ABSTRACT
With the foundation and inception of resource management, workload prediction technology has always been the hot spot in the field of computing research. With the rapid development of the cloud computing system, continuous expansion of the system scale, more complex resource management, dynamic workload characteristic, low prediction accuracy, and bad prediction algorithm instantaneity in cloud workload prediction research, an online dynamic workload prediction algorithm is proposed in this paper. In this algorithm, host workload is decomposed into several independent task workloads, an improved Kalman filter is used to predict the task workload, and an error compensation mechanism-based Markov chain model is added for correcting the final prediction results. The proposed algorithm has advantages of low computation cost and high prediction accuracy. Several experiments conducted on the classic workload prediction algorithms show that the proposed algorithm is more efficient and accurate.
