A novel performance constrained power management framework for cloud computing using an adaptive node scaling approach

Cloud computing is an on-demand IT resource delivery technology that is aided by server virtualization and load balancing. Power and performance management to improve operational efficiency and increase compaction are important considerations from a cloud service economic point of view. The objective of the present study was to draw new insights from existing approaches and techniques to design an innovative self-adapting mechanism to address the mismatch between server's energy-efficiency characteristics and the behavior of server-class workloads, which solves the power versus performance trade-off problem at cloud data centers. The proposed system was simulated and evaluated for highly variable cloud workloads. The results suggest that the proposed system functions reliably for cloud workloads and ensures an optimal server workload distribution (i.e., determines the allocations of the VM server), minimizing the average power consumption of the servers and ensuring that the average task response time does not exceed given performance limitations. (C) 2017 Elsevier Ltd. All rights reserved.