Cloud Computing Has Green Appeal
Large-scale deployment of cloud computing, backed by green datacenter design principles, may help contain runaway energy consumption.
Cloud computing refers to the use of computer hardware and software resources which are usually remotely located and are accessible over a network. The advantage of cloud computing is that the users will be able to access applications and data on the cloud from anywhere in the globe, making the cloud appear as a single point of access. Various types of cloud services are on offer today. Amazon Elastic Compute Cloud (EC2) provides resizable, remote computing capabilities in the cloud; Microsoft Live Mesh allows users to store documents, photos and other files on the cloud and access it all across the globe; Gmail is a completely cloud-based e-mail service; Flickr, Tumblr, Dropbox are other cloud-based services as well.
Just as almost anything we do nowadays possesses a carbon footprint; similarly, the Internet also has a carbon footprint associated with it. According to a report by Gartner, the Information Communication Technology (ICT) sector is responsible for 2 percent of global CO2 emissions per year. The same report concludes that the Internet releases a whopping 300 million tons of CO2 in a year (which is more than 50 percent of the carbon emission of the whole of UK). Network equipment in the US consumed an estimated 14 – 18 TWh in 2008, whereas the total European telecom network used 14.3 TWh in 2005. It is estimated that by 2030, the emissions by the ICT sector will increase by 60 percent. Drastic measures need to be taken to reduce this impact, and in this article we will explore one of the most efficient ways to achieve a significant reduction, namely, cloud computing.
We shall now elicit a number of ways in which cloud computing can be used to provide energy savings. The basic method is the pay-per-use or pay-as-you-go nature of the cloud. This means that users are forced to pay for the resources that they will consume, thereby ensuring that they do not waste or hog any critical or valuable resources for an extended period of time. Hence, energy and resource efficiency will automatically increase and over-utilization gets reduced. Cloud computing also assists in reducing data duplication. Large organizations often need to have some information available at all times to all their offices around the world. One way to achieve this is to set up a costly centralized server. But for those organizations who have distributed servers often have multiple copies of such important data. Thus, an efficient way to have information available at all times is to store it in the cloud.
Cloud-based file hosting and sharing services can be utilized to eliminate the need of portable storage devices, since the information stored in the cloud can be accessed from all places. Shifting to cloud-based e-mail services will also save on the energy and materials needed to create and print pages. With time, information stored and shared in the digital format via clouds, can effectively phase out their physical counterparts.
All these claims to reduce carbon footprint via cloud computing have also been verified by independent studies from various sources. Microsoft, Accenture and WSP conducted a study where they showed that the smaller is the organization shifting to the cloud, the greater is the energy and carbon savings. A study by the Carbon Disclosure Project, supported by AT&T, has decisively proved that companies adopting cloud computing have reduced carbon emissions, lower energy consumption and decreased expenditure while their operational efficiency improved.
Let us now briefly explore the science behind why cloud computing is really more efficient than other centralized or distributed systems. At its very core, cloud computing resembles a giant centralized system and yet it manages to be much more energy efficient. Let us consider an organization which maintains its own data and e-mail servers in-house. More often than not, these servers will be utilized to only half of their total capacity, and the remaining storage space gets wasted. But the organization cannot provide power to and run only the utilized part of the server. Instead, it has to expend a huge amount of energy to maintain these servers and as a result much of this energy gets wasted due to under-utilization. Now if this organization shifts to a cloud-based server, an appreciable decrease in energy and carbon emissions is seen. This is because the specialized cloud servers are already well in use, and as more organizations join up, the server utilization levels increase, thereby decreasing the energy wastage. Also, the organization no longer needs to provide for their in-house servers, thereby bringing their own energy and carbon wastage levels down to next to zero.
A diagram illustrating the differences in server usage levels for in-house and cloud-based email.
A number of metrics exist to measure power utilization of servers. In a nutshell they are: Power usage Effectiveness (PUE), Datacenter Infrastructure Efficiency (DCiE) and Green Power Usage Effectiveness (GPUE). PUE gives an overall idea as to how effectively a datacenter utilizes its power; DCiE is just the inverse of PUE. GPUE gives a much better idea of the power utilization because it also takes into account various other factors which determine how environmentally-friendly the datacenter is. Consider the PUE and GPUE values of some datacenters given below. (Lower numbers are more desirable.)
Certain datacenters like those of Google, Microsoft, Yahoo and Apple have similar PUE values. But the GPUE values of Microsoft and Apple are significantly higher than those of Google or Yahoo, thereby showing that these datacenters are actually much less environment-friendly. The chart illustrates the increased effectiveness of the GPUE over the PUE. Take note of the GreenQloud datacenter having almost equal PUE and GPUE values.
Finally, we shall consider some success stories in this field. Google, along with Yahoo and Facebook, leads the movement in using renewable energy in its cloud services and also actively supporting policies that leads to greater conversion to cloud-based services from normal in-house services. Almost all of Google's flagship products - Gmail, Google Earth, Google Documents, etc. – are cloud based.
Google uses 39.4 percent renewable energy and most of it is also generated on-campus by Google itself. According to Google, a normal user releases around 1.46 kg of CO2 by using all the services offered by the company, which is equivalent to mundane tasks like filling a deep bath or buying a bottle of imported wine. Thus, Google is doing its part to try and reduce the global carbon footprint of the Internet not only by promoting cloud services but also through the usage of renewable resources.
Another noteworthy organization is GreenQloud, based in Reykjavik, Iceland. GreenQloud provides cloud-based server hosting, online storage and high performance computing services with minimal carbon footprint. As shown in the chart above, its PUE is close to the ideal value of 1.0 and its GPUE is close to its PUE value. This alone quantitatively shows the minimal impact GreenQloud has on the environment. This is mainly because of its location in Iceland, where cold ocean currents help in natural cooling of datacenters and the easy availability of 100 percent renewable geothermal and hydroelectric energy.
On a concluding note, it can be stated that cloud computing is still in its infancy, green cloud computing is even more so. With time, the cloud is becoming essential in storing and transmitting data for various purposes. By transferring more and more organizations and the common user to cloud-based services that are fueled by renewable energy sources, we can ensure that the carbon footprint of the IT sector can be slowly but surely decreased to much more manageable levels, thereby helping avert an impending global crisis.
About the Author
Chiranjeeb Roy Chowdhury was born in Kolkata, India in 1992. He has completed his Bachelor’s degree in Computer Science, graduating from St. Xavier’s College, Kolkata. He is a hardcore coder and has an in-depth knowledge of a plethora of programming languages, like C, C++, Java and C#. He is interested in cryptography, graph theory, Internet, .NET Technologies and cloud computing. He has recently completed a project in implementing his own interpretation of a Bayesian Spam Filter. He is also a bookworm and an avid gamer.