Real-time models supporting resource management decisions in highly variable systems

Data centers providing modern interactive applications are enriched by autonomous management decision systems that are able to clone and migrate virtualmachines, to re-distribute resources or to re-map services in real-time. At the basis of all these decisions,there is the need of a continuous evaluation of the stateof system resources and of detecting when some relevant changes are occurring. Unfortunately, the load ofinteractive applications reaching the system is intrinsically heterogeneous with consequent highly variable effects on the resource behavior emerging from systemmonitors. Hence, existing algorithms for online detection of state changes are affected by low precision andscarce robustness when they are applied to modern contexts. We propose a novel model for online detectionof relevant state changes that combines a filtered representation of the raw measures with adaptive detectionrules. Experiments carried out on real and emulateddata sets confirm that the proposed model is able totimely signal all relevant state changes, to limit falsedetections and, even more important, its results are robust in highly variable contexts.