Home // Research Research Results Low Carbon Virtual Private Clouds
Low Carbon Virtual Private Clouds
Fereydoun Farrahi Moghaddam, Mohamed Cheriet, Senior Member, IEEE, Kim Khoa Nguyen Synchromedia Lab, Ecole de technologie supérieure, Montreal.
Published in 4th IEEE International Conference on Cloud Computing, pp. 259-266, Washington, DC, USA, 4-9 July, 2011.
Abstract—Data center energy efficiency and carbon footprint reduction have attracted a great deal of attention across the world for some years now, and recently more than ever. Live Virtual Machine (VM) migration is a prominent solution for achieving server consolidation in Local Area Network (LAN) environments. With the introduction of live Wide Area Network (WAN) VM migration, however, the challenge of energy efficiency extends from a single data center to a network of data centers. In this paper, intelligent live migration of VMs within a WAN is used as a reallocation tool to minimize the overall carbon footprint of the network. We provide a formulation to calculate carbon footprint and energy consumption for the whole network and its components, which will be helpful for customers of a provider of cleaner energy cloud services. Simulation results show that using the proposed Genetic Algorithm (GA)-based method for live VM migration can significantly reduce the carbon footprint of a cloud network compared to the consolidation of individual data center servers. In addition, the WAN data center consolidation results show that an optimum solution for carbon reduction is not necessarily optimal for energy consumption, and vice versa. Also, the simulation platform was tested under heavy and light VM loads, the results showing the levels of improvement in carbon reduction under different loads.
Goals and Motivations
Creating a Low Carbon Cloud (Follow the wind/sun concept)
Definition of greenness for a node
Calculation of carbon footprint
Creating a Simulation Platform for a Virtual Private Cloud (VPC)
Carbon footprint Optimization in VPC
Carbon footprint Opt. vs Energy Efficiency
Virtual Private Clouds
A VPC is a set of clouds which are connected to each other in a flat topology with direct high bandwidth links or VPN connections through internet. A VPC will allow VMs to migrate from one data center to another data center seamlessly without interruption in service. There are different works to help building a VPC [1,2,3,4].
CloudNet: A Platform for Optimized WAN Migration of Virtual Machines 
Definition of Greenness of a Node
Different source of energies have different carbon footprint:
Table I - Current State of Development of Electricity-Generating Technologies: A Literature Review 
Each source of energy has different greenness compare to other source of energies. Here, we define a conversion rate for power to carbon as follows:
Power-Carbon conversion rate
It can also be defined for nodes with more than one source of energy as bellow:
Power-Carbon conversion rate for more than one source of energy
We define the green percentage of a node as follows:
This formulation will give 0% for the most non-green source of energy (Coal) and 100% green for a source of energy with Power-Carbon conversion rate equal to zero. The rest of source of energies will fall in between.
VPC Carbon Footprint
In this research we provide a formulation for carbon footprint of a VPC:
Carbon footprint of a VPC
We also provide a simulation platform to measure and optimize carbon footprint:
Using provided formulation in the simulation platform we measure the carbon footprint of a VPC of 13 data centers with existence of sun and wind.
Carbon footprint of a VPC in 3 days
VPC Migration Management
In order to reduce the carbon footprint of a VPC, virtual machines need to move from data centers with lower green percentage to data centers with higher green percentage. As it is shown in the carbon footprint formulation, the carbon footprints of migrations are also in the equation. In this research a genetic algorithm module will read the whole network information as an optimization problem in each time interval and will provide the best solution as the next status of the network. Then, virtual machines will migrate to the new destinations accordingly. Carbon footprint formulation is used as the cost function for the genetic algorithm. Place of virtual machines are the variabels of the optimization problem.
Choosing the right time interval
To choose the best time interval for execution of genetic algorithm, different time intervals are studied and among all of them a time interval between half an hour and two hour is the best time interval for the re-optimization of the network. In all our simulation a time interval of one hour is choosen.
Small time intervals
Large time intervals
A VPC on simulation platform is tested under no optimization, server consolidation on each data center only, and VPC optimization on whole network. Simulation results show a significant decrease in carbon footprint on the VPC.
Carbon footprint of a VPC under different optimization modules
We also optimized the VPC for energy and compared the results with when optimized for carbon footprint.
Cabon footprint: Energy optimization vs Carbon optimization
Energy consumption: Energy optimization vs Carbon optimization
As it is shown above, when energy is the target of the optimization, we may save a little bit of energy but network have much more carbon footprint compare to when carbon footprint is targeted by the optimizer.
Finally we repeat our tests on the same network with different loads, and simulation results show that in any load of VMs the VPC optimization will reduce a lot of carbon footprint. The reduction is the highest when network in under lightest load.
Heavy load results
Light load results
VPCs enable optimized WAN migrations over normal internet connections.
In a VPC, when Carbon optimization is performed, Energy is not optimized and vise versa.
Carbon reduction is significant under different VPC loads when carbon optimization is performed.
Low Carbon VPCs are optimized on carbon but not on cost. Certain carbon regulations are necessary for them to be financially interesting.
Use real data for wind streams.
Add weather conditions such as clouds.
Consider different type of applications on VMs.
Consider migration failures in plan executor.
Scale the network.
Add more metrics to the carbon calculation.
Analyze the cost.
Apply to real testbed.
 Van der Merwe, J., Ramakrishnan, K. K., Fairchild, M., Flavel, A., Houle, J., Lagar-Cavilla, H. A., and Mulligan, J., Towards a ubiquitous cloud computing infrastructure, 17th IEEE Workshop on Local and Metropolitan Area Networks (LANMAN), vol., no., pp. 1-6, May 5-7, 2010.
 T. Wood, K. Ramakrishnan, J. van der Merwe, and P. Shenoy, CloudNet: A Platform for Optimized WAN Migration of Virtual Machines, University of Massachusetts Technical Report TR-2010-002, January 2010.
 Farrahi Moghaddam, Fereydoun, and Cheriet, Mohamed, Decreasing live virtual machine migration down-time using a memory page selection based on memory change PDF, International Conference on Sensing and Control (ICNSC), pp. 355-359, April 10-12, 2010.
 Timothy Wood, Alexandre Gerber, K. K. Ramakrishnan, Prashant Shenoy, and Jacobus Van der Merwe, The case for enterprise-ready virtual private clouds, In Proceedings of the 2009 conference on Hot Topics in Cloud Computing (HotCloud09). USENIX Association, Berkeley, CA, USA, 4-4, 2009.
 Lenzen, Manfred, Current State of Development of Electricity-Generating Technologies: A Literature Review Energies 2010, 3(3), 462-591.