 Alright, thank you everybody for coming to this session at this latest afternoon right before the party. And it's already our pleasure to miss you all, and I want to share with you the journey, the visions and the experience we have for this environmental sustainability, one of the latest CNCF tech. And with my pleasure, I want to introduce you to the co-chair of the tech model. Hi, and this is Wamin Chen. So I'm a cloud software architect, and Wamin Chen is a senior software engineer over at Red Hat. Oops, we don't have notes. We got a... Give me a second. One second. Hi, welcome to the first talk. We'll have some technical difficulties by the environmental sustainability tag. We're here because sustainability is becoming increasingly important, whether you are about carbon or merely trying to reduce the cost of running a data center in the cloud, in these odd times, power is becoming increasingly expensive. And you're in the correct place. I'd like to introduce... I am one of the three chairs for the environmental sustainability tag. Max and Leo are also the other two co-chairs. They are not here. And there's also Erin Boyd over there. She's our tag liaison to the talk. And with all my pleasure, I want to... Because during this journey of creating this tag, we have been meeting with different volunteers from different companies around the world. And I saw my pleasure to introduce the folks here, and Tony, Ariel, Kara, Jochen, William, Laura, Nikki, Scott, Chen, and Peru, who has been with us for a long time. Now, we have been discussing with these community contributors on different agendas, share with them the passions and share with them the visions. And so they have been contributing projects, ideas, and road maps to this tag. So we have been working with the proposals on multiple occasions, having multiple discussions. And this is not easy, so to speak, because we have to miss out different time zones. People are working out of their working hours. This has been a tremendous undertaking for many of us. So, yeah. So let's talk about some definitions. Is there... Let me give you a sec. So let's talk about some definitions. So when we start talking about scope, that's where we're talking about where your carbon emissions come from. So scope one is sources controlled or owned by the organization. Scope two is caused by production of energy purchased and used by the organization. That's the energy you're paying for to run your hardware or your cooling. And then scope three is caused by providers of services or products consumed by the organization. Scope three really isn't in scope for this, because we're trying to aim for cloud-native technologies. And so what vehicle you use to drive to your data center isn't really within our scope. Yeah. So that's also to share with you certain aspects of the cloud-native and carbon emission reduction. So one of the things you are thinking about is where the carbon come from. The carbon actually has been measured in many multiple different ways. If you are looking across the continents, North African continents have same amounts of carbon intensity. And over time, you will see these carbon intensity changes with the introduction of new wind power, hydro power, and nuclear generators. And the general response, if you are going to the sources we listed here, the general response is more and more continents and countries are reducing the carbon intensity, but they're still unbalanced. The other aspects of the carbon intensity is that they actually vary between the hours of the day. If you are using your electricity at night, you may have less power in intensity than during the day, which is understandable because not all the generators come on at the same time. And not all the power source are available, solar, and wind power. They may not have the same capacity during different hours. So the other source of powers have to come online to make up the difference. And I also want to share with the other discoveries, and we also, over the times, not all the workloads and not all the settings in the data centers are the same. What makes the differences between the data centers and the workloads is that do you have the virtualization or not? And what kind of workloads you are running? Are you running like the state-less, state-form or service workloads? This will make a huge difference in power consumption in general. And if you are going to the source of this study, it's based on a survey and measurements of a number of different organizations. The data center size, if you have small data centers, we'll come back to that later. The power efficiency in terms of power utilization effectiveness is different. And the workloads, if you're running, for example, the database, running for a long time, using a lot of CPU, a lot of memory, this type of workloads will consume a lot of energy. Consequently, the power footprint is also bigger. If you go to public and private clouds, the differences will be even bigger, as we are going to see in the later slides. So one of the top priority for us is to understand all these things. How do we get these things? We use the metrics. The metrics, we are also, we are talking about different levels of metrics. One is the power at the grid level. These power, the carbon intensity can be measured in different ways. One is how much fossil fuel used in the power generation. And what's the, even that is not good enough because the fossil fuels consist of a fossil, coal, or just a natural gas. They have different carbon intensity. The carbon intensity metrics, which are provided by many of the grid electricity generators, they are closer metrics. We all want to collect, just based on the information they can provide to us, to imagine how much carbon used per kilowatt per hour. The ranges varies across the continents, as we see. It's something anywhere between 500 to 700 kilowatts hour. Data centers, they could be having different metrics. And all the way if you come to the servers and it's going to be just another set of metrics. So then let's talk about the various ways that at least I break down the problem and other people have different ways of breaking down the problem as far as resource management. That's what I'm talking about. So the first piece is insure cluster. So which cluster do you want to schedule to you according to your metrics? So if it's during the daytime, maybe you want your data center in a solar power region. If it's at night and you happen to have wind in a storm, maybe you want to schedule to wind power. The second place is where do you place your workloads? So in truck cluster. So if you have everyone's seen this heterogeneous type of system where you have different types of CPUs, you have mixed CPUs, you have an area with an accelerator fabric, and then you want to land your workloads accordingly. So you might put your microservices on that first node which has had a particular type of CPU, your mixed workloads on your center one which has both efficiency cores and performance cores and your AI workload over on the one with attached the accelerator fabric. Similarly, when you're looking at intra cluster scheduling, if you have CPUs that are running hot because 30 to 50% of your energy within your data center goes to cooling, then maybe you don't want to schedule to that first node. Maybe you want to schedule to that center node and try to keep your hotspots to a minimum. The last area that I usually divide things into is node level. So if you currently, you know, you have a power config and you have idle CPUs and working CPUs and they're under the same power config, what you would rather have is have it to your idle CPUs or in a minimum power config, your first pod which has a performance job. You want a power performance. Maybe your last one, you want a power savings config because maybe you're not planning to use that much energy for those. Yeah, so I've come to cooling. Cooling is a big part of the energy consumption data centers and you see the, this is the data dating back five years ago. But you see general responses as you go big and the data center goes bigger and bigger, you go into the hyperscale. The cooling components will be much, much less. And that's why, you know, even nowadays we're kind of talking about 1.1 and even close to one. And even there's a people talking about an active carbon footprint. How could that happen? It depends on two things. What's the ambient temperature you want to set? Usually data center recommendation is that you can set as 18, anywhere between 18 to 75. South east, but depends, that's where how much it costs. The higher the temperature, the more likely they are going to be overheated and that's where I come to the life cycle of the hardware. You don't want to overheat your hardware. The second is that depending on the techniques you are used for heating, that you're using an air heating or just a liquid heating, they have different impacts or efficiencies and as well as the energy consumption of the heating system. The other way is the very last one is do you use a smart control? One of the best practice from Google data center dated a few years ago by DPAI is if you are taking all these inputs from IoT sensors you can reduce the data center cooling by as much as 40%. That's a huge savings. So about halfway. So does anyone have any questions before we launch into the charter and all the technical details as far as the charter specific items? All right, so we're going to go through the charter. This took us, this was a community effort. It took a lot of input within the community. Huamin was kind enough to run some of those meetings. I ran some of those meetings just to get that through. So this is what we came up with. So we want to identify, define and develop tooling to assess and improve environmental sustainability approaches including quantify the energy consumption of cloud native implementations individually as well as in common integration patterns. We don't have stuff like this yet. Recommendations and strategies to develop, package, distribute, deploy and operate cloud native implementations to reduce energy consumption. I won't read the rest to you, it's all the same bucket. Capabilities, benchmarks and processes to evaluate technological and architectural health of projects. There's already some work for this in the Green Software Foundation as far as some amount of measurements but we need stuff that's specific to cloud. Community outreach and engagement on the work of this tag. So you'll see we'll start doing more engagement. We just formed so we're looking for people to like yourselves to come in and get engaged and collaboration with other environmental or sustainability organizations, initiatives, activities that may fall outside of the CNCF. Next piece is out of scope. We're not forming an umbrella organization beyond the CNCF. The scope is huge just within cloud native. We do not have the time at least within all of our workdays in order to get all this done. We don't wanna establish a compliance and standards body beyond the cloud native space. There are people already working in that area. Let's leave it to them. But we will collaborate with them. We won't evaluate individual company infrastructures. That's probably a good business opportunity for someone who wants to go do that. And we don't wanna focus outside of cloud native technologies. We have a set of deliverables. The first one as a tag is a landscape for carbon and energy efficiency. We've already started this document if anyone wants to engage. We have a QR code so you can come in and if, or you can contact me and I can get you into the HackMD where we're trying to collaborate quicker but GitHub is fine as well. We want to do recommendations and optimizations to new and existing projects within the landscape so we'll help you as we can. And to use a community expertise. We wish to do reports on gaps in the environmental sustainability coverage of the landscape. All these tools are fairly new. There's nothing that's very well established at this point that I know of. And we've been looking for a while. We've used inputs and recommendations for proposed projects for CNCF hosting and advancement and suggestions for improvements for CNCF internal processes. For example, education for sustainability. So here's a QR code to the landscape docs. So now you can get out your phones and pull in information. We are looking for active contributors. As I said, you can reach out to me directly. I'm on CNCF Slack. I'm on Kubernetes Slack. Anytime, Buamina is also and is also very helpful. So if you're in, you know, if you're in Europe and... Oh no, you're here. We have some, we also have the Europeans. The other two tag chairs, they're in Europe. So if you're from Europe and you want quicker contact with them, go ahead and contact them. There's also, we're also gonna highlight a few projects that are currently going on in cloud native space. None of these are official CNCF projects at this point. So I'll let Wamin do this one because this is his... Yeah, so we do only illustrates how we are gonna contribute that can happen. So we want people to donate projects, donate technologies, donate to contribute to the regions and the directions. One of the best ways, in my opinion, because of CNCF is all about innovation and incubation. So we make the projects happen first. The projects here are listed may not necessarily be limited to what we are doing and the space where it's still exploration. So Kepler is a project towards that one of the direction we just met. How do we create metrics? How do we collect metrics? And how can we use the metrics to optimize energy consumption in data centers? So this is one of the projects we are gonna have a more opportunity to present to you in the coming days. And along that slide, some of the CNCF projects has already been working towards the sustainability angles. One of the things you have already, maybe you already attended to the talk provided by Nikki, is how do we make the CSCD pipelines greener using flux and using some of the metrics we can collect from projects like Kepler to present to you the energy consumption and you can use the heuristics to optimize and fine tune your workload deployments or cluster deployments. And yes, another one is the collaboration. One of the examples we can use for collaboration is that we work with pre-existing communities and we want to identify the use cases in those projects. How can we incorporate some of the ideas for even carbon reduction and energy consumption to improve the sustainability impacts of those compute workloads of the framework? So that's this framework and this workload will be more greener in the future. So additionally, we have an end user sustainability surveys. So we're trying to take information from what end users are doing as far as sustainability and what their needs are so that we can make sure that the community is targeting the people that are gonna use the technologies. So if you have time to go and do the survey, we really would appreciate it. Additionally, there will be a group of us meeting outside of room 620 from 1030 to noon tomorrow directly after the keynotes. It's all the way at the end of the hallway, beyond the keynote room. And so if you wanna talk to us and have discussions and get more information, you're also welcome to go there. And here's a link to our Twitter page. We have a Twitter, it does not yet have a icon, so we're still waiting for a logo from the CNCF. They've been really helpful in getting us started. We did just start directly before this, so asking for a logo the week before wasn't a reasonable ask, but we'll get to it when we get to it. Do we have any questions? Any suggestions? Any volunteers? So the question is also how we are at the local volunteers, how you're gonna be able to contribute and organize events, something like that along that line. I will take one of the answers and then maybe Maro can share your thoughts after that. So me, just illustrated several of the projects. I personally participate in these projects myself. I think it's all about the passion and the contribution. So you will definitely find the same passion and the contribution from that side of the project or ecosystem or community. And the idea is how we can work together to make things happen. So if you have the ideas and you understand how the impacts will work out in your organization, in your projects, in your community, feel free to do so and feel free to help us to make that like a case study for us so we can go, that's templates in our other engagements as well. We've also had some discussions about starting a podcast so we can start doing like a monthly highlight or getting that Twitter more active so we can start twittering out sustainability projects. So if you have some ideas of things you wanna do, please show up to the meeting. We're glad to have you there. We're glad to intake. We're still brand new, so we're still figuring out what our structure is. So we're happy to enable however we can. It's to find members but also start working on those deliverables. So we already have some places to put effort. So we have the landscape doc that I referenced. There's also a glossary. So we're trying to get a glossary of terms so we make sure we're talking about the same things across the board. So if someone says, you know, scope one, we all know that we're talking about scope one, right? So there's already places to land today. There's already some projects in motion. So if you come and have some specific asks, we're happy to help you but if you have other ideas of things that we need to be doing, that's also there. So it's not just get members, it's also we already have some projects in motion. And I would have continued as a line just like if you're seeing other CNCF practice but the tech practices and they also have creating these projects and based on projects, there are certain ways you can contribute to, like for example, the tech security, they have certain projects listed and there was going to be members over there as well. So yeah, that's going to be a certain ways you can create model and the rest of the chairs can probably think about the charter and the organization how that's going to be done. Yeah, I mean, I expect that some working groups will come out of this for particular topics as well, which is part of the point of the tag is to support those working groups. Thank you for attending. Yeah, thank you. It's good talking to you. Thank you.