 Hello everyone. Welcome to my session. My name is Max Kavecha and I hope you had so far a very great open source Summit. I'm very sorry that I can't be with you today in Dublin But I hope we will connect somewhere soon on the platform on LinkedIn or Twitter and I'm very happy to answer your questions also later So my session will be about seating and environmental sustainability into the cloud native community But first, who am I? Well My day job is actually about Q&A consulting and the cloud native advisor itself means having a look where the customer are moving to a more cloud agnostic approach with open-source software most of the time based on Q&A is obviously and Help them with their digital transformation and getting more cloud-ready cloud native And yeah, so I'm actually an enterprise architect by heart This is my profession from the last years, but around 2016 the end of 2016 I've somehow found Q&A this was in a nice project and Since then it somehow didn't release me therefore also contributed two and a half years to Q&A's release team itself and To all the companies related projects as well Initiated with the CNCF the environmental sustainability working group, which we're talking partially about today so the first great news is Depending on when you will watch this video Either we're still a working group or we're already a technical advisory group depending on the results If we will be a technical advisory group We'll be released later. So I cannot say it yet, but As this is recorded Depending on when you will see it. Yeah, I said one or the other so Before we can go deep into this talk I first need to give you some terminology because maybe some of you are experts in this field Maybe some of you are not and I want to ensure that everyone can follow along When I'm talking about emissions Then I mean every kind of greenhouse gases. Yes, this is another terminology but these are all gases which are Caused by human beings by burning fossil fuels or which I actually can be also naturally appear by Biological processes through animals through Deep freezing of the ground and releasing the thumb and so on Emissions can also include particles micro particles For example when you're driving car and all the burnt fossil fuel particles flying around So greenhouse gases therefore also from the way of how it is named and actually anticipate that It is heating up whatever is inside. How does it work? Well? By burning for example coal or gas Most of the time will really will release carbon dioxide to the atmosphere, but also other gases like methane and so on and They're invisible, but they're like a blanket around the earth and The sunlight which is traveling towards the earth. It's normally partially reflected from the surface and go back to space now The greenhouse gases are very good in catching the infrared light And some other lights, but mainly infrared light and throw it back to the earth So what happened is that everything what makes the Earth's earth surface hot is Coming back to us again, and that's why we have a heating a global warming If I'm talking about carbon footprint, this means like Everything which you can calculate together Which is related to either a person or organization can be a software About the carbon dioxide in total We just yeah embody it to this and Therefore we are directly in the next terminology embodied emissions are basically All kind of emissions which are released to the earth For producing something this can be a class of water It can be a headset a watch chair. You're sitting on And everything of this needs energy in the energy is most likely somewhere produced by burning gas or oil or Coal or whatsoever And so you see a lot of these terminologies are somehow Bill on each other, but have slightly different meanings And when you're really deep into this topic or you want to come close to this topic then you will also find something called scopes and Scopes are actually defining whether you have a direct indirect or very indirect way of Causing this carbon emissions. So a scope one means that if you drove with the car To Dublin for example, then this is a scope one emission because you have a direct impact on causing this Scope two is what for most people working with software with cloud providers is very relevant because This is where we have an actual impact on How well you have for example your application running on a cloud provider and the club provider somehow needs energy So the energy Obviously somewhere produced release CO2 But this also means that your service is causing this energy consumption and therefore also release of CO2 So you have a kind of indirect Influence on it, but you still could influence it and this is what we are going to talk about today most of the time and last but not least there's scope 3 emissions and scope 3 is Actually everything what is in your production line But you do not have a that direct influence on it so if you order a server rack full of servers and this and whatsoever Then all of the single components requires energy and they are getting delivered from some other Manufacturers who also have produced it and it's to transport it and you have people who needs to go to the work and so on and so forth and Actually all of these emissions needs to be put and calculated together for the scope 3 emissions Which are then embody it into the server or the server rack which you have in your data center And there you can see this can be a rapid hole and can be very difficult, but we will come to this later So for scope 2 what is actually the problem? Well, the global data centers are utilizing at the moment around 2% of the whole energy worldwide This is comparable and even more than some smaller minor countries And it's expected that was in the next years. This will crawl by additional 2% The most craziest thing is that there will be a peak And it most likely will not stop there, but this is the direction which some forecasts thinking of To reach around 12% of consumed energy globally, I'm sorry that writing mistake by 2040 Which is a huge number this is bigger than countries like Germany and Japan and France for example together, but how does this happen? Well We produce every day more and more data and we distribute this whole thing IOT is not just the buzzword. It's really happening But what I it is actually doing is producing data and send it back first And then we have the digitalization itself also for countries who are just stepping in to digitalization They need more servers more compute capacities Well, you see where it's going. It's getting more and more and more and it's not expected to getting less Plus we have obviously some old systems. They're difficult to optimize their needs old hardware because it cannot be replaced and They're like the foundation line of all of these Causes for for releasing carbon emissions and Carbon's emissions are actually everywhere. This is something which we need to be aware of And this is also outside of this room and outside of this topic from open source and cloud native Everything causes in some way or the other carbon And the pollution of the environment and this is what we have to work on so For me, it's important what we can do in my professional day life as an IT consultant I can have all the companies to be aware of that Even their software which they cannot physically touch are causing The the release of carbon emissions to our air and therefore making this planet more hot And this is something which we have to change. So That's what we are going to talk today partially about and what we are thinking what we can do from a cloud native perspective to get a little bit better So as a working group our goal is to advocate for and to develop and to support and to help to evaluate environmental sustainability initiatives, so everything what in the cloud native technology landscape is Somehow trying to make the things better Is where we want to go into and see how we can help them And what we want to help to reach is a net zero goal This means that in the end Ideally the software Doesn't cause any more any kind of carbon emissions As you have seen this can be very complicated, especially if you think about scope three So we also need to think about to simplify processes and make the life easier to really identify this You have a couple of main activities defined. This is nothing which is nailed down This can change, but this is the first step The first things we have in mind and what we can do So on the one hand side, we want to identify and define and also developing tools or help to develop tools To grow into this direction To get better We have one major issue We will need to quantify where energy is consumed from applications and be very precise to identify Okay, this component actually utilizes more. Well, this one it actually looks big, but it doesn't do so much and so on we need to develop recommendations to reduce energy and To get better with the software efficiency For this we need the community. We need to reach out to the community We need to partner with them. We need to help them to understand and to find first points where they can Work on getting better with it But we do not have to reinvent the whole wheel. There are a lot of foundations out there Since many years I need to say That are working on Certain kind of topics also in this direction like the green software foundation, but we'll come to them later but we need to collaborate with them and this is where we see the community this working group attack as A door opener as someone who can help to go Easily to the cloud native community and yeah Show best practices and explain how to be very efficient with writing software And therefore we also need to be able to evaluate How is the architectural house of the application of the software Of the projects in the cloud native landscape and where we can start doing the changes and implementing changes So I call this better together Because we have the green software foundation and the name anticipated. It's everything about software We have an open compute project initialized by Facebook, which is bought hardware and data center optimization and then we have the OC which are working on Helping actually the economy to switch their capital workflows from Climate harming investments to climate changing investments So because the long text is a little bit too complicated to read I try to put it more in some clear bullet points the Queens office on station is doing an awesome awesome job and The target is to build a culture And to bring this culture of building sustainable software into software development teams and make this a core priority and For example the software carbon intensity specification is one of these little things It's not that little there's a lot of brainpower go into But it's this little things which helps the team to Actually measure how good they are and if they can get better too Within the open compute project we see awesome initiatives like heat reuse it because if we do not Need to heat again because we can use it already This means that we reduce actually the amount of cost carbon emissions and There's also the thinking about future technologies so if you build specific chips for artificial intelligence and Deep neural networks and so on who are very efficient exactly for this use case Then they are also being better to be utilized rather than a general purpose compute instance, which needs tons of energy and I said you have the OSE who was thinking about to implement a data pool and software platform to empower sustainable solutions so how do I see this well in the middle we have For me a typical stack how I can find it nowadays nearly everywhere and For sure quite often the data center for me is not visible anymore, but everything above that is What you can see day by day? If you start on the top left corner, we have the green software foundation and how they can help actually to optimize and Teach coding best practices for the software, which is running for example in a container. It doesn't needs to run in a container obviously, but Well, I'm the kubernetes guy. So need a summary container at least The OSE is to about to empower the change of the use case and to focus more on Environmental friendly and more sustainable solutions rather than to environmental Harming solutions and they want to do this by a data-driven AI platform On the bottom you can see the open compute project, which is looking into optimizing data centers and hardware and so on well Obviously there's something missing and this is where the environment sustainability work in group comes in because the cloud native Environment and the universe of the covenetus on the round covenetus has such a tremendous impact it's something which we cannot skip in this stack and There's a few things which you can easily do like optimize container images and the build processes around it But also about scaling and scheduling and there's tons of little best practices What we can do already today? And which can help but we have still a long way to go Nevertheless as you can see When we put this foundations and this is just a selection of foundations that want to be Very careful here and there are a couple of other Very cool groups and people who are since years working on ideas how to how to get better here But I think these are the for us at the moment most seen ones or which fits into this explanation How we can think about this whole stack and how we can really optimize it from the first kilogram of bet on touching your crown building a data center up to What is your use case and what are you going to sell as a product and how? Yeah, in which way it is thought through as a product I personally believe that we can be only successful in this change if we are working together This means that we have to solve the different dimensions within the computing environments and If you just tackle one single piece of it This is like a drop in the ocean but if you work on all of these pieces and Get also there in a good thing Then this is like a river into the ocean at least So it will be even more important also in the future as you have heard in the introduction Everything is growing. We get more compute powers around the world. It's going through the edge It's going to be more data. We have to Digitalize more and more countries and so on so forth These scale we have to catch We cannot reduce it and I also personally believe it would be done to stop it because to connect the world also means to free and open the world to Allow everyone to have the same rights of getting information and the same rights to Implement the businesses and build something for them But we need to get also better and I think this is the point where we together can start and optimizing how We do it how we do software how we do infrastructure so cloud native What could Kubernetes and cloud native actually do? Well First and foremost, it's very important to understand that Kubernetes itself. It's not a hypervisor. It's not the infrastructure or something else From my perspective working very long time with Kubernetes. This is an extensible platform It is a powerful API Which actually can run nearly everywhere. It doesn't matter. You can run to an edge on a global scale well, there's no really limitations into it and it is very good at managing the workload and Handle it either to keep it alive, but also to scale it up and down And it is very good and integrated meanwhile with hypervisors and cloud providers. So What you can give us command to a Kubernetes platform Can also go directly into a cloud provider and this is important this connection Because otherwise we maybe can scale somehow Kubernetes in the workload on top of it But we still then also need to communicate and talk to the cloud provider. Well, it's easiest for API's and other stuff, but Well, you do not want to reinvent one layer after the other layer after the other layer because then it's also getting complicated to make the right fit on the area of these different layers and All these aspects together make Kubernetes actually a very perfect target to optimize Because it's acting a little bit like a clue between a couple of these dimensions but Hardest reading to say it Kubernetes is not always a solution. And I think we are also here have the responsibility to find what is the break even point for sustainability and performance and efficiency and Where it makes sense to utilize Kubernetes and where it doesn't make sense just to run a simple web server Where you have 500 people may be visiting it per week is not a reason to run a Kubernetes but If you have a very large scaled Environment of tons of applications who all behave a little bit differently scaling up down left right Have different resource read demands and so on then we talk about it And this doesn't need to be also hundreds it it is enough if you are Medium-sized already sometimes also startup-sized company because Even small companies often have very large-scale little applications so what we can do Nowadays and what we have to or to be a little bit more negative what we can't On the one hand side what we can do well that already we can optimize container images we can make them more smaller we can I'm sure that they're starting up fast and all these helps that we need less computing power less network traffic and so on We can schedule containers for high-density so rather than on one server is running one container We can pack them all together if it makes sense and We can also scale containers to zero because I don't need containers which Are the round and wait that once in two hours one request happens You can also scale the whole clusters right and And I think this is today one of the most important step, but it's often forgotten We have to optimize and know it's in the hardware. You can see it on AWS the arm-based Kravitan CPU instances at the moment are very favorable by customers We can optimize the operating system to run just containerized workload and delete every other packages and pieces of dirt We don't need on the one hand side to reduce the needed space but also In some times in some cases to reduce the amount of background noise operating system level however, what is very important to understand is that all the scaling and scheduling and Reducing sizes and so on this can happen, but it doesn't happen because we know Which carbon does it cause? We maybe know The amount of energy is used but also not that much so this is on our have-to side We have a very big issue. We need carbon data We need the energy data how much energy is used by a service to decide whether we can scale it down or delete it or whatsoever For me personally, I also have the big question in my mind Is there a future without containers? Can we even get more smaller more faster? Something which is even more resource-friendly? I Believe so, I think one of the potential answers is what assembly but this is a totally different topic Nevertheless, I think this can be again a point to improve Because rather than we need to transfer hundreds of megabytes and gigabytes for every image We need to bring somewhere we're talking about kilobytes and this means that all the processes getting faster and more efficient Nevertheless, I'm going into the wrong direction of the topic and also I very much love it I think what is very painful and these are the two last points We have to actually very seriously discuss is to design architectures for sustainability This will stay in conflict to security performance high availability and And Well depending on which kind of security requirements you have It's There's a lot of difficulties and also around the whole network communication encryption here Which all of these is most of the time at the moment quite unsustainable quite un-efficient and last but not least and I'm very sorry to say but We have to improve the power management of servers your standard server is Most of the time very poor in it and there's a couple of studies from a few universities Can find them and maybe link it also later or If you ask me I will find it in a share to you directly That has shown that servers who are in a standard motors running Are nearly using the same amount of energy like the server in the eco models There's just a little drop of energy consumption And this is because the server still needs some kind of base level of energy, but It's still quite high and I think this is also the reason why club providers develop and build their own Infrastructure their own hardware with their own scheduler with their own power management. Sorry And to be very much optimized because if nothing is running on the server, they shut it off and keep it on Just ready to be powered on if really really needed Because I think all these servers are laying around would cause millions and millions of dollars Which wouldn't make sense so Again a key missing link is on the one hand side the data about the energy consumption We have a few options which are slowly going into this direction. You can more or less get it Maybe estimate it If you're running on hardware or some hyper visors, it's even easier to get it It was even controller or redfish But on the other hand side and this is the very crucial and important point is data which is missed about the cost co2 per kilowatt hours and Here maybe if you're working at an energy provider, it would be great if you could open up some APIs to read this information For sure, you will maybe give this information also to your end user like the data center owners, but I think in a Digitalized word, it would be great to have this information on just on the hand and the fingertips to read it But yeah, I think without this information we optimize for energy consumption but we do not optimize for the cost co2 emissions and These are two different topics. I can run my software in a data center. It utilized tons of kilowatt hours and It doesn't cost any co2 and I can have one single application running in another data center Which causes tons of co2 because the only power source around it is based on coal for example So this is very important that you get better here So this is also the final boss fight which we come to we have the missing data and we have the scope 3 Scope 3 again is about the embodied carbon emissions about your hardware which you order the data center the building construction or whatever and What is everything in the production line and logistics of the production are related to this? Again, just to give you a high-level picture if you order a server There are some chips inside. There are some cables inside. There's some other microcontroller inside There's a piece of metal around it and all of these different things are somewhere else produced. So again somewhere else is someone using energy which must be produced and There are also transportation in between maybe it was a ship or a truck and then you have even systems around it like a ERP system to manage the Resources and the plan probably the resources So you see this is a very very crazy rabbit hole and in the end the scope 3 emissions are Guessed around 80% of all of the emissions when you think about a single service, so Thinking about your piece of software is running. Let's say in your private data center We're talking here more sorts of time scope 2 maybe a little bit scope 1 when you drive with the company car back and forth But everything else is in scope 3 and there is a very difficult to get this information There's a few initiatives to clarify this, but yeah, this is a long way to go. That's where a lot of Foundations need to actually help into so long story short we need your participation and if not us if you're not so much with the whole cloud native community to thinkie Maybe have a look at the other initiatives, which I have shortly named. They're also very great They have a couple of projects running which are all very very famous and Nice to see in which direction they are and how holistic they are thought of But if you're interested to work with us together and join us on the CNCF slack It will be easily able to find it search for environmental sustainability and you can find our Channel to discuss with us or find us in the CNCF GitHub repository Also, they're again search for environmental sustainability We have there's some discussions going on. You see the current work artifacts and so on So it's a good starting point if you want to directly give some small contribution or have your own idea which you want to throw into the room and Last but not least if you would like to stay updated also major decisions or whether we are getting from a working group to attack join the mailing list and Most importantly meet us virtually Our meetings are documented in the GitHub repository in the slack channel So it's easy to find us online and to join our conversations bring up your ideas thoughts and changes us a little something so Finally, we have to start now and to be serious with you. We should have started 20 years ago. Sorry two years ago but uh Better late than never I would say hopefully slightly optimistic here because Everyone needs to be with us in this boat We're talking here about open source and software, which is a part of this whole story It's not the major part of this whole story But if we want to prevent that in future 12% of the global energy consumption is caused by a team and it's maybe just six or eight or nine percent Then this savings would be leading back to the initiatives. We may be kickoff today and where we maybe start to optimize today However, we need every profession on board and everyone needs to be part of it from marketing to manufacturers to transportation and logistics Everyone has to think about and change something because only together we can change the world how it will be in the future and keep it worth living on it So thank you very much. I hope you found out how for interesting your hobby to join us come to us and I'm very excited to See you maybe in our meetings and have us joining in our discussions