Carbon Dioxide Emission Minimized Virtual Machine (VM) Placement in Cloud-Fog Network Architecture

Abstract

Cloud computing has provided economies of scale, savings, and efficiency for both individual consumers and enterprises. Its key advantage is its ability to handle increasing amounts of data and provide functionality that gives users the ability to scale their computing resources, including processing, data storage, and networking capabilities. Virtual Machines (VM), enabled via virtualization technology, allow cloud service providers to deliver their services to users. This, however, results in increasing carbon dioxide emissions from increased energy use. This paper introduces a Mixed-Integer Linear Programming (MILP) model that investigates the VM placement, focusing on the British Telecom (BT) network topology, in a cloud-fog network architecture when renewable energy sources are introduced in the fog layer located near traffic-producing sources. VMs can be placed on nodes hosted on the core, metro, and access (fog) layers. We first investigate the effect of varying traffic on IP over WDM power consumption in the backbone network and the number of optical carrier signals to serve the traffic over a period of time. We later extend the model to consider the CO2-minimized optimal virtual machine placement given the sporadic traffic quantity, and the consideration of solar renewable energy sources placed in data centers located in the access (fog) layer throughout the day, imposed on the VM and the minimum workload requirement of the VM to maintain a service level agreement (SLA).