Energy saving and maximize utilization cloud resources allocation via online multi-dimensional vector bin packing

Abstract

In data center, which use virtualization technology and provide virtual machines (VMs) for their customers, they often receive the VMs renting request with determined multiple resource demanded which tends to arrive at an arbitrary time. Resource allocation strategies become one of considerable problem as they can affect the energy efficiency and resource utilization significantly. In this paper, we explore the resource allocation strategies which can minimize the number of used servers, in order to save the energy and maximize the resources utilization. We model this problem as the Vector Bin Packing Problem (VBPP) which is a variant of classic bin packing, proposed Multi-Dimensional Cloud Resource Dynamic Allocation Model(MDCRA) based on VBPP. Since most of existed algorithms aim to solve offline VBPP. We generalized those algorithms to online and proposed Single Weight and Double Weight family online algorithms based on the offline work. Through both rigorous theoretical analyses and extensive simulations, we demonstrate that the proposed allocation strategies achieve energy saving and utilization maximization.