By Desiree L. DePriest, Purdue University Global 

Fig. 1.  "Network Earth" (by Geralt on Pixabay) 

The United States debates the effects of global warming in a perplexing way. It does not discount the weather changes or climate, nor does science or other belief systems not see the increases in the severity and frequency of these changes. The argument-to-action rides on fault, whether it is a natural phenomenon of Earth cycles or has causation based on human behavior. Human behavior contributing to carbon dioxide emissions, fossil fuels in agriculture (including deforestation), ocean warming and subsequent acidification are the top activities focused upon. While this fault shaming in America continues in perpetual inertia, it seems little to no blame gets attributed to the massive amounts of energy that is required for information and communications technology (ICT) companies to operate and cool data centers. More frequently, these have become hyperscale data centers as the vast majority of human activity is digitized and then stored. ICTs then sell the data to numerous industries to analyze and make transactional decisions on new business opportunities.

There is little to no discussion of the huge amounts of energy that ICTs use through stored data and networked systems. Data center storage capacity is set to reach 190 exabytes by 2021. Data centers are no longer just big buildings with multiple servers on one huge level but are multiple buildings and acreage with tiers of servers to ensure redundancy. Most recently, optimum data center activity was 99.67% availability with only 28.8 hours of interruption per year (OVH_SAS [FR]). At the end of 2017, hyperscale data centers were denoted as having at least 500 servers and 10,000 square feet of available space. In all, the U.S. had 44% of these hyperscale data centers in 2017. At the end of 2018, the U.S. only had 40% as a significant amount of construction occurred in the Asia-Pacific and other regions (Hardy, 2019). ICTs are described more recently as having several hundreds of thousands of servers – or sometimes millions (Miller, 2018). 

Most small to medium-sized companies never experience the full use of a dedicated server because they share these services through cloud computing. Technical management is avoided and it is shared hosting. Companies like Google Analytics provide data feed information to smaller companies from online transactions, visits by consumers to web sites, and customer service interactions. Smaller companies using cloud computing can scale up or down as needed and better manage costs however, the ability to spin up different systems on demand increases usage and leads to massive increases in demand on the energy grid. Although cloud hides individual companies’ use of electricity due to removing machines, hardware, distributed databases and capital risk, it currently relies on electrical grids using fossil fuels. Cloud servers still make API calls, dump into a dedicated server somewhere and still contribute to the use of energy through the allocated resources and maximum performance. All these transactions are housed in hyperscale data centers.

Currently, data centers produce at least 2% of global emissions. It is suggested the tech industry produces 1/50th of all carbon emissions now and is expected to produce up to five times more in the next seven years (Johnston, 2018). Other researchers estimate 3.5% of global emissions by 2020 and up to 14% by 2040 (The Guardian, 2017). 

These numbers are best understood when put into context. Data is the primary product from all industries, and everything is being digitized. Internet protocol traffic, all generations of mobile technology, wristband pedometers, and product UPCs are stored. Satellite data is digitized as well as emerging quantum computing data. The world’s banks, credit, and trading data are in these centers. Social security numbers from the moment of birth in the U.S., selective service notifications for males at 18 years old, voter registration and earbud usage are all in these centers. Television channel selections, time and length of usage (for all electronics) are housed in data centers. Data capture from cars, machines, robots and artificial intelligence are digitized and stored in data centers. 

Other ICT companies are mining data from social networks in real time. Pairing customer service representatives (chat, messaging, and voice/contact centers) with a customer based on their pre-determined personality type, race or ethnicity (profile based routing) is based on data center information. Facial recognition, fingerprints, electronic medical records (EMR), the Internet-of-Things (IoT), and technical expansion to developing countries are also a part of what is stored in data centers. Data from GPS chips in animals are stored in these centers.

One additional but necessary mention to the amount of energy consumed by data centers is the huge amount of cryptocurrency mining or cryptomining. It is a strenuous process with specialized software requiring constant generation of currency and securing of transactions. Each bitcoin takes huge computational energy with nearly impenetrable algorithms (DW.com, 2019) and it is suspected to consume more electricity than Ireland.

ICT companies are under attack publicly for myriad reasons from common carrier requirements, classifications of ISPs and most importantly what is done with the acquired user data. Arguably, the use case is simple – they sell the data anywhere and everywhere. More scrutiny placed on what these companies are doing to reduce the amount of carbon dioxide and other carbon compounds emitted and the gouging of electricity through fossil fuels, is the-elephant-in-the-room reshaping economy and the Earth herself.

Fig. 2.  "Electrical Data Center" (by Akela999 on Pixabay) 

Electricity is traded just like any other commodity and referred to as an electrical “grid” for this reason. For example, Arizona supplies electricity throughout the Southwest, not just in Arizona. Alabama and Arkansas generate more electricity than they consume, and typically send about one-third of their output to nearby states (Popovich, N., 2018). Data Centers, requiring so much electricity to avoid a loss of power, cooling or connectivity normally have many forms of electrical supply and backup mechanisms from grids that are immediately available where the data center is located to elsewhere. The backup mechanism used, due to a transmission line failing could be a diesel generator or large battery storage. This adds to the carbon emissions footprint.

This is not just a problem for the world’s climate, it is also extremely expensive for the internet companies. Business must continue to create value for their stakeholders while reducing the excesses as data centers grow. Larger data centers can demand almost a fifth of the output of a conventional coal power plant to stay cooled (Hanak, 2015). In Belgium, the optimum temperature for a Google data center is 68 degrees Fahrenheit with a peak operating temperature of 95 degrees Fahrenheit for a few hours. At that peak, humans can no longer manage the center (Humpries, 2012). Dell warranties its servers to operate in environments as warm as 115 degrees Fahrenheit but again, humans must be replaced with artificial intelligence tools (SteveFL, 2016). However, the risk is that CPUs can get fried by their own heat which triggered the ICT companies to see the value added in efficient cooling combined with AI technology. ICTs recognizing the need to reduce the costs of energy consumption in data centers, make money, and future resource planning, just happened to align with greener technologies.

Free cooling through solar, wind, biomass or hydro uses the environment to cool the systems. Cooler climates such as Denmark and Ireland, where outside air is cooler most of the year than the temperature inside, allow natural free air to cool data centers by simply opening windows (Hanak, 2015). It is cheaper and also eco-responsible.

The same can be done by these large and extremely profitable tech companies building solar, wind, biomass and hydro farms to generate the same amount of energy, and then buy and sell from each other. This is called carbon offset but it is not currently fully sustainable or consistent. There would still be some reliance on fossil fuels but it is a step in the right direction that internet companies are not getting credit for. Regardless of motives, these are the economics and ethics of the free market.

Organizations like Greenpeace have not ignored what data centers contribute to energy consumption in the world. Greenpeace began benchmarking the energy performance of the IT sector in 2009, challenging those companies who are the largest global architects and operators of the internet to commit to powering their rapid growth with 100% renewable energy (Greenpeace, 2017). As a result, many large tech companies have committed to buying off grid in addition to building solar, wind, hydro and biomass farms to generate energy. 

In a report from the EIRGRID Group, wind energy with contributions from hydro and biomass, provided 22% of all electricity in 2016 in Ireland (All-Island Generation Capacity Statement, 2017-2026). The share of electricity generated from renewable sources increased by 3.3 percentage points in 2017, to 30.1% towards the 40% 2020 target (SEAi, 2018). Northern Ireland is using solar photovoltaic energy generation. As this extends “all-Island wide” a deficit in one jurisdiction can be alleviated by a surplus plan in the other. 

Denmark is another country dedicated to abundant green power with a mild climate that allows low-energy year-round cooling. More than 50% of the energy in Denmark’s electricity grid comes from renewable sources, primarily wind power. Their renewable sources are an interconnected infrastructure with an up-time of 99%. There is no surprise that Apple, Facebook and Google have built, and are building, huge data centers in Viborg, Aabenraa, Odense and Fredericia (A Power Hub for International Data Centres, n.d.). The Danish government strives to be completely cleantech and fossil-free by 2050.

In America, Amazon Web Services (AWS) had achieved 50% renewable energy usage as of 2018. AWS has six solar farms in Virginia, and three wind farms in Indiana, North Carolina and Ohio. Amazon’s largest wind farm in Texas adds over 1 million megawatt hours of clean energy to the grid annually (Frangoul, 2018). 

IBM also purports “nearly” 50% of procured electricity from renewable generation sources. Colocation data centers use only 27.4% renewable energy, however, IBM manages most of its data center locations (Johnston and Currie, 2018).

Google describes its efforts as a “matching” commitment. For every kilowatt hour of energy it consumes, it adds a matching kilowatt-hour of renewable energy from solar or wind farms to a power grid somewhere. Google admits it is too big to run on 100% renewable energy at this time (Frangoul, 2018).

In 2017-2018, Newsweek “Green Rankings” named Cisco Systems, Inc., ranked #1 (U.S.) and #5 (globally) or 83.70% green. Apple, Inc. was ranked #8 (U.S.) and #17 or 71.50% green. Many ICT companies are making decisions on where to locate data center sites is based on the availability of renewable energies including colocations.

In a Microsoft report dated May 17, 2018, it showed 93% more efficiency including renewable energy. Microsoft determined that investing in sustainability is good for business, good for customers and good for the planet (Microsoft, 2018). 

Fig 3.  "Wind Farm" (by Coffee on Pixabay)

 

No one industry or government can address the problem of climate change in isolation. Each of us, in our homes and businesses, have opportunities to be more sustainable in our use of energy (SEAi, 2018). If one were to partition their mind limited to the ICT sector, there would still be net neutrality, data capture, cybersecurity, social media and fake news trolling and even ICT-shared/organized terrorism unrelated to climate change. The ICT companies continue to endure the fault debates on the front-facing issues that are usually related to how they use and protect data but not the energy consumption related to data. 

At the same time, great strides are occurring in eco-friendly, energy conservative, and climate efficient data centers to preserve and receive more of that same data. These companies have found the connection between profitability and going green. They will continue to find better and more economical ways to retain energy, profit from data, and propose solutions ahead of government-imposed regulations. There is a foreseeable future, totally run by AI, where hyperscale data centers will survive climate change even if sentient beings don’t. Data centers don’t need breathable air, just air to make inexpensive wind energy. Data centers don’t need drinkable water, just water to make inexpensive hydro energy. 

 

All-Island Generation Capacity Statement (2017-2026). Retrieved from http://www.eirgridgroup.com/site-files/library/EirGrid/4289_EirGrid_GenCapStatement_v9_web.pdf 

A Power Hub for International Data Centres (n.d.). Ministry of Foreign Affairs of Denmark. https://investindk.com/set-up-a-business/cleantech/data-centres 

Bitcoin energy boom stamps down colossal carbon footprint. Deutche Welle [DW], 2019) Retrieved from https://www.dw.com/en/bitcoin-energy-boom-stamps-down-colossal-carbon-footprint/a-41695365 

Frangoul, A. (2018, April 5). Google says it’s the biggest corporate buyer of renewable energy on the planet. Retrieved from https://www.cnbc.com/2018/04/05/google-says-its-the-biggest-corporate-buyer-of-renewable-energy-on-the-planet.html 

Greenpeace (2017). Clicking Clean: Who is winning the race to build a green internet? Retrieved from http://www.clickclean.org/usa/en/ 

Hanak (2015, Nov. 18). How data center free cooling works and why it is brilliant. Retrieved from https://www.masterdc.com/blog/what-is-data-center-free-cooling-how-does-it-work/ 

Hardy, S. (2019, Jan. 15). 430 hyperscale data centers worldwide at end of 2018: Synergy Research. Retrieved from https://www.datacenterdynamics.com/news/synergy-number-hyperscale-data-centers-reached-430-2018/ 

Humpries, M. (2012, March 27, ) Google’s most efficient data center runs at 95 degrees. Retrieved from https://www.geek.com/chips/googles-most-efficient-data-center-runs-at-95-degrees-1478473/  

Johnston, P. (2018, Oct. 2). What do climate change, Data Centres and Cloud have to do with each other? Retrieved from https://medium.com/@PaulDJohnston/what-do-climate-change-data-centres-and-cloud-have-to-do-with-each-other-9b908887c610 

Johnston, P. & Currie, A. (2018, Sept.). The state of data center energy use in 2018). Retrieved from https://docs.google.com/document/d/1eCCb3rgqtQxcRwLdTr0P_hCK_drIZrm1Dpb4dlPeG6M/edit 

Microsoft.com (May 17, 2018). The Microsoft Cloud can save customers 93 percent and more in energy and carbon efficiency. Retrieved from https://news.microsoft.com/2018/05/17/the-microsoft-cloud-can-save-customers-93-percent-and-more-in-energy-and-carbon-efficiency/ 

Miller, R. (2018). Hyperscale data centers reached over 390 worldwide in 2017. Retrieved from https://techcrunch.com/2017/12/30/hyperscale-data-centers-reached-over-390-worldwide-in-2017/ 

Newsweek Green Rankings. (2017-2018). Retrieved from https://www.newsweek.com/green-rankings-2017-18 

Popovich, N. (2018, Dec. 24). How does your state make electricity? New York Times. Retrieved from https://www.nytimes.com/interactive/2018/12/24/climate/how-electricity-generation-changed-in-your-state.html 

SEAi (2018). Sustainable Energy Authority of Ireland. Energy in Ireland: 2018 Report. Retrieved from https://www.seai.ie/resources/publications/Energy-in-Ireland-2018.pdf 

SteveFl (2016, July 7). Serve Rooms Temperatures…How hot is too hot? Retrieved from https://community.spiceworks.com/topic/1704357-server-rooms-temperatures-how-hot-is-too-hot 

Desiree DePriest is an IT/AI business intelligence professor at Purdue University Global for 14 years. Desiree’s expertise is in technology program creation, information systems and artificial intelligence in business environments.  She holds a Ph.D., in Management & Organization with emphasis in Information Technology, along with two masters degrees (Telecom and IS respectively). Desiree has a Bachelor of Science degree in psychology and certificate in ABA and I-O psychology which greatly assist in her work in the various areas of business intelligence, industrial and organizational motivation and attitudes. She is the Vice-chair of the Institutional Review Board at Purdue Global and attended UMKC Law School. 

Desiree developed and directs the Purdue Global Internship Program – Technology (PGIP-T) which is an internship for IT and business students wanting real world experience prior to graduation. She also created the Graduate Information Technology Association (GITA) for active and alumni IT/Business students, and serves as Faculty Advisor. Desiree recently won the “Best Practices” award for her work in the internship from the American Association of Adult Continuing Education (AAACE). Her publications include research in persuasive and predictive analytics, artificial intelligence and algorithms in decision support, and pattern recognition. Desiree enjoys staying abreast of technology in current events. Other interests include neural correlates of consciousness (NCC), cognitive coupling (CC) and quantum teaming (QT). Quantum Teaming is a quality management methodologies with particular focus on virtual team environments and is the intellectual property of Dr. DePriest. Desiree presents throughout the year at conferences in these areas.