A novel virtual machine consolidation algorithm with server power mode management for energy-efficient cloud data centers

With the increasing demand for resources from cloud computing users, enhancing the energy efficiency of data centers and ensuring the Quality of Service have become noteworthy research topics. Most Previous studies on Virtual Machine (VM) consolidation often neglect the delay and energy consumption associated with Power Mode Transitions (PMTs), leading to potential deviations in experimental results from real-world scenarios. In this paper, we propose a Two-Level Heuristics Virtual Machine Consolidation (TLHVMC) method that takes into account the overheads of PMTs. This method employs a Greedy Strategy-based hybrid heuristic Placement (GSP) algorithm for resource allocation. After incorporating underloaded host detection for host classification, GSP applies distinct greedy strategies on various host lists, which balances data center workload and minimizes active hosts. Moreover, Time Window-based Centralized power mode Control policies are implemented to centrally manage host power modes. These policies improve energy-saving by suspending idle hosts and reduce the Service Level Agreement violations by adaptively resuming hosts based on the data center’s state within a time window. Experimental results on real workload traces show that TLHVMC significantly improves the energy efficiency of the data center and maintains the required cloud system performance compared to some other VM consolidation algorithms.