Enhanced multi-objective virtual machine replacement in cloud data centers: combinations of fuzzy logic with reinforcement learning and biogeography-based optimization algorithms.

The process of remapping virtual machines (VMs) to host machines (HMs), which is defined as VM replacement, affects cloud data centers (DCs) performance. To enhance the performance, optimal replacement of VMs regarding conflicting objectives has been proposed in some research, most of which use the non-dominance method to compare generated solutions. Although the non-dominance approach reaches acceptable results, it is not desirable in the VM replacement problem since we need only one mapping of VMs to HMs as the optimal solution to this problem. Indeed, in this method, a solution is considered better if it outperforms another solution regarding all parameters otherwise, two solutions are considered equally good. Also, a solution can be better even though it demonstrates poor performance regarding some parameters. In this paper, we propose two enhanced multi-objective algorithms, Fuzzy-RLVMrB and Fuzzy-MOVMrB, to address the mentioned problems of the non-dominance algorithm. The proposed algorithms aim to balance the horizontal and vertical load between PMs in terms of processor, bandwidth, and memory. We simulated all algorithms using the CloudSim simulator and compared them in terms of horizontal and vertical load balance, energy consumption, and execution time. The simulation results depict that Fuzzy-RLVMrB and Fuzzy-MOVMrB algorithms outperform other algorithms in terms of vertical load balancing and horizontal load balancing. In addition, Fuzzy-RLVMrB surpasses other approaches in terms of energy consumption. [ABSTRACT FROM AUTHOR]