An improved K-means clustering and optimal random forest for scheduling workloads in virtual servers

The utilization of cloud computing systems has increased recently in many organizations and applications due to on-demand service as well as the quick response capabilities of cloud-based services. However, the dynamic load conditions on cloud infrastructure cause system failures. Therefore, it is essential to balance the load on the cloud server to efficiently process the tasks. Proper balancing of workloads on the cloud computing system optimizes response time, resource utilization, and system stability. To accomplish this goal, this paper proposes a novel energy-efficient load balancing technique named ‘improved k-means levy flight white shark optimization based random forest (IK-LWS-RF) algorithm’. This proposed algorithm performs two distinct operations namely clustering and scheduling to balance the workloads among the virtual servers. At first, the set of tasks needed to be executed is clustered according to their task lengths into distinct groups using an Improved K-means (IK) algorithm. The deviations that occurred on executing clustering workloads are reduced by utilizing the levy flight white shark (LWS) algorithm. The clustered tasks are then scheduled or assigned to virtual servers based on their processing capacities utilizing a random forest (RF) algorithm. Thus, the proposed IK-LWS-RF algorithm efficiently balances the workloads and as a result, resource utilization maximizes and task response time minimizes. The analytic results inherit the IK-LWS-RF algorithm’s supremacy over other compared techniques. The IK-LWS-RF algorithm takes very less response time of about 11.23 ms and, achieves greater memory utilization of about 94.7% and CPU utilization of about 98.2%.