Episode 55 - Decarbonizing Kubernetes with Flavia Paganelli and Niki Manoledaki

Flavia (00:00)
I got to meet Nikki and all the people at the CNCF and I had the opportunity to go to KubeCon in Paris last year, it was amazing because it's not just about the who and the what, but it's about the people. So the energy that came out of that group, I like, want to be part of this. I want to, you know, join forces to build something together

Gaël Duez (00:16)
Hello everyone, welcome to Green IO. I'm Gaël Duez and in this podcast, we empower responsible technologists to build a greener digital world, one bite at a time. Twice a month on a Tuesday, our guests from across the globe share insights, tools and alternative approaches enabling people within the tech sector and beyond to boost digital sustainability. And because accessible and transparent information is in the DNA of Green IO. All the references mentioned in this episode as well as the full transcript are in the show notes You can find these notes on your favorite podcast platform and of course on our website greenio.tech.

Cloud computing is nothing but material. It is just someone else's computer, as a popular quote says. Yet, using cloud services comes with its specific challenges for whoever is willing to seriously monitor its environmental footprint. And with the rapid adoption of cloud-based solutions, came extra layers of abstraction and remoteness with the bare-metal servers which ultimately compute and store the data.

One of these extra layers is the use of containers in the orchestration system. And they have been massively adopted in software engineering and cloud operation, the so-called DevOps. The software containers market is now a multi-billion US dollar industry with a double digit gross rate. Containers come with the promise of automation, scalability, and reliability.

Flavia (01:41)
Thank you.

Gaël Duez (02:08)
The question is how to add sustainability to the list without breaking its other benefits. To talk about these challenges, I'm glad to have two experts in Kubernetes, by far the most used container orchestration solution in the world.

Gaël Duez (02:24)
Flavia and Niki who will be at CubeCon 2025 in London next week.

Gaël Duez (02:32)
Flavia is CTO at Agnostic and a tech lead in the CNCF TAG Sustainability Screen Reviews Working Group. She has decades of experience in software engineering, and in her early days, she co-authored several O'Reilly books on AWS and also built an IoT platform She is originally from Argentina and now lives in Utrecht, Netherlands. Niki Manoledaki is a senior software engineer at Grafana Labs, where she's part of the platform engineering team but she is also an environmental sustainability advocate, keynote speaker and a community facilitator, starting with co-chairing the CNCF Environmental Sustainability Tags Green Reviews Working Group. She's based in Barcelona, Spain. And fun fact, reflecting the 2024 year for the Green IO Podcast, I was concerned by the lack of Spanish speaking guests in the lineup. And I've realized that in 2025, there are a majority so far. So I do hope that all this episode will help spark meaningful conversations, both in Spain and in Latin America. And at some point I will have to consider hosting a GreenOil conference in Barcelona, Or maybe in Amsterdam first.

Niki (03:48)
Barcelona is a great place to host conferences. We do have the Mobile World Congress already happening. Actually, it's happening soon.

Gaël Duez (03:57)
I also happens in...

Flavia (03:57)
It would be nice when there's a train that goes from here to Barcelona, but they're still in progress.

Gaël Duez (04:05)
I know. At least Paris is well connected by train, both from Barcelona, the Netherlands, UK, etc. So at least for Green IO Paris, it will be easy to join by train. That being said, hello, both of you.

Niki (04:19)
Hello.

Flavia (04:19)
Thank you for inviting us.

Niki (04:21)
Yes, it's great to be talking with you both today.

Gaël Duez (04:23)
Yeah, that's going to be a very interesting episode. And my first question might sound a bit dumb, but could you explain Kubernetes to our non-ops audience for a start?

Flavia (04:37)
There's maybe an analogy that I read once and I thought it's pretty clear. If you think about a very busy restaurant where you have a lot of guests and you need to make a lot of food and you need to organize this so everything comes out. So there's enough food for everyone at the right time. can think about Kubernetes in that way. Kubernetes is the chef organizing all the cooks. You have maybe the containers can be thought of as the individual chefs. And each individual chef has to be in charge of making one specific dish. And then you have the nods, which are the kitchen stations for different purposes, like a kitchen station for grilling, another for baking, another for chopping. And then the pods are like a team of cooks working on one single order. like in Kubernetes, you need to handle load or scaling properly, right? You need to make sure that the orders go to the right people so that they don't get overloaded. You need to put more cooks on a specific dish which is more popular, et cetera. If one group of cooks have a problem with something, then there's always. Yeah, the authority and the organization to fix that. So concepts like scalability, reliability, you want your meal to get to the table and on time. Those are maybe nicely included in this metaphor.

Niki (06:34)
And just remember for listeners who may be less familiar with the cloud, we're talking about hundreds of servers. So hundreds of computers and, how do you get the dish, which is the, application that a user is trying to, to access, available for hundreds if not thousands of users. So we need to make sure that this application is available on every server. And that's what Kubernetes does is it orchestrates that all of these dishes are available at all of the tables. So all of the applications are all of the servers for everyone to be able to enjoy them.

Gaël Duez (07:16)
Got it. And now that you've mentioned all these servers, let's go to our main topic, which is what are, according to you, the top sustainably challenged running containers and maybe more specifically Kubernetes.

Niki (07:32)
So monitoring and auto scaling is what I would say the two branches of sustainability in Kubernetes. So we need to make sure we run everything as efficiently as possible. So we need to make sure resources are allocated in the most efficient way. So to do that, we need to be able to observe these resources how much CPU, how much memory, energy, and various other things. And then on the auto scaling side, we need to ensure that we are not deploying more than what you need. So everything needs to be basically packed together as tightly as possible so that we don't have idle resources just laying around and not being used.

Flavia (08:23)
So that's a really good document. And I hope that these resources translate around what you are being used.

Gaël Duez (08:30)
And is it the case today? Because I've seen some numbers in non-scientific studies, so I won't quote them here, but saying that in general, we are facing a massive over-provisioning of resources due to auto-scaling and all of this, is it true or is it a urban legend according to your experiences?

Flavia (08:51)
Yeah, it is very true. Last time I was at FOSDEM giving this talk and I asked people, did it ever happen to you that you found servers that were running that were not they didn't let me finish my sentence. Everybody was like, yeah. So, I mean, yes, this is, this is a very common problem. And I think it roots from the fact that, ever since we started using cloud, we can spin anything, anytime very easily without realizing the consequences immediately in terms of, yeah, not only price, but impact the environment

Gaël Duez (09:36)
And this tendency to overuse them, how can we fight back, especially from a sustainability angle? I know that both of you, you've been working on a project named KEIT which, the goal to automate the monitoring of the energy consumption, but maybe it goes beyond and it goes all the way to carbon estimates. So you will correct me if I'm wrong

Flavia (10:02)
So with KEIT, we are basically showing in what way your software and infrastructure and hardware cause an impact and what is source of this impact. Basically we use the software carbon intensity formula which is an ISO standard it considers three aspects. It considers the energy consumed by the software. It considers the carbon intensity of the energy used and the emissions of the hardware. in a way, you can see with the software carbon intensity, you can see where you have the most potential to improve. Or at least you can, you can observe it. can make changes and then you can improve. You can see as well which part of your software is, is generating most emissions. For example, looking at different namespace, looking at how many nods you have, et cetera.

Gaël Duez (11:06)
OK, got it. And my question is, where does this number come from?

Niki (11:11)
From the past two, three years while I've been building the CNCF environment sustainability TAG I've worked on the open source technologies that underlie the KEIT project. So I haven't worked on the KEIT project directly large part of what it's based on is open source tools that I've been maintaining or helping to build. And that includes, for example, Kepler, is an energy monitoring tool within Kubernetes. that is a tool that we could get very technical. Like really, some of the measurements from the kernel of the server. We basically, through Kepler, we're able how much energy is being consumed on the server and what is it linked to? which application is emitting or consuming this energy, which is measured in millijoules. So that's one of the components of the software carbon intensity specification to the same of that we have.

Gaël Duez (12:26)
Niki, just to clarify, I've got a bunch of questions regarding KEIT. The first one being, when you extract the energy consumption from the kernel, is it a measurement or is it an assessment via some sort of a low-level model?

Flavia (12:45)
You have both options because either have the, I don't know what it stands for, but the RAPL

Niki (12:53)
Running average power limit.

Flavia (12:54)
Okay, yeah, so component in the chip which lets you measure the energy consumption of the hardware or if not the nice thing of Kepler is that there are models to estimate them if you don't have a chip with the RAPL. But yeah, in general, course, everything is an estimation. even this RAPLE measurement. It measures only, I believe, the CPU but not everything else around it. So we do have to accept as engineers, even if we don't like it, that everything is an estimation and just work from there. It's better to have estimations than to have nothing.

Niki (13:41)
Yeah, it's a model, right? So everything is a model. Some models are more useful than others. there are so many different ways to measure energy consumption. But then there's embodied carbon and the energy that went into building the physical components of the server, for example. But as Flavia mentioned also, we do have gaps there are gaps such as networking. like gateways and other networking components are notoriously hard to measure. I know there's different communities, like different open source communities, and there's conversations amongst us, but still networking is something a lot of people are working on and that is really difficult to measure.

Gaël Duez (14:34)
So, if I understood right how KEIT works, you've got this energy consumption, either via RAPL or via other estimated model, that you translate into carbon emissions. Maybe this is the missing point. I guess you're using solutions like Electricity Maps or WattTime. Am I correct to assume this?

Flavia (14:56)
Yes, exactly.

Gaël Duez (14:58)
Okay, and then you add on top of it the embedded carbon from the servers, but you could not really include all the networking part. Is it correct summarize what you've said before like this?

Flavia (15:11)
Yes, yes, I think so. So it's those three sides of it and their limitations, I would say, almost in all three parts of the formula, because for the embodied emissions, for example, it's very difficult to get information, already difficult to get embodied emissions, but they are usually incomplete because there's not information about the disposal or the recycling of it. It's not disclosed by data centers or by cloud providers, or they don't say how many years the hardware is being used because that also has an impact on the embodied emissions.

Gaël Duez (15:54)
And do you use the data from the manufacturers or do you have other sources of data?

Flavia (15:59)
We use Boavizta, which is from a French, non-profit. so they have a whole database of the different hardware. so they have an API. You can say, I have this. I don't know, instance type. Tell me what are the embodied emissions. in KEIT we do that dynamically because the instance types can change to include this in the formula and calculate the SCI score.

Gaël Duez (16:33)
Thanks for the clarification. And do you have any success stories to share yet about how KEIT has been used in some organization or within some teams?

Flavia (16:44)
Well, I do have to say it's at the very beginning, project, but We are the moment working with one client, which is the Consumers Association, a non-profit in the Netherlands. so we installed it because basically you have your cluster, can deploy pretty easily and then you have an overview of your score. But we also added this widget to show the number of nods that you're running. And then we immediately saw like, okay, we're running too many nods in an environment which is just for development. So we are now working on improving that on setting up carpenter consolidation to make sure that only the necessary nods are being used at a certain time that's a very, for me, very nice example where it's helping.

Niki (17:43)
And at Grafana Labs also we're deploying Kepler at scale. It's quite challenging because Kepler for example, we mentioned, which is one of the components in the SCI and it relies on certain information that it should be able to fetch from the kernel but in the public cloud such as AWS, example, or GDP or Azure, this information may not be accessible. So there's certain other estimate methods that it needs to use. So all this to say that running this energy and carbon estimation models at scale is very difficult. And from my side at Grafana Labs, We are trying to run this at scale and kind of find which issues we come across and how can we deploy it in a production environment with as little issues as possible. So all of this is still a work in progress, but it's really great that we can come together in the open source community and kind of exchange this knowledge.

Gaël Duez (18:56)
For the clarification that this project is still an early stage project. People interested in this specific project, they can join you at TAG sustainability group. And actually that leads me to a question I wanted to ask in the beginning of this interview could you help us maybe make some sense of the true alphabet soups around CNCF What is CNCF? What is TAG? What is the Green Reviews work group, for instance?

Niki (19:26)
So the CNCF stands for the Cloud Native Computing Foundation. it's a project actually of the Linux Foundation, which is a nonprofit organization that hosts a lot of open source projects. And the CNCF was created, I think in 2015, to host Kubernetes when it was

Flavia (19:41)
I think 2015 took host Cabrera's when the organization was initially created by the...

Niki (19:49)
Initially created by Google and then was donated to the Linux Foundation and the Linux Foundation created the CNCF to host it. And then other projects were hosted also by the CNCF such as Prometheus, which is a monitoring, well, a time series database for monitoring metrics. And then other projects joined and now there's, I don't know, like maybe dozens if not hundreds of sandbox projects. I'm not really sure on the number. But Kepler, which we talked about previously, is one of these sandbox projects that was donated by Red Hat and Intel the CNCF. So then with different subjects around this tool such as security, there was the creation of are technical advisory groups. And one of these tags that was created two years ago is the environmental sustainability TAG. And in this TAG, we promote and advocate for tools and use cases around cloud native sustainability, including Kepler, for example, and KEIT and other things like the SCI, the software carbon intensity specification. We talk about this at KubeCon, one of the biggest conferences that is related to the CNCF and FOSDEM as well. There have been talks about this cloud native sustainability tooling.

And finally, we come to about a year and a half ago, we wanted to have a technical project where we can really get into the nitty gritty of how do you deploy Kepler? How do you link carbon emissions metrics such as WattTime and Electricity Maps, how do you get the embodied metrics from Boavizta and how do you calculate the rates of the software carbon intensity of tools and we created the green reviews working group. The idea for that is create reports or we report on the sustainability metrics such as energy use and carbon intensity and other more traditional metrics such as CPU usage and memory usage. we've been doing that with Falco, which is a tool. It's a security tool, a project hosted by the CNCF to work with the students and care for the students. And other things so we've been basically trying and innovating and just new ideas come, new people come to contribute their ideas, they want to try something, and we have this space to make it happen. And KEIT is kind of emerged from this work, I think. Flavia, maybe you can tell us more. yeah.

Flavia (22:53)
Well, we had this idea of making, we initially called it a sustainability plugin. So trying to reflect in a Kubernetes environment, the environmental impact of the infrastructure and software. And we a lot of ideas on things that we wanted to see there, only I thought it was pretty complicated. So I thought, how do we do this? And then I started looking around what's out there then I found the CNCF environmental sustainability TAG and I joined I saw this green reviews project and the software that was being worked on. I thought this is at least very good reflection of what we want to build. But for generic Kubernetes clusters. So I learned about the SCI and deploy actually maybe these things, because they are not so concrete, you don't see the power in it. But when I got to meet Nikki and all the people at the CNCF and I had the opportunity to go to KubeCon in Paris last year. It was amazing because it's not just about the who and the what, but it's about the people. So the energy that came out of that group, I like, want to be part of this. I want to, you know, join forces to build something together.

Gaël Duez (24:15)
Actually, it's an interesting mention that you've done because I wanted to ask both of you a bit more about how it is to run source projects and to have this open source community working together, especially for, I would say, greater good project. Just before we jump on this of questions, to understand clearly what the Green Reviews Working Group is. I understood well, it's a bit of a permanent brainstorming work group on everything related to sustainability within the CNCF space. Or does it also has, I don't know, sub-team fully dedicated to maintain or code or create new products. I'm having a hard time understanding the connection between this work group and the tools that you've mentioned before, such as Kepler or KEIT.

Niki (25:20)
So the environmental sustainability TAG is kind of the broader brainstorming group. In the Green Reviews working group, we are creating a benchmarking pipeline. So we are doing benchmarking tests for cloud native tools such as Falco.

Niki (25:40)
to measure the softer carbon intensity rates. it's very similar to KEIT essentially, and we run benchmarking tests measure the different factors of the software carbon intensity specification. So that's the runtime energy, the emissions impact, the embodied carbon, and all of this using a unit of work, so like a rate. So yeah, the idea was, to take the software carbon intensity ISO specification when it became public last year, and to create an example using cloud native tooling to show how it can be done in Kubernetes.

Gaël Duez (26:20)
Okay. And one last logistic question. If people want to join you, do their organization needs to join? Can it be done on a voluntary individual basis? Do they need to fill a form or how does it work concretely?

Niki (26:38)
It's fully open source, so everything is completely out in the open. a Slack organization for the CNCF. And if folks Google CNCF community invitation, they will get a portal to where they can put their email address and get an invitation to join the CNCF Slack. I know that a lot of organizations join the CNCF, but we don't require that in the TAG or in the working group. So anyone can join the meetings. The meetings are twice a month. And we do a lot of planning during those meetings. And we talk about the different pull requests that people are working on or that need to be reviewed. then we triage some of the issues And we do have like a agile workflow. So anyone is welcome to join. And then we have our GitHub repo where people can see issues that are open and that are beginner friendly. And we have the issues board where people can see which ones are next and ready to be picked up if they want to contribute. So people often will join for a bit, maybe pick up an issue, contribute a way, learn. And then maybe they'll stay for more long term or some people come for a bit just to like learn. So just like any other open source project, really we're looking for contributors and long-term maintainers and we're open to everybody.

Gaël Duez (28:22)
Talking about open source projects and Flavia mentioned it a bit earlier. FOSDEM celebrated its 25th anniversary this year. Shall we say that everything is doing well for the open source community and maybe more specifically for the projects related to sustainability? Or are there some challenges that we don't necessarily see beyond the success?

Flavia (28:46)
Yeah, there's always challenges. especially that people find time to work on it. Because is an area where people get attracted to it because they want to do something good. They want to have a purpose. So you do find a lot of people wanting to join the environmental sustainability TAG or help in the green reviews project. But it's hard to find people who have the time or who are ready to spend some time on it.

Gaël Duez (29:23)
Flavia, before discussing more specifically the time constraints, I had a specific question regarding the KEIT project when I see how imbricated it is with other open source projects. How do you manage to build something that is so heavily relying on other components which are also open source?

Flavia (29:47)
Well, first of all, it's great to have those tools because otherwise we couldn't have built what we built. We just put them together. what we needed to check is the licensing to make sure that we are not infringing any of the licenses of the software that we are using. so far we've had good experience. for example, one of the projects that we are using is so there's electricity maps and there's an exporter for Prometheus to get that data into Kubernetes. It's called the grid intensity, From the GreenWeb Foundation. And it's also open source and there was something that didn't work exactly like we expected or we needed to make an improvement and we just made the change. Applied our PR. It was approved and problem solved. So in that sense, it can be easy because given the need, you can just make the changes yourself.

Niki (30:53)
I think this is the issue of open source tooling and how well maintained it is, or if there is enough contribution to keep it running is kind of a bigger problem in the industry and for Kubernetes and the Kubernetes ecosystem as well. Like even though a lot of these projects are hosted by the CNCF. Oftentimes, there isn't everything that you might need your use case. And maybe someone needs to find time to contribute back to the open source project. For example, like if there's a feature missing or if there's a bug. Oftentimes there needs to be that kind of instinct of like, okay, this is missing. I'm going to take some time to contribute upstream. Upstream meaning to the open source project. And that's how the tools can continue to exist. But there also need to be long-term maintainers who are sponsored by their company, maybe, who are given the opportunity to take time to contribute to open source projects as part of their work. one of the main challenges in open source projects, I would say is, people finding time to contribute to this project.

Gaël Duez (32:18)
Niki, that's very interesting that you mentioned people being employees of corporations, large or not, and having some of their professional time allocated to open source. Because I've been recently recording in releasing actually an episode on the WordPress sustainability group that has been quite brutally dismantled. And that was part of a bigger drama of WordPress governance it made me realize how much the WordPress community is dependent on automatic. Is it something that you fear also for the CNCF? Is it something that is more the exception than the rules or what is your? Yeah. Point of view on this dependency that we might find from time to time.

Niki (33:09)
the CNCF has managed to build a huge community of people who are excited and able to contribute to their projects. And this is an extraordinary feat of like modern technology and like community building. So I would say I'm not. I'm not worried about Kubernetes itself or the really large projects hosted by the CNCF, there's always an underlying worry or an underlying realization that we need to contribute back to what we're using. And maybe that's kind of part of the success of all these open source projects. if a lot of companies are depending on it for their operations and then they have an incentive to contribute back to keep things running. a sociological aspect, I find it fascinating

Gaël Duez (34:09)
And maybe because you've sort of already explained it, but if both of you, had to, I don't know, bring three ingredients to create the perfect recipe for a thriving open source community, such as the that you belong to, what would it be?

Niki (34:29)
Great question.

Flavia (34:30)
I would say purpose is one all these people also that you see in the open source community, like in conferences, like FOSDEM, they really believe in something. They are not just techies, but they, I mean, they are techies, but they, believe in the power of the people building something and being free and open. So purpose is one. Now what I was going to say, so everybody who is there really wants to be there, contribute to something bigger than themselves. And I wanted to mention because the TAG ecosystem in the CNCF now is going through a restructuring because well, after many years of the development of the Kubernetes software ecosystem, things changed. And so we used to have the TAG environmental sustainability, TAG security, TAG application delivery, and a couple more. And now this is all going to be restructured. Does this have anything to do with the current situation? Because the NCF, even if it's a… and nonprofits, it is based in the United States. Does it have anything to do with the political situation in the US? I don't know, maybe. So the TAG environmental sustainability will be renamed, will be part of a different TAG. My point being that even through all of this, there's all this bunch of people that I'm part of, which is want to continue. We don't care the naming. We just want to continue with what we are doing. I don't know in the case maybe of WordPress I don't know because I couldn't finish the episode of last time I don't know if the people who were working on it will continue because in the end it's all open source. You can still do what you want.

Niki (36:35)
I mean, I think that the open source community enables us to work on things that otherwise would be difficult to do on the day to day. Because there will always be changes in the business direction, for example. Whereas open source work means you can always get back to it and continue some work and do it for as long as the community decides to work on that. So it does give a lot of freedom. another thing I would say is that new problems require new solutions. think innovating in the open is a great way to build new solutions. For me, innovating in the open is the best way to really bright and motivated people involved to come together and try to solve this together rather than behind closed doors, which is what business often is like. So it's really great to be able to maintain those spaces. And I'm glad that the Green Reviews Working Group will continue to exist as a project in the CNCF, despite of the reorganization that is happening. And we'll know more around KubeCon Europe, which is going to happen in April

Gaël Duez (38:01)
Thanks a lot for sharing your opinions and your feedbacks on how to run successfully in open source community. I could not not hear some or expectations regarding this reorganization? Do you already have some information which has been shared or will it all come as a big surprise at KubeCon Europe as you say there, Niki?

Flavia (38:22)
It was already shared, the reorganization. So there's going to be like five TAGS and sustainability will be part of, operational resilience, I think. So it will be more up to us to make this visible and to keep it going. the community is there, people want to improve. So we'll keep going.

Gaël Duez (38:57)
Concretely speaking, it's a general reduction of the number of TAG groups within the NCF, am I right? Or is it just specifically targeting the environmental sustainability TAG?

Flavia (39:11)
It's a reorganization of all the tags. So some are merged, some are new

Gaël Duez (39:13)
Okay. Okay, got it. And so now you've got bigger tags, five only, and I guess they will structure in subgroups, subworking group, like some of them working more on operation or resiliency, efficiency, sustainability, or how will all those different projects will be maintained with this single big TAG, for instance.

Flavia (39:38)
Yeah, there will be, I think, like sub-projects. in theory, working groups will continue running. They will just be part of a different TAG. So we will have to see if the chair of the talk where we are in, we hope that they all that is also someone who considers sustainability important. it's all volunteer work, but we get from the CNCF, we get resources for, for example, infrastructure and where to run our software, pipelines, our databases. So yeah, we'll see how this develops, but we all want to continue.

Niki (40:25)
I'm excited also to see how sustainability can fit in the operational resilience kind of story, includes, for example, it observability, more concretely, like traditional observability, as opposed to like carbon observability, right? I think it's interesting to see how it will become part of that operational resiliency story. think it might actually, in the best case scenario, might help folks to explain how sustainability fits in software.

Flavia (41:09)
It's software.

Niki (41:10)
I guess

Flavia (41:10)
I guess.

Niki (41:11)
If we're talking about resiliency, I don't think it's too far away. I'm hoping to learn from that. I think having closer proximity to other domains will mean there will be a lot of like cross pollination of ideas and lessons learned and more contributors maybe, maybe contributors who would not have approached sustainability, learn more about it by being in the same TAG. So I'm hoping to find like positives from this and I guess we'll talk about it more at KubeCon and we'll see how it will go.

Gaël Duez (41:52)
That's a good point and it's a very optimistic way to see things that I really enjoy. Maybe we will have to redo an episode in a few months or before the end of the year to see how, yeah, to be continued. Yes, absolutely. to understand a bit how things developed. The WordPress situation was a worrying signal, but it doesn't mean that all the open source community needs to be continued, shut down sustainability or reorganize it. So if you feel optimistic, that's great. And I think on these very positive notes we can close this podcast.

Flavia (42:29)
Yeah.

Gaël Duez (42:32)
I really want to thank both of you for joining and explaining both the projects you've been running, but also the way you've been dealing with these open source projects and how this open source community at CNTF works. that was really enlightening. So thanks a lot for joining the show today.

Niki (42:53)
Thank you so much for having us.

Flavia (42:53)
Thank you, it was fun.

Niki (42:54)
Thank you so much for having us.

Flavia (42:54)
Thank you, it was fun.

Gaël Duez (42:56)
Thank you for listening to this Green IO episode. If you enjoyed it, please share it on social media and TAG your connections working in cloud operations and DevOps. If you attend KubeCon next week, this would also be a useful tool to kickstart conversations. And of course, don't forget to give us five stars on Apple Podcasts, Spotify, or your favorite podcast platform. It helps us reach out to more soon-to-be responsible technologists. 

In our next episode, we will celebrate a birthday with Sarah Hsu. A year ago, her book, Building Green Software, was released. And she will tell us everything about its impact, the feedback she received with her co-author Anne Curie and Sarah Bergman, and the trends she sees in green software, including AI training and inference. Stay tuned! 

One last thing, Green IO is a podcast and much more. So visit greenio.tech to subscribe to our free monthly newsletter, read the latest articles on our blog and check the conferences we organize across the globe. 

Green IO Singapore and Green IO New York are just around the corner, respectively in April and May. Early bird tickets are gone. But you can get a free ticket using the voucher GREENIOVIP. Lucky you. Just make sure to have one before the 50 free tickets are all gone. I'm looking forward to meeting you there to help you, fellow responsible technologists, build a greener digital world.