Dynamic Virtual Machine Scheduling Using Residual Optimum Power-Efficiency In The Cloud Data Center

Abstract

Abstract Cloud data center serves tremendous workload demand due to the ever-increasing usage of internet services. Scheduling of these workloads over the physical servers is the combinatorial problem that resembles to an NP-complete problem. Furthermore, the workload is dynamic and changes at each scheduling interval result in high power consumption and Service Level Agreement (SLA) violation. Virtual Machine (VM) migrations provide the opportunity to balance this dynamic workload. However, it results in additional power consumption and performance loss. Therefore, this paper aims to find the optimal VM scheduling with minimum VM migration to minimize power consumption and ensure the SLA. This paper proposes a Residual Optimum Power Efficiency (ROPE) aware Improved Clonal Selection Algorithm (ICSA) for dynamic VM scheduling. ICSA-ROPE algorithm finds optimal VM schedules at each scheduling interval guided by two optimization functions Total Datacenter Residual Optimum Power-Efficiency (TDCROPE) and VM Migration Cost (VMC). TDCROPE ensures that servers operate at optimum power efficiency as they consume less power and are less prone to SLA violation, while VMC ensures to find optimal VM schedule with minimum VM migrations. The proposed approach is implemented on a CloudSim simulator, and results show that the ICSA-ROPE is 95.54 %,90.34% and 88.49% more significant in terms of performance efficiency than the LrMmt, DthMf and VMS-MCSA.