Energy consumption model for virtual machines in cloud data centre

Abstract

Cloud computing is a growing demand that allows a business to drive their goals using good Information Technology (IT). The tremendous increased in computing resources usage introduce several challenges to IT service providers. One of the key issues is the high energy consumption of the data centres. Virtualization has been used as a technique to reduce hardware footprint and thus realizing power efficiency in the data centre. However, improper placement technique has caused many virtual machines being provisioned inefficiently. Consequently, the physical servers installed as hosts are utilizing high power in data centre while the provisioned virtual machines are underutilized. This problem defeats the very reason why cloud computing was first introduced. This study focuses on identifying a relationship between virtual Central Processing Unit (CPU) of cloud servers, physical CPU, and total power consumption of its hypervisor. We developed a model to estimate power requirements for virtual machines (VM) in a cloud data centre and introduce an algorithm to place the virtual machines within the virtual data centre automatically. The model is applied to steady-state cloud data centres or in systems engineering design phase. The model can also be extended in determining a cloud server’s chargeback model based on an individual could server proportional power consumption.