Sustainability – Full interview with Mark Butcher | Watts the Future
Mark Butcher works for a technology company and he's very, very good at getting to the truths and some of the mistruths when it comes to sustainability in the IT sector. So, hi, Mark. Could you tell us a little bit about the work your company is doing in and around sustainability? We're very focused in three key areas, which is pretty contextual for today. So it's the world of FinOps, DevOps and the newly developed area of GreenOps. And one of the terms you mentioned was GreenOps.
Explain that a little bit more, if you would. GreenOps effectively comes down to considering sustainability in all areas of IT, looking at operational processes, standards, design, methodologies, everything. So it's operating in IT as efficiently as you can be from the context of sustainability. And it's embedding it within your overall organizational structure.
So how big of a challenge is this for the tech industry? It's pretty giant because the problem with sustainability in general with an IT is, it's a bit like the Wild West right now, and there aren't really many firm and solid standards. There isn't really a consistent way of measuring sustainability or measuring emissions. So you've got this blend - now this lovely hybrid blend in most organizations - of data centers, co-location, cloud services, software as a service. We have to start measuring the impact of digital because people have just realized how big it actually is. Data centers are now consuming vast amounts of power, increasingly as we start to see, an uptake of things like AI. So, is that what GreenOps is all about? Is that kind of steps towards how we kind of acknowledge this and try and be more efficient in the way that we build and run and operate these systems? The very first step is that bit you said earlier, which is actually acknowledging the vast footprint related to IT services because none of us have really thought about it before because it's always been someone else's problem.
So we don't see the huge numbers of servers and storage and network switches - everything that's running the services in the background - we just consume them. People are increasingly aware that actually these big data centers are effectively operating a bit like power stations with the sheer amount of power they're actually consuming. So there'll be people watching this that if they are thinking about power consumption inside data centers or resource usage inside data centers, will be thinking about electricity. But it's more than that, right? There's other resources that we are drawing into these big data centers that are being powered, right? What people haven't realized is these huge gigantic data center facilities that are being built inside the cities, they are sucking so much water, it's unbelievable. This infrastructure generates a lot of heat and that has to be cooled. And in general, they tend to use a lot of water.
And yes, there are technologies that can be used to offset and minimize the amount of water, but a lot of these data centers don't use them. An average medium-sized data center will consume the water equivalent to 30,000 to 50,000 people. And you think of the scale of the much larger ones, the hyperscaler ones, you're talking about entire cities' worth or areas of a country's worth of water being consumed just to drive the requirements for cooling for these gigantic facilities. And so, it's something we will suddenly see becoming much higher on the agenda and that will impact actually where these data centers are built. But actually the big bit that we're missing is what's called the embedded emissions, which are way, way bigger than just those associated with electricity, power and cooling.
Because the embedded emissions is everything that comes from the construction of the infrastructure. So, going all the way back to the raw materials that are used to build this stuff in the first place. A lot of people who will be watching this probably won't really comprehend maybe what the scope of a data center is. I mean, this is a big factory, for want of a better analogy. If you look at some of these data centers now, it used to be a couple of kilowatts of power per rack and now, if you get a big kind of AI on machine learning platform, some of those are easily going to 30, 40, 50 kilowatts per rack.
And the amount of heat and consumption of water required to drive those is absolutely vast. The density of the kit, the power, the GPUs that are involved. There's an awful lot of embedded emissions in those.
What we're talking about there is all the sustainability of the facilities and of the services. We're not talking about the sustainability in them. It's not exactly a great secret that if you go into the average organization, there's an awful lot of waste in the platforms and the services: oversized servers, you've got vast amounts of storage that maybe isn't actually required because it's either data that's never being looked at, is never gonna be looked at again, or wasn't required in the first place.
You've got things that could have been switched off. You've got development platforms that don't need to be running out of hours. Whichever way you look at it, there's a vast amount of actual waste. So it's actually quite easy for us to start looking at ways that we can firstly, measure what we're doing and be accountable for it, but also start remediating it, fixing it and making things more cost-effective and efficient. If you think back, one of the things that we've always focused on was kind of shrinking the footprint, reducing the costs, increasing the efficiency, but it's always been very financially motivated.
But sustainability suddenly seems to have been coming up at the same sort of thing. It's as important, in some cases, more important than actually reducing cost is now increasing sustainability and reducing emissions. Are you seeing the same? I kind of moved my business in a direction of sustainability about two and a half, three years ago.
We were having huge numbers of conversations but people just weren't spending the money in addressing the problem. But since COP26, it's fundamentally shifted. And the key thing is, all these organizations have made net zero claims for deadlines that they're going to meet, but they're now suddenly being asked to be held accountable for achieving those deadlines. It's external investors, it's venture capitalists, it's shareholders agitating.
You've got customers saying, "We're going to make our choice on the brands that we work with or buy from simply on the basis of do they align with our beliefs? Do we want to be seen buying from them and using their products and services?" Employees are selecting who they want to go and work for, so their new employers are based upon, again, how they align with their ethics and their beliefs. Then there's obviously external governance as well. So whichever way you look, there's kind of pressure on them to actually deliver. IT is really unique, I think, personally in that effectively take away the subject of sustainability and replace it with the word "efficiency" and it's sort of the same thing in many ways because you're going to be operating more efficiently.
You'll have shorter design processes, operational processes, you'll have a better use of infrastructure and services, whether it's in the cloud or a data center. So actually, being sustainable could add potentially multiple percentage PBT to the bottom line of particularly a high-tech organization. How do we become more sustainable without slowing down the pace of innovation, though, Mark? If anyone says being sustainable slows down the pace of innovation, they need a big slap around the back of the head. In the nicest possible way because so, take a big bank that's spending hundreds of millions or billions of dollars a year on IT services. If I could show them how being sustainable would release, say, 300-400 million dollars back straight into their bottom line, that money could be used to actually invest in the services, the people and the processes that they don't necessarily have budget for today.
Being sustainable actually isn't fundamentally shifting, it's making sustainable design decisions. It's looking at things like automating standardized processes, setting line of business requirements through to development processes, through to the design standards, operational standards and critically building it into your procurement processes as well, so that you're buying from the right brands in the right way and you're getting your metrics in place. But there is a flip side to that.
The big auditors and the big consulting organizations are all rapidly building sustainability functions so they can carry out audits for organizations. What that means is, you're going to be having a carbon budget as well as a dollar budget within IT services. And what that's going to mean is a total shift in thinking, a total shift in different approach.
Do I put my workload in the U.S. where the carbon intensity is way, way higher so the emissions are higher? Or do I put it in Sweden where they're really, really low? That sounds like an easy decision. But to contrast that, if you're an AWS user, the cost of putting it in Sweden is probably about 25% to 30% higher. And what that means is someone's going to have to be accountable for that decision.
People will be directly held accountable for deciding to be more or less sustainable. So one of the - I'm gonna call it a scourge of this whole sustainability topic - is something called greenwashing, where effectively people position sustainability and the things they're doing positively around sustainability as really more of a marketing effort than anything else. So, is that something that you're starting to really see as well? Partly, it's accidental greenwashing in some places where people don't even realize that it's happening. Some of it is, sadly, a little bit deliberate and coming from the corporate end of things. But a lot of it actually comes from lack of knowledge and lack of experience in the sector as well.
Because if you think how this has to flow down, Scope 1, 2 and 3 emissions... Scope 1 essentially being things like the diesel to power the generators in the data center. Scope 2 being the electricity you consume in the data center.
Scope 3 being everything else. When you're consuming or buying servers or storage or that kind of thing, most of your emissions sit in Scope 3 from an IT perspective. But the problem we have in digital sustainability is the data quality today is not fantastic.
I've actually seen what you guys have been doing in NetApp and what you've done with the release of your carbon reporting, your analysis. Essentially, from what I've read so far, it looks like you've actually done quite a thorough job because you've looked not just at the Scope 1 and 2, but you've also considered your Scope 3. You haven't ignored it. And if I contrast that with what I've seen from a few other technology vendors who have done the polar opposite and they really have just a cursory guesstimate of their Scope 3 and they're not auditing their supply chains and that's when we get into those realms of accidental greenwashing. Scopes 1 and 2 are probably 5% of the actual emissions that are created by a company that sells a lot of goods because you're responsible for those goods throughout their lifecycle. Even when you get into the idea of people misunderstanding the terms, like carbon neutrality and net zero, because defining a net zero goal actually means you have to be net zero for all of the emissions in your supply chain, including your Scope 3.
Whereas carbon neutrality is a little bit more relaxed. If you are going to be aiming for this net zero, the first thing you have to do, which not many digital organizations have done, is to actually start baselining your emissions. So to work out where you are today and start measuring it proactively and just getting something in place where it becomes part of your regular reporting cadence, so that if in a year's time, your business says, "This is now strict governance, we've got compliance rules, we have to get accurate figures," you're in a place to actually do something about it.
Carbon accounting is becoming a specialist area in its own right. So, if I'm buying a big storage array and I want to reduce my embedded emissions or my overall emission from using it, one of the levers that I could pull is if I currently have a refresh cycle of three years, can I reduce my average in-year emissions from an accounting perspective by extending that to five years or seven years? If I do that, what will that impact be on my power consumption? Technology is becoming more and more efficient as we move on, it's becoming more dense, it's requiring less power on a per unit basis. Where and how do they make the decision to renew infrastructure based upon the carbon budget as well as the dollar budget? Should I be deploying power optimization technologies in my data center to reduce the amount of power being consumed by my servers and storage? Should I move from my current on-premises facility, which is quite sort of ugly, aged and inefficient, into a much more efficient colocation facility? If I do that, what will that impact have on the availability of my servers, my applications, my services? Will I have latency issues? Is that gonna accidentally increase emissions elsewhere? And I don't think there'll ever be a perfect solution or a perfect thing, there'll just be decision points that you'll be asked to be accountable for. When you talk about colocation, so let me take that one step further, which is cloud. Now, the cloud providers are doing a good job. Let's kind of lay that out, first of all.
Give us a little bit more of a take on where does the cloud fit in? Is that the answer for this problem? They do a fantastic job of operating facilities as efficiently as they can and they are investing a lot of money in new technologies, renewable energy, so they are doing all the right things. I think they could be moving faster in some areas and there is a, in some places, I do think there's a bit of excessive use of PR because some of the comparisons they do are based on a very aged and very biased comparisons where they're taking the ugliest and most inefficient on-premises facility and comparing it to the most well-designed, well-laid out and most efficient cloud deployment you could ever possibly have in the nicest possible region. Are you saying the vendors do that, Mark? I can't believe that. I know. Who would've thought that there'd be a biased ROI model out there? Never.
Never in a million years. But it does come back to the fact that actually the average hyperscale facility is massively more efficient and effective than the average - whether it's even colocation or an on-premises data center. But that doesn't mean it's exactly the right decision because one of the things that you see in the world of cloud is actually a vast amount of cloud waste. At the lower end, it's about 30%. The average figure I've seen and from my own personal experience is between 40% and 45%. And if you think about cloud spend this year, it's likely to hit $500 billion this year.
If 40% of that is wasted, that's what? $220-230 billion worth of waste out there? With the cloud, the controllers shift completely out to the edge and entire sub-sector's appeared of people managing cost and managing efficiencies. It's easy to find savings in the cloud, but it's actually really hard to convince people to make the savings, to go back into your DevOps teams and effectively tell them their baby's ugly. That goes down, as you would probably imagine, pretty badly and they're very good at finding the million reasons why you're wrong and you don't understand it. The bit that isn't helping is the transparency of the carbon calculations. All the big hyperscalers - well, the top three - have released carbon tracking and reporting tools.
One of the cloud providers is only counting their Scope 1 and 2 emissions. Because from your context as the consumer, all the emissions are your Scope 3, and your Scope 3 is made up of their Scope 1, 2 and 3 in terms of the emissions. The problem you've got there is you're only getting a picture which is between 5% and probably 12% to 15% of your actual total emissions. That's a big figure you're going to be misreporting that is potentially millions of tons of carbon misreported. Then that flows into how they're doing the calculation. I'll pick the top three.
The carbon intensity in real time of the power you're actually consuming. One of them calculates it based upon or gets the figure on an hourly basis, one of them gets it on a monthly basis, and the other one gets it from three months ago. An hour ago is about as close as you can get to reality. You can't really get much closer than that, but three months ago is effectively useless. Sorry to go on about it, but then you've got the next problem, which is their extensive use of averages.
One of them calculates at what you call a region level. So U.S. or Asia Pacific or Europe is an average figure rolling downwards. You cannot, at any point, say that Instance A is more carbon efficient than Instance B because it's based upon an average calculation. So this very much goes back to the whole carbon accounting then.
It's very difficult to do carbon accounting when you're being presented with figures that are at best, averages, at worst, so far out of date that they're not really valid to make decisions or decent decisions around. We are seeing an economic slowdown right now. Is an economic slowdown likely to impact our sustainability efforts? More efficient IT services are more cost-effective IT services. If you can reduce your cloud spend by 30% to 45%, that's a lot of money which you can either return directly to the bottom line or it can be used to cover existing roles within the organization if you don't have budget for the roles and the people, or it can enable you to focus that money on building and delivering new services that your business needs to accelerate the delivery of new things for them to be out there selling these organizations. So actually there are opportunities that exist outside of the bubble of IT for better defining, if fact, you call it your line of business requirements, to do things much more effectively and efficiently, which can massively drive down consumption. I hark back to when we used to have a lot of regulation or when regulations were really starting to take effect.
I remember people saying, "It's hard work. It's not a good thing. It's difficult for us." I think it's the same with sustainability. There's actually a whole bunch of benefits.
So yes, there is upfront effort but actually it'll manifest itself into a more positive outcome afterwards. And the other thing I would say is that what I found fascinating about executing on sustainability. So you go into that engineering development team and you tell them, "You're wasting loads of money."
I can guarantee you, 9 times out of 10, they'll not do anything about it because as far as they're concerned, they're delivering features, products and services that drive the business forward. They don't care. You tell them actually, "Here's an opportunity for you to do something more sustainably that can reduce our emissions," and you give them KPIs and targets that are measured in a way that makes sense to them. They will start doing something about that because they can see the outcome, they can measure it, and it's something that they probably care about personally for their family, for their children and the children yet to come of we want everyone to have a planet. But the very first step of that is coming back to that first point: just start measuring where you are, understanding where you are.
I like the idea that you have to appeal to different people with different sets of values. Do you want to save money? Do you want to save the planet? One of those will be more important to one type of person than they will to another. We've already got an economic crisis and now we're going into an energy crisis. You probably will tell me that that's a good time for people to be looking at sustainability, right? It is such an easy win right now. So actually you're about to be hit by massively spiraling colocation or hosting bills or power bills from within your own facilities and there are so easy wins that you can do to deploy proactive measurement and monitoring technologies that will optimize your power consumption.
A live example from a telco I worked with where they had a retail application where they were streaming out video content in ultra HD, the highest setting they possibly could with full audio encoding because that was what they thought was wanted and needed by their customer base. We got them to actually look at who was consuming it and what they found was: more than 95% of it was being consumed on mobile devices. And also, 98% of them had the audio switched off. And what they've done is they've switched their thinking - their design thinking - into they now push it out with the audio switched off in standard definition. A mid-range storage array, petabyte of storage over five years, uses about 150,000 kilowatt hours of electricity.
So that 150,000 kilowatt hours number, that's just the system. We're talking about 30, 40, 50 thousand pounds worth of electricity just for a storage device. If you were to take that and put it into a location that had a lower PUE, you just saved money. If you were to look at what's in it, we see about 68% of data is never ever used again after it's been created. So you also know that about two-thirds of what you're storing on that device actually has no value whatsoever. It's a ginormous challenge and it's one that's never changed.
The hardest thing with this entire area is getting businesses, organization and people to accept the change that's being made. But sustainability is now a really outstanding angle that you can use to leverage that. Looking backwards at the vast amounts of stuff that you've already created, there's loads of tools, and I actually think the 68% figure is wildly underestimating the amount of redundant data when you factor in the governance, the compliance people setting rules for retention of data. Setting things that we have - the multiple years worth of backups for no good reason, the multiple backup copies that weren't needed, the offsite copies, the disaster recovery copies.
I think there's more value in looking upstream at changing your processes in the future than there is looking backwards too much and trying to fix something that actually you're gonna struggle to get your business to accept the change for. I'm not saying don't do it, I just think get to it at an appropriate time when you've got all the other wins that are bigger. You've got one tweet that you're able to send out, Mark, to kind of summarize what you think people should do. What would be that tweet? Start now. Start now and baseline would probably actually be a better one because you cannot improve what you haven't measured. Look across your entire IT organization and work out exactly all your processes and actually map everything out and get the people involved because your people within your organization will want to help once they understand the impact of their actions, so motivate them to deliver the change.
Because the impact you'll get at the back end will be - not only will you be helping to try and save the planet, but you'll also probably save an awful lot of money in the meantime as well and make your services probably more resilient, more available, and more cost-effective. I've always had this vision of a little plant on the corner of my desktop and the more emails I send during the day and the more data I create during the day, this little plant's starting to die off. Trying to gamify some of these things to bring them to the attention of the wider audience actually is part of the solution, right? It is, and it's keeping the creative spirits alive. It's that organizational transformation, it's the cultural change and it has to go outside of IT as well. Because it has to be all aspects within your business and your organization.
They have to accept the change. But if they can understand from a sustainability perspective the scale and the size of the emissions that would be created as a result of it. Particularly if they start getting something like a carbon budget as well as a dollar budget, you'll suddenly find their requirements will shift. And sometimes, it starts with a movement. DevOps started as a movement and then became a practice that was much better defined.
That's a really interesting point to consider, and one of the things I think we need to be doing as an industry is coming together better to solve this problem in a consistent manner. The data center industry is measuring sustainability differently to how the infrastructure industry is doing it, which is different to how software developers do it, which is different to how cloud providers are doing it. We've covered a lot of different topics here, a lot of kind of short-term things, a lot of medium-term things that we hope will happen and I think they are starting to. We're certainly seeing indications that they are.
But #WattsTheFuture? What is this gonna be? Where are we going to be optimistically if we look out over the next 5 to 10 years, Mark? So if the question in that is from an IT perspective, a digital perspective, if we do things right, we can very easily, comfortably reduce the emissions related to IT by 40%, 50% without even thinking. Yes, it requires effort but it's easily achievable. But what I think is also going to happen, what will drive that happening and enforce it happening, it's going to be a combination of all of that along with actually pressure from consumers and employees as well. Taxation from governments being put in place. It won't just be governance requirements and compliance standards.
It won't just be data center standards that are coming through from different countries and groups. Because you put all that together and it is a monumental amount of pressure. And what it will do is start people thinking about why we're using what we're using? How can we do it better and embed that as a culture.
Mark, thank you ever so much for joining me today. It's been great to have a chat with you. Always good to get your perspective on things and stay safe. Thanks, Matt, great to be here.
2024-02-07 07:54
