Published Paper Details:

Virtual Machines (VMs) in Cloud systems are scheduled to hosts based on their instant resource consumption (e.g., hosts with the greatest accessible RAM), rather than their overall and long-term utilisation. Furthermore, the scheduling and placement operations are often computationally intensive and have an impact on the performance of deployed VMs. In this paper, we provide a Cloud VM scheduling method that considers existing VM resource consumption over time by assessing previous VM utilisation levels in order to schedule VMs while maximising performance using the PSO technique. Because Cloud management activities such as VM placement have an impact on previously deployed systems, the goal is to minimise such performance deterioration. Furthermore, because overcrowded VMs tend to grab resources from neighbouring VMs, the task enhances the VMs' true CPU consumption. The results reveal that our method refines traditional Instant-based physical machine selection as it learns and adapts to system behaviour over time. The idea of VM scheduling based on resource monitoring data taken from previous resource utilisations (VMs). The PSO classifier reduces the physical machine count by four.