Runtime state change detector of computer system resources under non stationary conditions

All runtime management decisions in computer andinformation systems require immediate detection of relevantchanges in the state of their resources. This is accomplishedby continuously monitoring the performance/utilization of keysystem resources and by using appropriate statistical tests todetect the occurance of significant state changes. Unfortunately,the complexity of today systems and applications and the unpredictability of user request patterns result in highly variableand non stationary time series which are difficult to analyze. Asa consequence, present solutions for detecting state changes atruntime are affected by excessive time delays or false positives.We propose a novel “agile” runtime detector that solves thedelay vs. false positive tradeoff: it is able to detect the relevantstate changes as fast as the best reactive models with the lowestpercentages of false positives. All evaluations carried out for alarge set of scenarios confirm the efficacy and robustness of theproposed model.