Edge Computing and Data Center Resource Usage

In 1969, the term sustainable development first appeared on the world stage in a document signed by 33 African nations. In the same year, the EPA was formed in the United States. Here it is, over 40 years later, and we’re still having similar conversations about how to balance growth and sustainability. In tech circles, the conversation has long skewed towards the ways technology can help us be more efficient energy consumers and how we can embrace business practices that also embrace sustainable values. Yet that has not always come to fruition.

For data centers, the businesses who partner with them, and the consumers who enjoy the enabled services, that means increasing scrutiny on resource utilization and “green” initiatives. It’s essential to make disaster recovery plans because disasters seem likely.  We are proactive in that regard.

Also Read: Digital Transformation Trends

Climate change which is, and has been, accelerated by human causes, is, in part, responsible for the increase in natural disasters from violent storms to mudslides to severe flooding. Not only must we have disaster recovery plans, but data centers and the regions where they are built will, looking forward, need to consider disaster prevention plans. That is, how can we utilize resources to build toward a more sustainable data center industry?

Quick Links

What are the Environmental Concerns Facing Data Centers?

Every industry in the modern world must weigh risks and benefits, and must find a balance between the services or products it delivers and the impact they have on the environment. For many industries, those concerns include emissions, power usage, raw material consumption, water usage/pollution, waste production (though largely post-consumer), and land use. Data centers are no different.

Emissions and Power Usage

As little as 5 years ago, it was estimated that emissions from data centers would soon rival that of the airline industry and that, by 2025, they would account for 3.2% of the world’s carbon emissions. Further, in 2018, it was found that data centers accounted for 17% of technology’s overall carbon footprint.

While the power/electricity industry is responsible for 26.9% of greenhouse gas emissions overall, due mostly to our reliance on fossil fuels, it’s unclear how much of a data center’s carbon footprint overall is related solely to power usage. But, when one considers that data centers are running full throttle, 24x7x365, data centers alone are responsible for 2% of the country’s electricity usage.

However, while experts predicted that data center power consumption and emissions would grow rapidly with their expansion, between 2010 and 2018, consumption rose by only 6%, a fraction of what was anticipated. In large part, this is attributable to several factors related to the industry and technology itself.

Raw Material Consumption

There are several environmental concerns regarding the computer industry and the raw materials required to build parts. With an increasing demand for growth and more hardware, it’s easy to see how raw material consumption could be considered an area of concern.

First, while computers are made from some common metals, some of the metals used in computer parts are incredibly rare. In fact, hafnium, used in processors, may run out within the next decade. While there are potential substitutes, it’s worth exploring potential alternatives so as to not deplete the resources we currently have available.

Further, the acquisition of these materials is often an environmental concern, as well as geopolitical and human rights issues.  Not only does the mining of these materials cause toxic water pollution, but gold, used in computer parts, creates 80 tons of waste for just one ounce.

However, much like emissions and power usage concerns, the groundwork is being laid to minimize the impacts from hardware demands as well.

Water Usage

Many of us remember the statistics we learned in grade school about water. While the world is 70% water, only 2.5% of it is fresh, 1% accessible, and sadly, these days, even less of that is potable. In short, fresh water is a limited resource. In fact, 40% of the world today is facing water scarcity issues.

And yet, between water for cooling servers and the passive consumption from power plants, data centers use a stunning amount of water. In fact, in 2009, Amazon estimated that one of its massive data centers was using a shocking 360,000 gallons of water a day. That’s one data center. While not all data centers are the size of Amazon’s, Google’s, or some of the other tech behemoths, there are still 8.4 million data centers worldwide. That’s a lot of water.

Waste Production

One often overlooked consequence of massive data centers is their contribution to a growing e-waste issue. It’s estimated that 44.7 million tons of e-waste was created in 2016, with just 20% of that recycled. Mega data centers may be utilizing millions of servers (and the components to support them), though certainly most data centers are scaled to meet the needs of the organizations they serve.

As most hardware has a 5-10 year lifecycle and then is repurposed, data centers may not contribute a significant amount of electronics to this growing problem, but certainly can look to be part of the solution moving forward.

Land Use

A final environmental issue facing data centers is their land use. Again, the large tech companies are purchasing land by the 100s of hectares for data center build-outs. While that breaks down to under a square mile, the impact of overdevelopment on the environment is significant. Not only does it deplete resources, but it destroys vegetation, and creates a chain of other issues that contribute to climate change and weather disasters.

What Resource Concerns Should Data Centers Have?

The primary concern among data centers right now is power usage. A quick Google search will reveal that of all the environmental issues facing the industry, its massive consumption of power is a stumbling block. Further, while data centers are using so much energy, for many, power use is not only not a priority, but many aren’t even aware of their energy efficiency.

However, cities and regions growing in popularity for data centers, which are needed to support growing population centers and their digital needs, are facing significant challenges with water as well. In fact, Austin, which suffered drought issues in the past, is looking ahead to water conservation as water scarcity is expected to become a growing concern in that region.

While many of us enjoy the digital advantages and tools that cloud computing, supported by data centers, provides, more communities are becoming aware of the large environmental footprint they have, and warn against inviting them in without studying impact first.

In essence, there are growing concerns across all populations with investments in the data center industry regarding resource utilization and ways to create more sustainable growth, as growth is inevitable.

What is Resource Utilization?

The term resource utilization has to do with efficiency. In this case, it refers to natural resource management and utilization as it relates to the ways data centers function. As noted above, significant resources are needed for the daily running of a data center (power, water, land use), etc. and there’s no doubt that it’s having an impact on the environment.

How Can Data Centers Optimize Resource Utilization and Meet Resource Concerns?

While concerns will continue to grow, likely on pace with data centers themselves, there are options for the industry to respond to these concerns. In fact, some measures are already in place. Further, as climate change accelerates and governments and industries respond, prioritization of these initiatives is likely to follow as well.

Emissions and Power Usage

Several changes in the industry are already decreasing energy consumption. Not only is hardware becoming more energy-efficient, but innovations in storage, like helium drives, also promise energy savings. Further, server virtualization software has enabled applications to run on single servers and large-scale data centers often feature more energy-efficient cooling systems.

However, most experts agree (and Google is utilizing AI for this purpose, a process called auto-scaling), that better modeling and better tracking of energy consumption are a key part of any sustainable initiatives. This means monitoring the Power Usage Efficiency (PUE), considering locations (cooler climates require less cooling), and relying upon alternative energy sources as some of the greenest data centers do.

Finally, in a setup that functions a bit like carbon credits and emissions trading, a data center in Norway is utilizing its “waste” heat to heat nearly 5,000 homes in the city. Like many climate challenges faced by nearly every industry, innovative and creative problem-solving will inevitably be one of our greatest strengths.

All of these measures, designed to limit power usage created by the hardware and cooling needs will, hopefully, translate into decreased emissions as well.

Raw Material Consumption and Waste Reduction

One of the best ways to reduce the consumption of raw materials is to not use them at all. For that reason, one of the best ways to decrease consumption is to reuse existing materials. Some components can be recaptured and recommissioned to be used elsewhere, and as noted above, some are simply recommissioned for other uses.

When recommissioning parts isn’t possible, recycling e-waste becomes another option. In recycling and reuse processes, they can not only reuse parts but also break down and separate plastics from metals, allowing for the reuse of raw materials as well. However, there are costs, both financial and environmental, associated with these processes as well. That said, the e-waste recycling industry is one that is solidly growing.

The final avenue left to explore here is simply the production and manufacturing of hardware itself. Over the last decade, significant innovations in electronics production has seen a decrease in the plastics and harmful materials used to manufacture, and, with any luck, similar steps will be taken in the computer industry as well.

Water Usage

As noted above, the copious amounts of water used by data centers is primarily used for the cooling process, so any measures to reduce water consumption will likely focus on how to reduce those cooling needs as well. That might include more energy-efficient cooling systems, but other options are also on the table.

One way some of the larger companies have handled this is through geographical considerations. As noted above, by locating data centers in colder regions, cooling needs are reduced. This practice has been going on for a few years now and is expected to be a viable option in the future as well.

However, it’s not feasible, especially with edge computing on the rise, to locate all data centers in remote and cold locations, so other options must be explored as well. Additional methods already in practice include the recycling of wastewater, the use of cooling towers, and air cooling systems.

Much like the rest of the industry, innovations will grow here as well, particularly as water scarcity becomes a much larger issue across the globe.

Land Use

Obviously, as mentioned above, moving to remote and cold locations changes the demand for land resources. However, as this isn’t necessarily feasible or viable for many data center operators, other options need to be considered.

One of the options on the table is colocation which enables multiple companies to utilize the same data center, thereby eliminating their need to build their own facility.
Another option, already in use, though very rare, is subterranean data centers, like those seen in Norway and under the Swiss Alps.

Further, the industry must examine the reasons for land usage and decrease it where possible. For example, many larger data centers overprovision with equipment like diesel generators which take up a significant amount of space. Being careful to build out only what is needed is a fundamental piece of this puzzle.

Finally, tapping into existing structures and “recycling them” is another option, as is land reuse. Former military installations have been seen as potential location matches for existing structures while land that is unsuitable for other purposes may be utilized for new construction.

As with all the other sustainability challenges, the industry is prepared to meet this one as well with several options to meet a variety of needs.

Also Read: SOC 2 Compliance Checklist

The Impact Edge Computing May Have on Resource Utilization

Edge computing, which relies fairly heavily on the existence of plentiful, but smaller, data centers outside of urban and population centers presents both opportunities and challenges for the data industry. However, one of the advantages of edge computing’s flexibility is also one of its greatest strengths in combating some of the environmental concerns related to data centers.

While an initial examination, on the surface, of edge infrastructure might garner a response akin to more data centers equals bad news for the environment, edge computing actually may require less natural resource usage than traditional data centers. This has, primarily to do with the decentralized nature of edge computing.

More specifically, the growth of edge computing will allow for an increase in the Internet of Things (IoT) and, as such, increase the computational power of smaller devices within our homes, cars, and even in our hands. This smaller level of computing and a decrease in the need to transmit data for processing at a centralized data center means those data centers don’t have to work as hard. In fact, recent estimates suggest that over 90% of data was transmitted to a centralized data center, but with the rise of edge computing and the IoT, that number is likely to drop to 25%. In other words, most of the power usage will be in the devices rather than the data center.

Further edge computing has other advantages as well. Edge data centers are designed for efficiency. Because resources are restricted to what a device can handle and the server is optimized to handle that it’s using fewer resources. Further, edge computing, the IoT, and device capabilities will only grow to enable optimization and reuse of existing hardware, again saving energy consumption and other resource needs.

Ruben Harutyunyan

Back to top