An Efficient Approach of Task Scheduling in Fog Computing

Abstract

In Cloud systems, Virtual Machines (VMs) are scheduled to hosts according to their instant resource usage (e.g. to hosts with most accessible RAM) without considering their overall and long-term utilization. Also, in many circumstances, the scheduling and placement processes are computationally intensive and impair performance of deployed VMs. In this work, a Cloud VM scheduling algorithm that takes into consideration already existing VM resource usage over time by assessing historical VM utilization levels in order to plan VMs by optimizing performance by utilizing KNN with NB approach. The Cloud management activities, such VM deployment, affect existing deployed systems hence the aim is to avoid such performance degradation. Moreover, overloaded VMs prefer to take resources from neighboring VMs, thus the work maximizes VMs real CPU consumption. The results reveal that our method refines traditional Instant-based physical machine selection as it learns the system behavior as well as it adjusts over time. The notion of VM scheduling according to resource monitoring data taken from prior resource utilizations (VMs). The count of the physical machine gets lowered by four utilizing KNN with NB classifier.