EEDVMI: Energy-Efficient Dynamic Virtual Machines Integration

The rapid development of cloud computing technology has resulted in a great energy consumption, but the utilization rate of resources in the data centers is often relative low. Therefore, if the virtual machines in operation are integrated into several servers and the idle servers are switched to low-power modes, the power consumption of data centers can be greatly reduced. The traditional research on the integration of virtual machines is mainly based on the current load of the host to set a high load threshold or periodically perform the migration. However, the accuracy of these approaches on time series prediction is very limited. To solve this issue, this paper synthetically considers the influence of a multi-order Markov model and the CPU state at different times and proposes a novel K-order mixed Markov model for predicting the CPU load of the host for a period of time. By conducting large-scale data experiments on the CloudSim simulation platform, the host load forecasting method proposed in this paper is compared with some conventional approaches, and it verifies that the proposed model greatly reduces the number of virtual machine migrations and the data center energy consumption. Additionally, the violation of the SLA is at an acceptable level.