Determining SDN Stability by the Analysis of Variance Technique

Over the past years, many algorithms have been proposed for task scheduling and congestion control in typical Networks, such as Round-Robbin, Greedy, and many others. Whilst these algorithms are very effective in their capacities to address the targeted problems yet to compare their performance concerning the level of stability they offered in respective network systems has been a gap in the research environment that needs to be addressed. Proper scheduling mechanism along with reliable algorithms does not often guarantee a stable network else, this article is proposing a very simple but popular technique known in the statistical world as Analysis of Variance (ANOVA) as a tool for relatively determining the level of stability based on the data that are generated from the network. An OMNET++ simulator was used to conduct the simulation of the network environment and the SDN toolkits called INET framework was installed on top of it to enable the deployment of both Greedy and Round-Robbin scheduling algorithms to run. The simple analysis from the ANOVA was able to determine the level of stability between the two samples of algorithms used in these experiments. The performance evaluation while determining the response time from each experimental setup discovered that assuredly ANOVA analysis is capable of determining network stability level as well as proof that the Greedy scheduling algorithm performs better in terms of stability level than Round Robin (RR).