A learning automata-based ensemble resource usage prediction algorithm for cloud computing environment

Infrastructure as a service (IaaS) providers are interested in increasing their profit by gathering more and more customers besides providing more efficiency in cloud resource usage. There are several approaches to reach the resource usage efficiency goal such as dynamic consolidation of virtual machines (VMs). Resource management techniques such as VM consolidation must be aware of the current and future resource usage of the cloud resources. Hence, applying prediction models for current cloud resource management is a must. While cloud resource usage varies widely time to time and server to server, determining the best time-series model for predicting cloud resource usage depend not only on time but the cloud resource usage trend. Thus, applying ensemble prediction algorithms that combine several prediction models can be suitable to reach the mentioned goal. In this paper, an ensemble cloud resource usage prediction algorithm based on Learning Automata (LA) theory is proposed that combines state of the art prediction models, and it determines weights for individual constituent models. The proposed algorithm predicts by combining the prediction values of all constituent models based on their performance. The extensive experiments on CPU load prediction of several VMs gathered from the dataset of the CoMon project indicated that the proposed approach outperforms other ensemble prediction algorithms.