CPU Load Prediction Based on a Multidimensional Spatial Voting Model

Resource performance prediction has become more and more important in cloud environment as CPU load prediction is key for system maintenance and application schedule. This paper presents a multidimensional spatial voting prediction model to predict real-time CPU load accurately. We improved the real-time CPU load prediction accuracy by gray prediction model under the one-dimension prediction, we also applied voting mechanism to find a more appropriate classifier prediction model for predicting the CPU load in real time. Our experiments showed that multidimensional spatial voting prediction model led to better predictions than classic models. Our model is not problem-specific, and can be applied to problems in the fields of other predictions.