Chances are, your IT workloads could be greener.
And while information technology may not be the first thing you think of when the words "climate change" are uttered, technology's role in greenhouse gas emissions is growing. That means IT leaders have some work to do around sustainability.
Here, Matt Buchner, senior director of cloud solution architecture at NTT Data Services, an IT service management company headquartered in Plano, Texas, discusses a handful of green IT considerations and why it's important to think about the environmental costs of technology, even when it is delivered via the cloud.
How can CIOs and IT leaders incorporate environmental sustainability into operations?
Matt Buchner: Ideally, a sustainability initiative starts at the board and C-level. Part of a company's emissions come from its IT workloads, so a plan specific to IT needs to be defined.
The first step is to measure how much CO2 your IT workloads are emitting. That gives you something to work from.
IT needs to develop a goal, a plan that can then be implemented to make IT workloads more sustainable. It does takes work and effort to make IT more sustainable. It's not magic.
So, we need to think about all the benefits you can get from moving toward green IT, and it is more than saving CO2.
For example, sustainability is a way to differentiate yourself from the competition. What we are seeing in the market right now is a great challenge to find talent. You can use a good sustainability plan as a differentiator to attract talent into the organization.
There are many reasons to develop a plan and implement it, but then the question is: How do you actually do this? Migrating to the cloud may be a good first step because cloud providers tend to be more efficient when it comes to energy management and cooling.
If you really want to optimize further, you need to modernize your applications. That takes more time and more effort, but it is proportional to the benefits. For example, one aspect of a modernized application is scalability. Instead of running a set number of servers 24/7 regardless of the load, you can increase and decrease this number, change the server specs as you need. And if you just turn servers on only when you need them -- just like with a light -- then you reduce your emissions.
If you want further gains, it's about making applications smaller. Say you're running one application that has tons of software, is very complex, is always computing, and the CPU is running hot, the fans spinning really fast. If you can create that same application in a much smaller and lighter way, that's much greener. Container, serverless and event-driven architectures have much smaller footprints than traditional applications.
If you're looking at machine learning, there's a whole different set of things to look at. Machine learning can use a lot of compute power to build models. While machine learning is not new, its adoption is now exploding with cloud technologies. There are legacy and modern implementation approaches. You can reuse pretrained models so you don't have to do the computing from scratch every time. You can save a lot of time, energy and money by doing that.
What is a potentially overlooked aspect of green IT goal-setting that CIOs and other IT leaders should keep top of mind?
Buchner: You could say, 'I want to reduce my emissions by 50% within two years.' But if in two years, your business expanded three times, your amount of compute needs increased significantly, and yet you need to decrease your emissions. So, it can be really challenging. A more achievable goal is to tie emissions goals as a proportion to the business size.
The [overarching] goal is to achieve net-zero emissions. But that will take some time. If you look at the commitments from cloud providers, we're looking at five to 10 years to rely on green energy 100%. Additionally, results from the 2021 S&P Global Corporate Sustainability Assessment (CSA) show that most companies haven't even set their initial emissions reduction targets, let alone net-zero goals. The CSA found only 23.7% of those surveyed have set a net-zero emissions target. The first step is to set these goals and make commitments within the organization to reduce emissions.
Sustainability is not just about CO2 but also water use, for example, since so much water is being used to cool data centers. If you look at the sustainability pages from the cloud providers, they discuss emissions but they are also setting goals about water use because data centers are consuming billions of gallons of water.
How can IT teams start to measure their carbon emissions?
Buchner: If you are using [a large cloud provider such as] AWS or Azure, then you can find in the console a dashboard that will show you how much CO2 you're emitting. If you don't have this type of dashboard, then you can do the math to understand how much energy you are consuming.
What about for organizations that are currently on premises?
Buchner: You may know the amount of energy you're consuming in kilowatts because it's part of your bill. And then you can use this to calculate your emissions.
Part of that is knowing whether electricity is generated by windmills, solar farms or nuclear plants -- and in those cases, there will be much lower CO2. If the energy is generated by coal plants or gas plants, there will be more CO2 generated. So, you have to look at the how that energy is produced for the locality.
Using cloud services doesn't eradicate environmental issues -- for example, companies will still have legacy tech e-waste -- and cloud computing comes with other environmental costs. What might be helpful to consider about those issues?
Buchner: [Among its benefits], I believe the cloud has the advantage of shared hardware. But I will say -- in counterpoint to my bias toward the cloud -- that in the cloud, things are so easy to consume. You can just click or come online and boom, you're using now a lot of cloud resources.
In a data center, you can enter the server room, you feel the heat, you hear the fans. Someone who has never walked into a data center can now run large computing servers. Using technology is so abstracted that we can forget what it takes to power it behind the scenes.
That's our job in IT: to educate and make sure people are aware and that they can make good use of the technology that is provided to them. Technology is made so easy that [its energy use] can be abused. But if we are tracking, if we are measuring and educating people about the issues around sustainability and technology, then we're in a better position.
What are some of the ways that IT teams in particular can start being more conscious of their environmental footprint?
Buchner: Seeing how much energy you are already consuming is one thing. The next level of green IT is to estimate the future consumption for a given application and choose accordingly. If you have two similar options performance-wise but one is greener than the other, you can make a decision based on this information.
[If you're using a cloud service], make use of automatic shutdown of servers. Some of the IT cloud services will have an automatic shutdown feature built in. For example, AWS Cloud9 is a service that, by default, goes automatically in hibernation mode after 30 minutes of not being used. If you look at the servers you're running already, there are probably many that aren't actually used. You need to look into this, make sure that servers shut down [when not in use], so configure an intervention or automatic shutdown option. This is not always built in, and you will need to implement it and test it, but it can bring a lot of cost and emissions savings.
When migrating to the cloud, it's important to look closely at the cloud region where your data is being stored. Grid emissions factors are an important part of the cloud carbon footprint. Grid emissions factors list the type of local energy sources that cloud regions use to power its infrastructure, and each cloud region will have different levels of carbon emissions efficiency. Of course, other factors such as data locality, latency and price will also come into play, but choosing the most efficient region will have a significant impact on your overall workload emissions.
This interview has been edited for length and clarity.