 Good afternoon, everybody. Thanks for joining us today after lunch, after a long week of KubeCon. My name is Mary Karocche. I'm a software engineer at D2IQ. Previously I worked at Red Hat, so I have a bit of experience working with Kubernetes and working in tech. And I'm Zinnia Gibson, like Mary. I also am a software engineer over at D2IQ, and I previously was at Red Hat working with OpenShift, specifically in migration. But outside of tech, I'm heavily involved in local community environmental groups, so I've done everything from planting and girdling trees to planting 5Ks to help spread environmental knowledge. I definitely need to ask you what girdling a tree is. But yeah, I'm also really excited to be talking about sustainability and environmental responsibility today. I'm from New York City, but I'm always trying to run away and get outside into nature. So today we're going to start it off with a bit of a poll. Feel free to just shout out your answer on three. But we want to start by having you think about which is the biggest contributor to carbon emissions globally? Is it manufacturing, electricity, transportation, or forestry? So one, two, three. Okay, manufacturing. Manufacturing screamed the loudest to me. It's actually electricity. And if we look into these numbers more deeply, we'll see that manufacturing is definitely up there. But electricity's emissions are about double that of manufacturing. And of course, our industry contributes to both. But while researching for this talk, these ratios really surprised me because while you can see the impacts of the others, like you can visualize them more easily, you can imagine a car emitting exhaust, you can picture a deforested area. But electricity's emissions are occurring far away from our computers and far away from us. So it's not something we think about all the time. And to take a more localized look, these are the numbers for the Netherlands. So the electric impact is actually higher. And we'll talk about this a bit more later. But we're leading with these numbers because our goal today is just to encourage you all to be more mindful and intentional about your choices behind the screen, because they reach a lot further than intuition allows you to believe. So here's our breakdown for today. We're going to dive a bit more deeply into our industry's impact on the environment. And then we have some use cases we're using to visualize how Kubernetes's environmental benefits scale down from large companies all the way to individual contributors. And then we're going to talk about some ways you can get involved. Cindy? Great. So before I get into industry impact, I just want to clarify that throughout this presentation, Mary and I are going to be using terms such as environmental impact, environmental cost, energy efficiency. And what we mean by when we say these numbers is what we're about to get into. But basically just that our industry does have a tangible effect on the environment. So what do I mean by that? Well, the cloud now has a bigger carbon footprint than the entire airline industry. That number was as of 2019. But even after the recent travel boom, the cloud is now growing at a faster rate of carbon emissions than the airline industry. So where is this carbon footprint coming from? Well, about 80% of this footprint is coming from the manufacturing and disposal of the electronics that keep the cloud running. But a quarter of that is just from the electricity that's needed to run the cloud. So where is the cloud running? I'm sure a lot of you know that that is from data centers. But a single data center uses the equivalent energy of 50,000 homes, which was pretty mind blowing to both of us because we don't really think about data centers when we go on the internet. So what is that breakdown? Well, we learned in our research that data centers are among the largest and fastest growing consumers of natural resources. From the water they use for power and cooling to the land they use up and then of course the generated electricity they require. So in terms of numbers, 88% of the energy that data centers use is allocated to cooling and maintaining redundant fail saves. So this isn't even used for running processes. This is just to make sure that the cloud can stay up 24 seven. That means 12% is used for those active computational processes. And then 40% of the electricity usage consumed is for cooling alone. So of that 88%, it's just to make sure that these data centers stay cool. And something we wanted to highlight was I feel like when we think about electric powered things, like when we're talking about like electric vehicles, for example, our instinct is to think that it's automatically a very clean option. But it's important to remember that that electricity still is powered by something, it's still fueled by something. And these are the EU numbers. Globally, the renewable percentage is a lot smaller, but only 17% of that in the EU is coming from a renewable resource like solar or wind. Everything else is fossil fuel based. So like your natural gas, coal, things like that. And this matters. And in the EU, there's been a legal pressure to make these numbers better. And under a new European climate law, EU countries have to cut their greenhouse gas emissions by at least 55%. So that's, you know, more than half by 2030, which is not that far away. It's less than seven years away. And their goal is to be climate neutral by 2050, which is also not not that far away. And the tech industry has a big role to play, like we hold a lot of weight in reaching this goal. So we want to talk a little bit about how our infrastructure choices at work can affect this. So now we're going to dive into some use cases to show how Kubernetes specifically affects our environmental impact, starting from large companies and going into individual contributors. But before we do that, we're going to do something that's a little scary for tech people. We're going to get a little vulnerable. So if everyone could just indulge us and go ahead and close your eyes, it's okay. It's going to be okay. Close your eyes and imagine that you're standing in a field of hyperallergenic grass, of course. And you smell the breeze coming off the trees, and you look before you keep your eyes closed. A path, you follow that path. And you see before you two treasure chests. You go ahead and you open your eyes and you realize that those two treasure chests are our metaphor for two large gaming companies. If you indulged us, thank you. But you can open your eyes now. The bid is over. But yeah, we want you to imagine two large-scale gaming companies. They're creating almost identical products, but one has already adopted Kubernetes, and the other will be moving to Kubernetes from uncontainerized VMs at each stage. So imagining that transition, taking each company in parallel, we're going to look at different stages of their growth, starting from the large and see how they scale down to the individual contributor. So our first case is a large company. You have a huge team that's shipping a new multiplayer game to an existing platform of games, and they're maintaining a large community of users. And in the gaming industry, this would be a AAA company, meaning they have a lot of money, a lot of time, and a lot of resources. So you can imagine the scale of something like this. And while we're not talking about a specific example, just to give you some context, Riot Games, they have a client where they host five multiplayer games. They're definitely AAA status, so you can imagine something of this size. So imagining in parallel are hypothetical companies. Company K is running Kubernetes at this scale, while Company V has made it this far with its VMs. And we're at KubeCon, so you all know the benefits of Kubernetes at scale. But we just wanted to highlight that those benefits are also environmental. Company K is experiencing enhanced DevOps efficiency, which means that their energy usage is lower, their cost is lower, and they're using less resources from those data centers. And I just want to call back to mine because it just blew my mind when I found out how much energy is spent on cooling. So you can imagine if you have infrastructure at this scale that is mismanaged, something that might feel like a small mistake takes a lot of energy to keep cool and online all the time. And another thing, if you're working with those VMs, even just development is a lot more complex. And that's something we wanted to highlight as well, as it's not just the experience of having things up and running, but also the experience of developing those things that have an environmental impact. If we scale down, instead of our multiple, our platform of multiple games, if we have just one game, still maintaining a large community of users, maybe it's a case of an indie acquisition where a company of the scale is being acquired by a giant like Riot, which happened a few years ago to Hypixel Studios, and this is their one flagship game. We're still seeing a lot of the same benefits that we saw at the larger scale. But the thing we wanted to highlight here was the training and migration. So if Company K was already running Kubernetes as they were merging into this larger infrastructure, they have a lot less training to think about, they have a lot less migration worries than you would if you were like Company V experiencing more infrastructure costs and a lot of effort to train people to migrate over to Kubernetes. Let's go smaller. So now let's look at this indie game before they're thinking about acquisition. So again, they're a small team of developers, they're shipping a multiplayer game, just beginning to think about scaling in terms of user base. So this is a real world case that applies to Kubernetes. Maybe you saw Dominic Green's talk at a previous KubeCon where he spoke on Netspeak Games and their migration to Kubernetes. So the main things that we want to take away from this example is that there was a single developer doing this migration and he was able to do so within the game delivery timeframe of six months. So it was a quick time. And the reason why this was so quick was because he used open source libraries, which we'll get into a little bit later. And as just like a little nugget, he was able to containerize Unreal Engine, which if you've done any game development, you might understand how unreal that is. So now we're looking again at Company K and Company V. Company K at this point, they're a small company. There's a lot of complexity to Kubernetes. Otherwise, you know, this conference wouldn't be happening. So with that, with a small team, there's going to be some more risks. But alongside, they have more scalability and automation benefits. So as they're thinking about, you know, a larger user base, they want to think about enabling cluster auto scaling, but then also event driven auto scaling, which is going to save a lot for the user experience standpoint, and then also for the environmental cost. And then again, you know, there's resource efficiency with using Kubernetes. We see this even at the smaller scale. Whereas Company V, again, using uncontainerized VMs, there's a lot of difficulty with managing VMs, especially when you're talking about a larger user base. And the key thing here that we wanted to point out is that there's non ephemeral resources for testing. Whereas Kubernetes, you can scale something up, test it, scale it down. So you're not running all those resources. It's likely that Company V is having a testing platform running all the time, which costs money and costs in terms of environmental. So again, cost considerations with that one. Now, if we take a step back and say, this talk inspires you tonight to go make your own game. But unfortunately, you don't have any friends just in this example. So you're a single developer shipping a single player game. The goal is to ship obviously, not yet to scale. So a real example, again, using Kubernetes is CubeDoom. And I encourage everyone to go check this out after this talk. It's a real open source testing game. We're not sponsored. It's just fun. But you can test your own Kubernetes clusters resiliency by killing pods in a first person shooter game. Yeah. It's just as fun as it sounds. So now we're looking at person K, person V. The big things to consider here again is that there's a huge learning curve when it comes to Kubernetes for the first time. I'm sure we've all experienced this. Maybe you're experiencing this at this conference. But with that, there's a lot of resource requirements that Kubernetes has. And if you're just a single person using this game for fun, maybe that's more engineering for the infrastructure than you need to. But if you're looking to the future, if you're trying to make friends at this conference, and you think eventually you might want to make this a multiplayer game, then Kubernetes is great for scaling. And in the future, this is going to save you a lot of time and effort. Whereas person V, there's a smaller learning curve to start with these uncontainerized VMs. So there's a quicker setup. But again, there's non-ephemeral resources for testing. So depending on how into this game you want to get, that may or may not be a con. So what do these use cases all mean? What do we want you to take away? The benefits of Kubernetes scale down to a point. So at a large scale, obviously you're looking at multiple data centers. When you have such resource efficiency, this is going to make a greater difference between company K and company V. Whereas as a single person, you might be over engineering and wasting resources. Migrating early can positively affect long-term impact. We saw this with Netspeak Games. They were able to have a quick migration with one single developer. Whereas a larger company, you know, that's going to take more time and effort to migrate and therefore more environmental cost. And what we're about to get into is that Kubernetes is not the perfect solution for environmentally conscious coding. It's great. But as a company or as individuals, there's more steps that we can take to be environmentally conscious coders. Yeah. So and that said, the most important thing we can do as coders, Kubernetes enthusiasts, anyone who's here is to get involved. You don't have to stand up. But if you want to take a second to stretch, maybe get the crick out of your neck. We wanted to invite you to do so. Hell yeah. Thank you. Brave people. We love it. It's been a long week. Yeah. If you want to. Thanks. So yeah, we're going to break it up into three different ways you can get involved. Talking about some upstream projects that were cool. We wanted to highlight offsetting emissions, because it's something we hear a lot about and we wanted to dive into the impact it actually has, as well as some development practices. Great. So we've mentioned this a few times already. I'm sure you contribute to upstream projects already, but if not, we just wanted to highlight some that we found along the way. So in terms of developing, green frame is a great way to get started with environmentally conscious coding. Basically, it measures and reduces your website CO2 emissions by detecting carbon leaks. For testing, I'm sure a lot of you use Prometheus. We do over at D2IQ, but it's just better than trying to reinvent the wheel with creating your own testing. And in terms of hosting, maybe you saw the Agones talk before this, before lunch, but it's a open source way to host and run dedicated game servers on Kubernetes. So very cool. And in terms of community groups, CNCF Environmental Sustainability Working Group is something that Mary and I are both a part of. And if you are getting interested in environmental awareness, which if you're at this talk, I'm assuming you are, it's a great thing to get involved with. And they have bi-weekly meetings. Now next, we've hit this a couple of times during this talk, but in terms of offsetting emissions, at this point of the talk, you're probably wondering, what the heck is this? Who the heck is this for? And is there any hecking ESG benefit for all of you business people out there? So what is it? It's something that gets talked about a lot, but it's a reduction or removal of emissions of carbon dioxide or other greenhouse gases in order to compensate for emissions made elsewhere. So there are ways to channel funds to conservation and sustainability development while reducing climate emissions if they're done well. But if they're done poorly, they're just a really good marketing campaign. So who the heck is this for? Obviously, any companies that want to get involved, there's four different ways that you can offset emissions. Two of them work really well and are the most cost effective ways. So renewable energy and energy efficiency improvements. So think switching light bulbs for LED light bulbs. These are the most cost effective, but the two that are used a lot to varying success are carbon sequestration and aviation offset. So what is carbon sequestration? It's basically the planting trees campaigns that I'm sure we've all seen. The issue with this is that these take a really long time and a lot of companies don't plant enough trees to offset their amount of carbon emissions. Now don't get me wrong, I love planting a tree, but it's just good to be informed about it. So is there any ESG benefit? Well, if you've read any ESG reports that companies have widely available, they mentioned offsetting emissions quite a lot. And it can be a great cost effective way to begin sustainability movements. And a lot of hyperscale data centers like those we see at Google, Amazon, Facebook are pledging to transition those sites to carbon neutral via carbon offsetting. So I encourage you to explore their ESG reports if you are interested. And then finally, we wanted to also highlight some better development practices. And there was a great talk earlier by Christina of Admin Control that I wanted to shout out if you want more of a deep dive on technical things you can do once those recordings are out. But to give you our short version, we wanted to emphasize starting small, both when you're learning, when you're adopting Kubernetes for the first time, and also testing small. So starting with as small a cluster as you can, seeing if things pass there and then scaling up, not starting, you know, at the biggest capacity first. And then we also wanted to remind you to not reinvent the wheel. There's so many awesome upstream projects like the ones we mentioned that can be a great way to save your own time and your resources and to build on top of something that already exists. And I also wanted to highlight that tools like kubekos, Prometheus, even if they're intended for business savings, if you have a cost saving, you're probably also having some, you know, savings in your carbon emissions. I wanted to highlight those. So our takeaways for this talk, tech companies have a tangible environmental impact, even if it's not something that we see every day. And while Kubernetes is not a complete solution to environmentally conscious coding, it is an awesome tool. And if we implement things like cluster auto scaling, event driven scaling, it can help us waste less, waste less resources. And those benefits do scale down to a point. So, you know, if you're an individual working on a project for the first time, it might be overkill to do everything with Kubernetes. But if you're planning on scaling in the future, it will make that migration a lot easier. And that also has a huge emission saving. Yeah, so we have linked a lot of really valuable resources that we couldn't really dive into because of the limited time. But these slides are available on, under our talk on the website, everything from offsetting environmental impact, and better developing practices. And then we also have, you know, some academic papers that we've linked the research to everything about environmentally conscious coding or environmentally conscious gaming, whatever you're into. And we love feedback. We're pretty sure this QR code works. So go ahead and leave us some feedback. Find us on LinkedIn, send us an email if you aren't able, if you think of a question later to ask us. But with that, if there's any questions, we are willing and able. Someone's running. Hello. Yeah. I just wanted to ask how would a developer or an organization measure their impact? Because right now, scaling up and down might be okay. Okay, well, if I divide my workload in half by less pods, then maybe 50% energy saved. But there is probably other ways which are a little bit more ambiguous. I'm not sure. Am I actually saving energy if I do this? Right. There are a couple of really good tools. We have some of them linked. I know Microsoft has a carbon measuring tool for Azure and then Green Frame, which we mentioned earlier for specifically for websites. But this is something that Mary and I are interested in diving into more of developing how individuals can really deep dive research into their carbon offsetting, because it's a little ambiguous as is of where your electricity is coming from. It's depending on which data center you choose, which region, because different data centers have different environmental impacts just depending on their country. Yeah. That's a great question. Thank you. Hi. One of the areas that I'm really interested is reducing workflows specifically around testing, so not standing up an entire suite of applications. And I saw a talk on telepresence and it looked really interesting for personal intercepts so that you'll need to send traffic to one pod. Anyways, just curious if you had any thoughts on that or maybe just leave it there as a comment. Yeah. I'm not familiar with telepresence, but that sounds like something to look into, for sure. Definitely. Yeah. Definitely in the theme of starting small and getting bigger. Yeah. Thanks for the talk. I was wondering about technological choices. For example, we still mostly, most of us still use Intel or AMD servers and there is a rise of IRM servers and it can create a lot of, like, serve the same processing power for less energy, but of course, if you have to replace all the servers, there is a cost-benefit analysis because it's a lot of energy investment. But it can be probably worth looking into, so I don't know if you have thought on that. And in a related note, there is another technology that's rising right now, which is WebAssembly. And it does, like, it can be convenient, but I think we might want to look into the energy impact of that because it's probably less efficient. So, yeah, if you had any thoughts on this technical aspect. Yeah. That's a great thing to point out. I think specifically with switching servers, which you mentioned at the beginning, the big thing with the carbon footprint of the cloud is disposing of old electronics. So what I hope to see in the future is more emphasis on recycling those electronics. I think there's a few different programs, depending on which country you're in that does that. But we're from the States and it's pretty difficult to recycle. But in terms of, sorry, I lost my train of thought. Mary, do you have any? Yeah, something I just wanted to add was that it's hard to find a perfect answer to any environmental or sustainability question because there's just so many moving pieces to the whole conversation. So I'm of the mindset of, you know, we can't do everything personally that the world needs, but any good we can do still helps. Any other questions? We'll also be here after if you want to chit chat, learn what girdling a tree is. All right, thank you.