Internet public network, while client browser used it. In other words, the public Internet network only affected client browser.

Based on 10 times bandwidth measurement (appendix F), the Internet speed was unstable. The Internet speed was unstable because it was a public network. As a public Internet network used by other people as well, the speed of the network was depending on the high/low network traffic. The traffic of public network was unpredictable: some users might do heavy activities at one time, and few seconds later the users did not do anything.

Hence, the instability of public Internet speed caused the instability of the total time taken of Hadoop’s activities. MySQL (non
Get Access