Cloud Instance Resources Prediction Based on Hidden Markov Model

Accurate prediction of the type of cloud resource instances is increasingly important for public cloud users and cloud service providers as it relates to reasonable reservation of resources to minimize costs for both parties. However, existing methods cannot predict the instance type of cloud resources based on the workload requested by the user’s application, and rarely consider the characteristics of workload data changes for the real-time prediction. To address these challenges, this paper proposes a cloud resource instance prediction model to predict appropriate cloud instance resource usage in a timely manner. First, we use trend degree (TD) to classify all requested workload into three types of waveform trend patterns. Next, a method based on Hidden Markov model (HMM) is established. Finally, a reasonable cloud instance type can be predicted according to the requested workload. The simulation results show that this method can predict cloud resource instance types, and has higher prediction accuracy than the existing related methods.