Genetic Algorithm for Virtual Machine Allocation using Server Power Profile

Abstract

A Virtual Machine allocation problem is one of the interesting topics in cloud computing. The cloud service providers allocate computing infrastructure, such as a virtual machine which is configured with CPU, memory, bandwidth and storage capabilities. More than one virtual machines can run on a host and different types of hosts have varying efficiency in energy consumption. To reduce energy consumption, this paper proposes a Genetic Algorithm which is a search heuristic to find the order of hosts to be utilized so that energy consumption in cloud data center is minimized. Our technique is based on a formulated problem of resource allocation which consider energy consumption from the CPU of VMs and hosts. The simulation focuses on power profile for each type of hosts including genetic algorithm. The experimental results show that the proposed algorithm of two-point crossover performed well when hosts are similar, while the one-point crossover performed well when hosts are a mix of high and low idle power.