Overload Pattern classification For Server Overload Detection

Received Jan 12 th , 2013 Revised Feb 1 st , 2013 Accepted Feb 26 th , 2013 In automated overload control system, it is difficult to accurately predict the condition of high performance server. Overload Control mechanism mostly depend on the accuracy of detection and prediction. Therefore, it is important to accurately determine whether a server is or is not in overload state. So we can prevent from overload saturation before it occurs. An important step towards overload prediction is the ability to detect it. In this paper, we described overload pattern classification by using Random Forest classifierand overload detection mechanism was proposed for server overload control system. In proposed mechanism, we make a synthetic dataset by using performance counter pattern and do experimentation on various classifiers. Keyword: