Combining predictive models with heuristic methods for VM placement to reduce SLA violations in the cloud environment

Today, with the rise of cloud data centers, power consumption has increased and cloud infrastructure management has become more complex. On the other hand, meeting the needs of cloud users is an important goal in cloud infrastructure. The process of determining the load status of physical machines and placing virtual machines on suitable physical machines can reduce energy consumption and prevent service level agreement violations. To address these issues, a virtual machine placement solution with a prediction capability is required to effectively place virtual machines in the proper hosts at runtime. The aim of this study is to provide a cloud management strategy that uses regression, moving average and simple exponential smoothing predictive models to identify overloaded physical machines and heuristic methods based on energy consumption, CPU utilization, number of virtual machines and memory to determine the appropriate physical machine for virtual machine placement, so provides a proper trade-off between reducing service level agreement violations and energy consumption and also decreases the number of virtual machine migrations. The cloudsim simulator version 3.0.3 has been used to evaluate the proposed model. The simulation results show that the proposed model averagely reduced the service level agreement violations by 45.65%, energy consumption by 28.96% and the number of virtual machine migrations by 46.49% compared to similar methods.