Chapter 8
Cloud Computing and Sustainability: Fact Versus Fiction
The nation behaves well if it treats its natural resources as assets which it must turn over to the next generation increased, and not impaired, in value.
— Theodore Roosevelt
This chapter does not focus on climate change. You bring your own beliefs to the table with that political football, and it doesn’t really matter if you consider yourself “green” or “anti-green.” Right, wrong, or indifferent, it’s time to set aside what you think you know about the topic to focus on the factual business reasons to pursue sustainability goals.
The concept of sustainability comes down to perception versus reality. The perception is that we all want to do what’s right, which is to lower our carbon output to stop or slow down climate change. This balancing act translates into many heated discussions about which paths a business should take to achieve these competing goals: good for the planet, or good for the business.
A happy coincidence is that doing things to make a business more sustainable often saves money as well. Plus, if businesses promote their achievable sustainability goals, they gain PR credibility. Finally, the environmental, social, and corporate governance (ESG) scores being assigned to companies can make or break your enterprise’s ability to obtain investments and new customers. Those are facts.
Figure 8-1 shows how public interest in sustainability accelerated over the last 10 years, which happens to be the same time interest in cloud computing grew. Cloud now approaches the 50 percent mark in the way that we do computing. The world began to focus on limiting carbon output around the same time cloud computing began its rapid growth. A frequent question soon arose: Is cloud computing green?
FIGURE 8-1 Public interest in sustainability grew at the same time cloud computing grew. The obvious question became: Is cloud computing good or bad for the planet? The answer, as usual, is, “It depends.”
IT in general outputs a significant amount of carbon and has for a long time (see Figure 8-2). IT’s heart beats on electricity, and fossil fuels still produce most of our power. Voila, carbon output. This will likely continue in the short term because renewable energy sources (such as wind and solar) are unlikely to become efficient enough to overtake traditional means of IT power generation until the long term.
FIGURE 8-2 IT has a huge impact on carbon emissions. Although there are green benefits to IT systems, this chart depicts the actual relationship between computing and sustainability from 1960 to 2008.
Figure 8-2 depicts an aggregation of trend data over the time period of 1960 to 2008, or the start of cloud computing, including the growth of IT and the corresponding rise in IT energy usage. This period starts when businesses, educational institutes, and the public started to use mainframe computers until 2008 when computers and technology became systemic to everything we do. The year 2008 also happens to mark the rise of cloud computing, for all practical purposes. Of course, SaaS clouds predated 2008, and time sharing predated SaaS clouds. However, 2008 is when the industry saw the value of cloud technology shift and began to track it as a new technology thread, which leads us up to today.
There is some good news, such as the fact that today’s servers are much more efficient than servers of 20 years ago. Today’s power supplies, processors, and storage systems are engineered to use less power. The problem is, we’re using much more of them. If you use 100 times more electronics than you did 10 years ago, power consumption will go up despite the devices being more “power optimized.”
Traditional in-house data centers also continue to leverage more and more servers, where hardware capacity is typically purchased well ahead of the need for that capacity. In-house data centers usually operate and power servers at 5–10 percent utilization. Thus, we run on hardware that is underutilized by default (see Figure 8-3). Yes, the utilization will increase over time, but it will never reach full utilization. In-house servers use the same amount of electricity at 10 percent capacity as they do at 99 percent. Therefore, every percentage of underutilization is a waste of power.
FIGURE 8-3 When using traditional approaches to computing, you need to purchase hardware well ahead of need. As you can see in this chart, those are the square lines, when capacity is raised by purchasing hardware outright. Thus, the hardware is running and using power, and the unused capacity is wasted.
You’re probably guilty of this yourself, if that desktop computer or laptop on your desk runs idle 80 percent of the time. Perhaps you use only 20 percent of storage and perhaps as much as 50 percent of your processor and memory. That extra capacity exists because you’ll need it at some point. Until then, those nonutilized hardware resources consume power to function, but you’re not leveraging that much of the capacity that you’re powering.
Initial Thinking: Cloud Data Centers, Bad
In the early days of cloud computing, representatives from various environmental organizations often contacted me. Because I am a cloud computing subject matter expert, they wanted to express their concerns that a massive amount of data center construction was underway to support emerging public cloud offers by the major cloud computing providers. They claimed that these new centers created higher power consumption, which would lead to more carbon and thus more climate change.
In some respects, they had a point. For decades, we built in-house data center after in-house data center to support the pre-cloud growth of enterprise IT. Now, the cloud providers were doing the same thing. Yes, those new cloud data centers will make things worse. However. Although the explosion in data center construction projects alarmed the environmental people, they often overlooked the big picture.
The purpose of this chapter is to explain what the growth of cloud computing means for sustainability.
More Data Centers, Bad
Let’s back up about 60 years when computing emerged as an acceptable business tool. Few businesses could afford to have their own data centers, so they accessed data centers owned by large tech firms (IBM, Oracle, and so on) or universities via time sharing, serverless “dumb terminals” that emerged in the 1960s and 1970s. Enterprises used these computing systems as a utility and paid only for the time they leveraged the computers (and, of course, the power to access the systems).
Because this was a time-sharing service, businesses usually tapped into these services along with many other enterprises via a shared multiuser server.
This “old days” multiuser configuration is a bit different than multitenant, which is the model that public cloud computing services employ. An enterprise’s cloud computing services can share coarse-and fine-grained IT resources such as storage, compute, AI, databases, and so forth, as if the enterprise owned these resources.
Conversely, traditional multiuser time-sharing services leveraged dumb terminals that provided a view of the computing system, typically as applications, that a few mysterious people in IT used to support business operations like customer databases. Those systems now seem quaint when compared to the processing power of almost any smartphone.
We were good at sharing computing resources in the early days of computing because we consumed less power per computing user. In other words, we started the computing journey greener than we are now because we were much better at sharing powered hardware resources.
So, what happened? Figure 8-4 shows the trend we dealt with for almost 50 years. We went from extensive time-share resource use to owning our own hardware for the exclusive use of an enterprise. This trend drove the growth of data centers to house and operate these new hardware resources that were solely owned by an enterprise. At the same time, the use of time-sharing systems diminished to almost nothing. Today, we’ve come full circle, as time-sharing systems are back in vogue with the use of cloud computing.
FIGURE 8-4 History repeats itself. We went from a time-sharing model in the early days of computing when data center costs were beyond the reach of most businesses. Then we shifted to privately owned equipment and data centers. Since 2008, we have gone back to sharing in a different way with cloud computing.
As the number of in-house data centers grew, most ran grossly underutilized hardware resources that consumed power and produced carbon. Remember our discussion of 5–10 percent utilization? You can see that cloud computing had a perception problem in the early days of cloud.
In comparison, data centers that are built to house cloud computing services can be shared by tens of thousands of users from hundreds of companies. A single company could use 10,000 times the power and space for an in-house data center than the equivalent usage in multitenant cloud computing services. This is what makes cloud computing truly green: sharing.
It’s like the services we see today that enable excess capacity sharing for underutilized vacation homes, private car usage, and more. Cloud computing’s capabilities to easily share and manage resources allow us to do more with less. In the case of enterprises, cloud allows more computing and storage and other types of business processing with fewer hardware resources and much less power. Yes, public cloud providers may have more data centers to house their cloud computing services and offerings. However, they operate at many times better power efficiency than traditional approaches to in-house data centers where everything is owned and nothing is shared.
More Shared Data Centers, Good
Figure 8-5 depicts two very important past trends. As the number of systems and the number of enterprises that owned data centers grew, we got a lot worse at utilizing those resources.
FIGURE 8-5 As we increased the number of owned systems and expanded data centers, we became very bad at system utilization that led to underoptimized power utilization.
Virtualization is another tool enterprises use to better utilize in-house data center resources. The capacity utilization problems remain with virtualization, in that we must purchase computing assets well ahead of need, as discussed earlier. However, it was a step in the right direction. Indeed, virtualization benefits were often confused with cloud computing, with the solution patterns being similar. However, multitenancy is more efficient than virtualization considering that more than one user, or one organization, can use the same resources at the same time.
Conversely, public cloud services can adjust utilization around changing needs. Many enterprises need additional compute power at specific times of the year, say, end of quarter/fiscal year, or during seasonal/holiday usage spikes. A public cloud service can provide an enterprise with what’s needed, when needed, and then remove those additional resources when they are no longer needed.
Cloud computing will not save us from climate change. However, it’s a step in the right direction. Cloud can at least turn the tide of wasteful enterprise computing operations.
Figure 8-6 illustrates the growth of public clouds from 2008 to today, which leads to higher system utilization through shared resources and lower use of power per cloud user and per enterprise.
FIGURE 8-6 The multitenancy models that public clouds employ allow for much better compute and storage resource utilization, meaning that we can do much more with the same amount of power-consuming resources. This is a huge step in a more sustainable direction.
This is not rocket science. We have always understood that our ability to share resources leads to less waste. The prevailing wisdom was that everything needed to be owned and not shared for security reasons, and that any other approach was risky. It turns out that cloud computing is typically less risky than traditional approaches, with the added benefit of being greener as well. Let’s break this down a bit and look at all aspects of cloud computing for sustainability, including the good and bad aspects that need consideration.
The Politics of Sustainability
So, who said this wasn’t a fun topic? The political aspects of sustainability mean that those who try to create a sustainability path using cloud computing (or other technologies for that matter) can end up with a solution diluted by so many different opinions and debates that the white noise pushes any sustainability progress offtrack. Here’s how to manage a path forward.
First, don’t get into dumb arguments. Sustainability for the enterprise is not about who said what, it’s about facts that follow along with many of the ideas presented in this chapter. Don’t just provide an opinion; provide data to support your opinion. If there are other ideas, they should be considered if they also come with legitimate data.
Second, out of all the presentations I do and articles I write that discuss sustainability, the audiences or readers often focus too much on the topic of sustainability as a concept. I really don’t care what people’s opinions are on the concept of sustainability. Instead, we need to discuss strategies to reduce power consumption and better align power consumption to capacity needs.
That’s why I typically don’t get involved in discussions about the politics of sustainability and suggest that you don’t either. Focus on reality, and objectively consider the negative and positive aspects of cloud usage for sustainability.
Power consumption optimization leads to the by-product of sustainability, which in turn saves money as much as it saves the planet. We must progress toward that objective for many reasons.
Finally, Sharing Is Possible
The current practice of building private data centers to house hardware that we may eventually use is itself unsustainable. Having those systems standing by and consuming power is not a good use of resources (again, see Figure 8-3). We shouldn’t pay for power that doesn’t create any business value. Incidentally, it’s also not a great thing for the planet to waste power.
Cloud computing is not the ultimate answer of consumption-side sustainability, but it’s a much better alternative to power-hungry compute and storage services running at 5–10 percent capacity. Those of you who run data centers, I urge you to look at the utilization reports for your standard and even virtualized servers to directly see the capacity and power that’s being wasted. Remember, these reports do not account for the power needed to cool the data center, lighting, or the fact that people must commute to the data centers to maintain this equipment.
Data centers are necessary, but they are a huge power draw. Most consume as much power and water as a small town. It’s better to have hundreds or thousands of enterprises use one public cloud data center than it is to have hundreds or thousands of individual data centers.
Once upon a time, we were very good at sharing compute and storage. We moved away from shared services when our data needs grew more sophisticated and compute resources became cheaper. We wanted complete control of our compute resources from the software to the hardware.
Cloud takes us back to the economies of scale with shared resources that are now better than solutions we leverage on owned data center resources. Also, R&D investments made by technology companies have shifted to cloud services over the last 10 years. Some enterprise software providers have already “sundowned” in-house data center services, and others continue to set expiration dates, which means you will soon have no choice but to move to cloud-based systems.
There is no bad news here regarding sustainability. Cloud moves to a more efficient technology consumption model, with additional applications and resources that are (or will soon be) far superior to those in an in-house data center.
The Greenness of Multitenancy
Multitenancy is what makes cloud computing sustainable. Simply put, it’s the ability to share IT resources using mechanisms that allow for better sharing approaches, where the resources are as close to 100 percent utilization as possible. Although the multitenancy systems employed by each specific cloud provider are a bit different, they all have the same basic objectives:
Share resources but disguise all sharing aspects from systems and humans who leverage those resources. Although this seems obvious, some multitenancy systems are very bursty and thus don’t function well as sharable cloud services.
Protect from other processes that you don’t own. No other process is allowed to affect your processing such as a program running in memory or sharing a processor.
Share all I/O resources including storage, such as databases running on shared storage, memory, network, and anything else that may consume and produce data.
Allow scaling to occur without interruption; more resources are automatically allocated as needed to support a greater processing load. Those who use a multitenant system (typically a cloud provider’s system) don’t have to think about when and what to allocate to obtain greater scale.
Return resources to a resource pool after use. For example, at the end of a holiday season, we can automatically eliminate excess capacity usage on a shared storage system, and that resource is returned to the public cloud’s resource pool where it can be shared by other processes and humans.
Figure 8-7 illustrates a multitenancy model’s nonlinear use of physical servers and workloads. Again, the core advantages here are resource sharing and offloading ownership, operations, and maintenance of hardware and software from enterprises to cloud providers. The use of excess capacity in a multitenant system drives more power efficiency out of the same number of physical servers. Yes, virtualization could create a multitenant structure for one enterprise. However, those who believe virtualization can replace public clouds often miss the bigger picture when it comes to efficiency of scale, including power consumption, considering that cloud computing makes resources more sharable and thus more efficient in terms of lowering power consumption across many different companies.
FIGURE 8-7 Multitenancy models use fewer power-consuming servers, which means that public clouds knock sustainability out of the park.
Not All Clouds Are Equal
A dirty little secret in the world of cloud computing and sustainability is that all clouds are not created equal in the race to sustainable results. Even different data centers owned by the same provider can be at different levels of carbon efficiency, or the same centers at different times of the day or week or year. This happens because each cloud provider must deal differently with load requests from cloud computing users depending on
The current load the cloud provider is experiencing. Other centers in different geographical locations and servers within those centers may be activated to deal with increasing loads. Those centers and servers could use more carbon-producing “dirty” power than the provider’s primary centers.
The availability of green power around the location (point of presence) of the cloud computing data center you leverage. A provider might show you a wind farm and solar-powered facility that drives some of their data centers.
However, in many areas it’s still impossible to provide renewables at sustainability levels; thus, you’re left with fossil fuel–generated power. Cloud providers do not try to hide that fact from you or the public. But if you choose a certain point of presence that’s close to a major factory your enterprise owns, the provider could have few opportunities to create renewable power in that location.
Most cloud providers cannot control the kind of power being placed on the grid. During off-peak hours, you might leverage primary power on the grid from nuclear power, which is as green as renewables. If consumption moves to a certain level, your power sources could also include fossil fuel–generated power to augment the cleaner power sources.
In most instances, the cloud providers just purchase power. The power companies still control the source of that power. However, some cloud providers do generate their own power using renewables, but typically not for all points of presence.
A simple way to measure if or how much a cloud is sustainable is to consider Figure 8-8. Answer the power consumption questions with yes/no and percentages if possible, and this decision tree will help you figure out whether your public cloud provider supplies sufficient carbon-neutral computing to meet your sustainability goals. Also, assume the enterprise data center and the public cloud provider will hold at an average utilization rate of 50 percent. This means that 50 percent of the hardware will be utilized for workloads. Note: This is just for the sake of this example. In the real world, enterprise data centers average less than 20 percent utilization, and most public cloud providers are north of 90 percent utilization.
FIGURE 8-8 When it comes to development with an eye on lowering power consumption, there are many issues to consider.
It would be remiss of me not to mention that we can often fool ourselves with a binary rating. It’s almost never completely carbon neutral or completely free from the use of dirty power. Instead, it will be somewhere between the extremes of more green or less green.
Continuing with our example in Figure 8-8, we have two location choices to host a system: public cloud or an enterprise data center. (For now, let’s ignore the fact that the choice is never that simple.)
From there, we can find a path through other factors that determine if the place we host our computing and storage services is green. Does the data center (public cloud or enterprise) leverage fossil fuel or renewables? An owned data center that uses 90 percent renewables would win out over a cloud provider that uses only fossil fuel–generated power, or vice versa. Know what you’re using.
Figure 8-9 shows an example of a path through this decision tree that determined a less sustainable conclusion with a public cloud provider that powers their data centers with fossil fuels.
FIGURE 8-9 Although many consider the use of public clouds to always be the most sustainable choice, in this scenario the public cloud burns fossil fuel for power. The public cloud route in this instance is not optimized for sustainability.
Figure 8-10 is an example of a more sustainable outcome, even though you leverage your own data center that happens to be powered by renewables. So, this non-cloud solution ends up being more sustainable than some cloud solutions.
FIGURE 8-10 Although it may seem counterintuitive, data centers that employ renewable power could be the best choice for sustainability.
If this approach seems overly simplistic, you’re right. However, it’s the way most enterprises evaluate their deployments for sustainability, cloud or not. This kind of static analysis is helpful for basic decisions and to understand what’s sustainable. However, it’s often unrelated to the real world, and actual results will be very different, especially when using actual utilization versus an equated average.
A better model to consider is Figure 8-11, which has an added dimension: utilization. The reality is that utilization is not static. Ever. Although we can assume static utilization for some models, a more realistic model would be to look at different levels of utilization for different approaches to computing (cloud and enterprise data centers), and for different types of power being supplied at different levels of utilization. This example uses 25, 50, 75, and 100 percent levels to consider.
FIGURE 8-11 In this example, we look at utilization of the hardware to determine if a deployment is sustainable. This model more closely resembles the real world.
At higher levels of utilization, more will get done on a single, powered piece of hardware. The higher the utilization, the greener the deployment. Remember, cloud computing is typically the more sustainable solution because it can share hardware resources and thus do more with less. This is the most important point in this chapter.
What’s interesting about this model are the very different conclusions made from different levels of utilization that are more reflective of the complexities of these decisions. For example, an enterprise data center that leverages renewables at less than 25 percent utilization may have a larger carbon footprint than a cloud solution that leverages fossil fuel at near 100 percent utilization. Why? It takes carbon to build the hardware (more than you think), and to build the data center as well. Just using renewables does not mean a data center is fully carbon neutral…and very much not carbon neutral if utilization is suboptimal. Indeed, some enterprises can achieve near 100 percent utilization with cloud providers that use dirty power and end up with a more sustainable solution than a deployment that uses renewable power sources.
Here are a few things to keep in mind:
First, sustainability is often not what it first appears to be. Public relations and marketing campaign claims of sustainability can be incorrect for both cloud providers and traditional data center real estate providers. You must consider all variables in your modeling. Account for the carbon impact of building a data center, the data center infrastructure, and the impact of building all the hardware that will be contained in the data center.
Second, sustainability is a dynamic process. What’s true today may not be true tomorrow. Literally. You must reanalyze the true carbon impact of your cloud and non-cloud deployments on a regular basis. Not every day, but perhaps annually. For example, the source of power that supplies the grid often changes based on increasing demand. While you might brag about how green your cloud deployment is in 2025, that may not be the case in 2027. If cloud computing sustainability is an ongoing concern for your company, then it’s a wise call to add a sustainability monitor to your cloud operations (CloudOps). This is especially true if your enterprise is graded on sustainability by outside regulators, or if it’s monitored by the press. The enterprise could end up with a huge PR problem if you’re outed for using dirty power, no matter if you meant to or not.
Resource Optimization and Sustainability
Much of the sustainability effort concentrates on ways to reduce our carbon footprint by using more sustainable compute and storage methods such as cloud-based resources. There are more obvious things to consider as well. The big one is to reduce the number of virtual or real server assets required. This means doing the same (or more) with less. Better optimizing the number of resources reduces your need for power. Let’s explore this a little.
A hiding-in-pain-sight secret of enterprise computing is that we always use more compute and storage resources than we really need. This comes mainly from a lack of architectural planning that leads to inefficiencies, that in turn cause more resources to be deployed than needed. For instance, that old data warehouse system that was built in the 1990s using 500 terabytes of storage because the same data is replicated a dozen times during processing. If redesigned, it could do the same or a better job with only 50 terabytes of storage.
The system works and nobody wants to reopen that issue, so the problem gets kicked down the road. When this old data warehouse system is lifted and shifted to the cloud, we’ll pay a much larger data storage bill than we should. Although that’s bad enough, we must also understand that this top-heavy system has an impact on power consumption as well. If you’re green, the old data warehouse system isn’t. If you’re just a pragmatic businessperson and good corporate citizen, the system is also unoptimized for its business value. Fixing this dinosaur could kill two birds with one stone.
The average enterprise has between 100 to 400 systems in dire need of better optimization for the resources they leverage. What’s likely to happen to these systems is that the cloud will shine a new light on their inefficiencies. In the past, resource-hogging systems installed within enterprise data centers did not seem to be a problem. We had to power and operate those dedicated systems anyway, so it didn’t matter if we leveraged 67 percent capacity versus 22 percent. The bill at the end of the month was the same.
When these systems are moved to the cloud, the inefficiencies become much more visible. That’s when you get a surprise $20,000 storage bill for a specific system that will generate many more questions about how to reduce that cost. Eyepopping cloud bills like those are the drivers of the current trend toward FinOps, which can monitor and optimize cloud costs. FinOps encompasses a dedicated set of people, processes, and tooling that will ask the hard questions until some sort of better optimization occurs. The good news here? When you solve the efficiency problem to save money and make the finance team happy, you’ll also reduce carbon output. Win/win.
Green Application Development?
Being an old developer, I recall having to deal with fewer resources. Memory and storage were expensive commodities back in the day. Thus, we were forced to conserve the number of resources we could leverage, which led us to build and deploy much more efficient applications that used a minimum number of resources. We didn’t even consider developing applications that used power more efficiently; that outcome was simply an unplanned by-product.
Now we want more sustainable computing platforms to run our applications and store our data. We’re also finding that the applications themselves can be optimized to burn less power and thus become more sustainable. This is known as green development (or similar terms), which means that we focus on the optimization of applications and data storage to burn much less energy than unoptimized applications. It’s a huge paradigm shift for a generation of developers who looked at resources such as CPU, memory, and storage as cheap commodities. Beginning with the application bloat that began in the mid-1980s, application developers just kept growing the number of resources that software needed to operate, which drove the need to upgrade to larger and more power-hungry hardware resources (pre-cloud). Just try running a modern operating system on an older PC and you’ll see that it’s almost unusable. Those who built the operating systems in those days operated with the understanding that additional resources were cheap and readily available. That’s why they created software systems that were much less efficient than those built in the 1970s and early 1980s.
“Bloatware” is what many now call it, which refers to the fact that many developers got lazy when it came to resource optimization. Bigger and better CPUs and memory systems could always accommodate the resource-hungry applications that were created over the last 20+ years. What we also failed to notice? Bloatware applications also require larger and more powerful resources that burn more power…perhaps a lot more power.
Today there is a new focus on green development, both on public cloud platforms and traditional systems. The idea is simple. A check is done during development, typically within a DevOps toolchain, where the applications are tested for security, performance, reliability, stability, power optimization, and many other things. A code analyzer scans through the code and analyzes the behaviors of the executables to determine how optimized the application is around power consumption. There are huge sustainability gains that can be made here that will also lower your cloud computing bill.
Green development has several benefits. First, it puts the burden on the developer to write and deploy more efficient code. Doing so will also optimize the resources used. Note the word optimize versus minimize. Developers must learn to do more with less by using proven approaches to lower the number of resources leveraged by a single application and to do so in a way that nothing else is sacrificed. Developers should not feel obliged to sacrifice performance and usability to optimize resources. If that’s what happens, the developer is not doing “green development” right. Creative approaches and tooling are now available that allow us to program both better and more optimized code. The bigger objective is to build and deploy applications that provide the same or better business value with fewer resources and less power.
When optimized applications and systems are operational, FinOps should monitor resource optimization for ongoing sustainability. Remember, in optimizing our cloud spending, we also optimize power consumption, and thus we can monitor both at the same time using many of the same cloud operations tools. Another win/win.
Multicloud as a Sustainability Weapon?
The use of multicloud to find greener cloud and traditional computing solutions is an emerging area of discussion. If done correctly, multicloud can leverage any number of providers and services that we consider best-of-breed. We can pick solutions that leverage green power sources and thus provide a boost to sustainability.
Figure 8-12 depicts how this works. In this multicloud deployment example, we have three different clouds that run under our single layers. This may include cross-cloud security, operations, governance, FinOps, and other common services that holistically manage the multicloud deployment. I discussed this in Chapter 6, “The Truths Behind Multicloud That Few Understand,” so we won’t dive in too deep here. The important thing to remember? When deployed correctly, a multicloud can use any number of services for any number of reasons and leverage the best services for the job, including best for functionality, performance, cost, and, yes, sustainability.
FIGURE 8-12 Multicloud deployments can leverage cloud services that provide the best sustainability rankings. Cloud service brokers are beginning to track sustainability as a ranking for services that can be promoted over less sustainable services.
Related to what you see in Figure 8-12, let’s say we leverage a database from a point-of-presence where only a fossil fuel—powered energy source is available. However, another cloud provider supports the same databases but runs within a special cloud computing data center that only leverages renewables, including a well-publicized wind farm that was created specifically for that cloud computing data center.
Multicloud allows you to capture a sustainability advantage by using that specific database, running on that specific cloud provider, over the other cloud brand that does not offer a sustainable solution. Of course, you need to consider cost as well, including the ability to support the core business requirements, which could include growth, scalability, and reliability. Don’t make the mistake of leveraging one cloud provider over another for sustainability reasons only to realize that specific provider is less reliable and less cost effective. You need to consider all factors when selecting a cloud provider and cloud service for a multicloud deployment. Sustainability can’t be the only evaluation criterion.
What Is Your Real Impact?
Sometimes I think we focus so much on sustainability that we miss more important issues that need to be addressed, such as the ones listed in Figure 8-13. We must approach sustainability with a holistic understanding of what it means to be truly sustainable and recognize when we’re working toward the wrong objectives.
FIGURE 8-13 Sustainability is not just one thing; it’s many. People, culture, and technology must work together to make sustainability happen. It’s more complex than many understand, and the danger is that all the focus on sustainability will have unintended consequences.
For instance, we might focus too much on the power source for our system deployment and not consider the true business impact. I’ve seen enterprises move to more sustainable deployments, typically in search of a better rating from outside agencies, and at the same time give up reliability. That solution could cost the company more in bad PR due to the cost and stress of dealing with system outages and outraged customers than any green advantages they gained.
Sustainability needs to be a change in culture more than a change in technology. We need to educate everyone in the company about doing things that are more sustainable, and best practices and approaches that work for other companies. For anyone who grew up in a household where a parent figure constantly harped about turning off lights when you left a room until it was second nature, you know what I’m talking about.
In the enterprise, teach the mechanisms staff can leverage today to make more sustainable choices for the business, such as cloud computing itself. This also means looking holistically at the business, knowing that a few sustainability gains may not justify other negative impacts on the business. The law of unintended consequences often arises if you make sustainability the main goal. Recognize when you may hurt other areas of the business and even cancel out any sustainability benefit by being penny wise and dollar foolish.
For example, let’s say you move all data storage to a single public cloud provider for their sustainability benefits. When audited, you discover the provider is violating several data privacy regulations and just flunked an audit during the first month it hosted your data. Aside from the numerous fines the provider may incur on your behalf, the carbon output required to fly enterprise executives and a legal team to that jurisdiction to clean up the legal and PR mess will remove any carbon benefits you should have received from that move for the next 100 years. But at least your heart was in the right place. Remember, in that and many cases, moving to sustainability can backfire in ways you can’t anticipate. If my suspicions are correct, we’ll see many of these types of sustainability case studies in the next 10 years.
Sustainability is a hot topic right now, but we can’t blindly chase a politically correct ideal that ends up hurting the business. Enterprises tend to move quickly when a regulatory requirement will also provide good PR and save money. However, they will find that sustainability is much more complex and less forgiving than most people realize. The name of the game here is to find a trusted path that is sustainable with a net benefit to the business.
Hopefully, you got the overall sustainability message in this chapter, even though we focused mainly on sustainability’s impact on cloud computing.
Call to Action
When it comes to cloud computing, we need to weaponize automation to provide better sustainability. Rather than rely on humans to shift workloads and data stores to get the best carbon impact, we must leverage automated tools to dynamically make those decisions. Those tools can use advanced approaches such as AI to ensure that we’re as optimized as we can possibly be, with all things considered in terms of what the business needs.
Figure 8-14 shows the rise of automation along with the rise of cloud computing. Automation provides us with mechanisms that can support more sustainability—for instance, the ability to dynamically move workloads from cloud provider to cloud provider in response to better sustainability rankings, as well as lower costs and improved reliability. These automated processes need to consider everything about the business to make the right decisions. If we focus them solely on ways to be more sustainable, they’re likely to inflict unintended consequences that can remove any benefits.
FIGURE 8-14 Widespread automation manages decreasing and increasing uses of cloud and even traditional IT resources, as needed, as demand ebbs and flows. By not under-or overserving the enterprise with resources on-demand, automation provides the most sustainable path available today.
Automation provides an even further benefit. We can leverage automation to find the right number of compute, storage, and databases instances that need to be launched to support our enterprise IT needs. These needs will likely grow and contract day to day and month to month. Automation can better align need with the number of resources leveraged. It’s this kind of automation that’s often overlooked. We need to consider what’s running versus where something is running to gain sustainable results with long-term benefits to the business and the environment.
This effort is about more than just sustainability. This is about being more efficient, which will naturally lead to cost and environmental savings.
Bottom line: Efficiency should always be our goal.