#216 January 9, 2024

NAIS, with Johnny Horvi and Frode Sundby

Hosts: Abdel Sghiouar, Kaslin Fields

This week’s guests are Johnny Horvi and Frode Sundby from NAVs (Norwegian Labour and Welfare Administration) platform team. We talked about NAIS. A kubernetes-based team centric platform aiming at providing the tools needed to deploy and operate apps easily.

Do you have something cool to share? Some questions? Let us know:

News of the week

Kubernetes 1.29 features:

Kubernetes 1.29 release lead Interview

Cisco acquired Isovalent

Cilium 2023 Annual report

KubeCon and CloudNativeCon Paris 2024 Hackathon

OpenFeature incubated as a CNCF project





IBM Websphere

Apache Mesos

Links from the post-interview chat

Nais on GitHub

ABDEL SGHIOUAR: Hi. Happy new years and welcome to the Kubernetes Podcast from Google. I'm your host, Abdel Sghiouar.

KASLIN FIELDS: And I'm Kaslin Fields.


ABDEL SGHIOUAR: This week, we speak to Johnny Horvi and Frode Sundby from NAV, the Norwegian Labour and Welfare Administration. We talked about NAIS, a Kubernetes platform that allows developers to focus on building apps without worrying about the underlying complexity of Kubernetes.

KASLIN FIELDS: But first, let's get to the news.


Since the release of Kubernetes 1.29, the community has been busy publishing detailed writings on the features in this version, starting from the out-of-tree cloud integration removal to contextual logging and more. You'll find the list of these articles in the show notes, if you want a deep-dive into what's new in 1.29. Also, don't forget to check out our interview with the release lead.

ABDEL SGHIOUAR: Cisco acquired Isovalent for an undisclosed amount. The network giant merging with the cloud-native firm known for its leadership position on eBPF and Cilium aims to strengthen Cisco's position in the cloud space.

KASLIN FIELDS: In Cilium news, the community released their annual report. This is a significant one for the project, as Cilium reached graduated status in 2023. The report highlights the growth of the project, both in contributions and adoption. The link is in the show notes.

ABDEL SGHIOUAR: KubeCon and CloudNativeCon Europe in Paris will bring the first ever CNCF hackathon in collaboration with the United Nations Office of Information and Communications Technology. The hackathon will focus on advancing the delivery of the UN Sustainable Development goals. The in-person hackathon will run alongside KubeCon Paris from March 19 to 21, 2024. The winners will be announced during the keynote on Friday, March 22.

KASLIN FIELDS: OpenFeature is now a CNCF incubating project. OpenFeature is an open specification that offers a vendor-agnostic API for feature-flagging and is compatible with various feature flag management tools. Feature flags are a software engineering way to enable, disable, or alter the behavior of certain features without modifying the source code of the application. And that's the news.


ABDEL SGHIOUAR: Well, hello, everyone, and welcome to a new episode of the Kubernetes Podcast. I'm your host, Abdel. Today, I'm talking to Johnny and Frode. Johnny and Frode are developers at NAV's platform team. We are here today to talk about NAIS, spelled N-A-I-S, but pronounced "nice," like nice things, a Kubernetes-based team-centric platform aiming at providing the tools needed to deploy and operate apps easily.

In other terms, the framework is a non-pretentious platform engineering tool. Welcome to the show, Johnny and Frode.

FRODE SUNDBY: Why, thank you.

JOHNNY HORVI: Thank you very much.

ABDEL SGHIOUAR: So we're going to talk extensively about NAIS and NAV, but let's just start with a very high-level introduction. Who are you? And what do you do at NAV? Maybe we can start with Frode.

FRODE SUNDBY: I'm now a developer working at the platform team. I've been in NAV for, I think it's 12 years now. And for the duration of my time in NAV, I've been working on making it easier for developers to get applications out to production, ranging back to the days when we built a platform for JBoss and WebSphere applications to now containers and Kubernetes.

ABDEL SGHIOUAR: Cool. What about you, Johnny? Can you introduce yourself?

JOHNNY HORVI: Sure. I'm Johnny. I have worked in NAV basically since 2009, started off as a consultant in 2016. I joined NAV as an internal employee. And all along, I've been working with providing infrastructure functionality and starting off with basic configuration management and deployment mechanism automating this and moving on to more infrastructure. And then in 2017, we started building NAIS.

ABDEL SGHIOUAR: Speaking of NAIS, let's just get going. We'll talk about NAV later. But can one of you tell us a little bit, what is NAIS? What's the history of this platform?

FRODE SUNDBY: Yeah, as I mentioned in my introduction, we had some deployment tools and some ways of making applications go to production for developers in NAV. We've been having that for a long time. As Johnny said, he started working on this thing back in 2009 or something, so it wasn't a complete foreign concept to NAV even going back a long time.

But then in 2016, I think it was, Johnny and I attended a course with Kelsey Hightower, where he showed us the magic of Kubernetes. And, yeah, we were quite blown away. And so, oh, here's something we could make use of. This is a lot more efficient than automating JBosses. Let's try it.

JOHNNY HORVI: So we've been looking at containers and Docker for a while. And at this time, it wasn't that clear that Kubernetes was the absolute best choice for everyone because you had Mesos and you had other alternatives. But after this workshop with Kelsey, we figured it was time to go.

ABDEL SGHIOUAR: So you had a workshop with Kelsey and then you decided, OK, Kubernetes is the way to go. Did you started building NAIS from day one, or was it more like, oh, let's just start using Kubernetes and figure out if it works for us and then eventually down the line figuring out we need a better tool? What was the story there?

JOHNNY HORVI: Initially, when we started building this new platform, NAV was going through a large change in its organization as well. We were going from a kind of traditional IT where developers were simply developing applications and delivering software packages to the organization, and NAV had been doing this for a long time. And then it became obvious, even to the Norwegian government, that NAV wasn't able to actually deliver functioning services and software even though it had a very large budget.

So in around 2015, '16, the entire management was switched out and NAV started focusing more on building autonomous teams, allowing them to take full ownership of their applications for the entirety of the life cycle of the application. This was the point where I joined NAV as well. And we figured we had to build a platform supporting this new way of working, basically, allowing the teams to be able to see and understand what's going on with their applications.

ABDEL SGHIOUAR: Nice. And that's actually why in the introduction I said it's a non-pretentious platform engineering team because I feel like this whole space of platform engineering and people talking about it, they're all more or less trying to solve the same problem. It's just that sometimes the focus is too much on the actual tooling rather than the concepts of like-- or rather than the focus is mostly, let's try to make developer's life easier, right? So speaking about that, can one of you give me like a quick introduction of what is NAIS? We know what it's trying to solve, making developer's life easier, but what is the platform itself?

FRODE SUNDBY: Technically, it is pretty simple. It's just Kubernetes. And we've built some abstractions on top of Kubernetes, making Kubernetes more accessible. Like, if a developer or an application needs X amount of runtime, they need an ingress, they need a database, they can just express this in a one-application manifest and then deploy mechanisms, and we will make sure that all the Kubernetes resources are available and they are functioning together. In a nutshell, it's just making Kubernetes and the runtime environment around the application easier, that's what it is.

JOHNNY HORVI: And just for me, for higher abstraction, the problem that we're trying to solve is to allow our developers to primarily focus on solving business problems. And today, a developer kind of have to care about a lot of things if they are to actually develop and run their application.

And this ranges from, like I mentioned, from runtime environment to, how do I secure my application, how do I provision databases, messaging queues, observability tools, so they know what's going on. And what NAIS is is trying to take all these different aspects and package them into products, deliver this to the application teams so they can choose the things that make sense for them.

FRODE SUNDBY: Yeah, building blocks, sort of.

ABDEL SGHIOUAR: Yeah, that's actually interesting because I think a lot of times when people talk about abstraction layers on top of Kubernetes because, I mean, let's face it, Kubernetes isn't the most developer-friendly tool out there. But when people talk about abstraction layer on top of Kubernetes, they mostly focus on the runtime part, which is what Frode described, so the, my application, my ingress, my secrets, my volumes, and all that stuff. But NAIS does more than just that. Like, you can provision dependencies if you want to call them this way, like databases and other stuff that an app might need.


ABDEL SGHIOUAR: Then, the next question, maybe if somebody listens to us and they don't really understand even what Kubernetes is, assuming everybody does because it's the whole point of the podcast, but let's take a step back. Can you help us understand it in the context of existing, maybe, third-party or open-source tools? What is the closest thing you can compare NAIS to?

FRODE SUNDBY: I think we're not entirely there yet, but more in the realm of fly.io or Heroku, where you can just point to your code and say, please run this for us, and then the platform will give you all the tooling and all the insight you need as well.

JOHNNY HORVI: After running this platform now since 2017, what we see is-- or the pattern and the use cases that we see is that you've covered a lot of ground if you're able to provide a solid runtime environment and good solutions around logging, and you have a SQL database and a messaging service and maybe a bucket, then you are pretty far along. It's bread and butter, basically.

FRODE SUNDBY: Yeah, that is actually quite interesting because we've started, we thought that, OK, hey, hello, guys, let's open up this platform. Here's Google Cloud for you. You got your own project. You can do whatever you want. It's yours. You've got a namespace, it's yours. You can do whatever you want. All the resources are available to you.

And what do we get? We get pods, databases, and buckets. That's all people are interested in and actually just their application, making it work, and the fancy things. We'll dabble with it, but not too much. We thought we would get so much innovation, but it's just bread and butter, as Johnny said.

ABDEL SGHIOUAR: It's actually interesting. I've done some consulting in part of my career, and it's interesting to always think that people have unique requirements, but then it's like 99% of the use cases are pretty much the same. I need to run some code and write to a database and maybe store some files somewhere, right?


ABDEL SGHIOUAR: That's interesting. So no spanner, nothing too fancy. No Vertex AI, nothing too fancy.

FRODE SUNDBY: Some teams have dabbled with it, but it's not a big thing going around here. And I think that perhaps is showcasing the point of the platform is that the application portfolio in our organization, it is pretty standard. They just need the basic things. And if we make a platform that makes the standard stuff, the bread and butter, make that easy, that means the teams that own the applications, they want to run on our platform because we provide exactly what they need and we don't try to do everything else at the same time.

So the things that we build, it is sort of like a golden path for the application team saying, OK, we provide a database. We haven't given them too many options. It's a managed Postgres, that's what you get. You're allowed to do, use whatever other database technology you want. You can set it up yourself and manage it, but we've built Postgres, and pretty much all the teams use it. And even though we're not saying, use Postgres, because that is what we have decided, people use it because it's easy.


JOHNNY HORVI: I think what we've discovered is that a lot of, or most of the developers, as long as they're able to just run their application and get this basic functionality and it just works for them, they really don't care. They don't care about Kubernetes. They don't care about the infrastructure parts. They might have some knobs and buttons that they want to tweak, but in general, it's pretty low interest.

And this was kind of interesting because starting out, we thought that because we were probably biased ourselves, because we are interested in these things, they must be too. And we're kind of exposed to those developers that are super interested in these things and they are very active.

But I think that the large majority of the developers don't really care as long as this works and that they are provided with sufficiently insight into what their application is doing, how it's behaving, how the thing that they are creating is affecting the world. So we don't create a magic layer that fixes everything for them but we try to kind of remove some of the complexities and make it easier for them to understand what's going on without being super experts in every single aspect.

FRODE SUNDBY: Well, that is a good point because we've been running Kubernetes since-- oh, we've been dabbling with it since 2016. And we know our services and deployments and secrets and everything. But still, we use our own tools because it's just so much easier to just write an application spec than to remember and to fill in all the different YAML files. It's just easier.


JOHNNY HORVI: So the application spec is the abstraction that we've created on top of all that is needed to kind of stitch together a functioning deployment workload in NAIS.

ABDEL SGHIOUAR: Can you walk me through the life of a developer on the platform? So I'm a new developer, I have an app. I built my app. I probably have a container, or maybe I don't. And I want to have a running application with an URL, a database, login, monitoring, whatever the platform offers. What would be that golden path?

FRODE SUNDBY: Well, starting in NAV, our smallest interaction with the platform is from a team. So you are probably a member of an existing team or you are starting. So let's say you are starting in a brand new team. You would go to our console and say, I'm starting in team X.

You go in, you plot it in, say, this is my new team, and you hook off all the things that your team needs. You need a Google project. You want a deployment key. You want a GitHub repository and all the bells and whistles that your team needs. And you say, who are the members of my team?

And then the console will set up all the things in the ecosystem around, like the things that you want, and you can start writing your code. You write your "hello, world" application, you push it to your repository, and you use the deployment action to say, we want to deploy this application to this cluster using the key for my team. And your workflow will then build your code, test it, and deploy it to the cluster of your desire. And it's up to the teams then to build their workflows in a way which they are comfortable with, like deploy it first to the test production, run all your tests in the environment, then deploy to the production environment, for example.

And the way that you describe what your application needs, as Johnny said, in an application manifest, which is just the Kubernetes' custom resource definition where they say, "this is the ingress I want. This is how many pods I want, minimum, maximum. This is the amount of CPU memory I want. I would like a database and a bucket, please." And then once deployed to the cluster, our operator will take over and create all the required resources in the cluster in their namespace.

ABDEL SGHIOUAR: OK. Yeah, so basically, you describe intent in the form of a YAML file, and then an operator plus a bunch of CRDs will generate the low-level Rho Kubernetes objects for you.


ABDEL SGHIOUAR: And then you can define-- I guess, you could call them dependencies. Do you consider, for example, login and monitoring as an optional or is it like they're out-of-the-box, available for people?

JOHNNY HORVI: We use Prometheus for monitoring. So in your application manifest, you simply specify which endpoint that you expose your metrics on, and the platform will automatically scrape this endpoint and put it into Prometheus. And then we have a Grafana instance where this is hooked up as a data source, and it's available there. And there we have some standard dashboards, and you're able to make your own.


JOHNNY HORVI: And same with logging, basically.

ABDEL SGHIOUAR: OK, so it's basically available out-of-the-box. And then you can optionally choose, I want a database, I want to expose via an ingress on a specific domain, those are like extra features, I guess, right?

FRODE SUNDBY: Yeah. So you can start in your application manifest just saying, this is the image that I've built and pushed to the central repository, and that's it. That's pretty much all you need to tell us. This is my team, this is my image, run it, please. And then everything else is sane defaults or optional add-ons.

ABDEL SGHIOUAR: I see. Cool. And then talking about this, like, is the platform itself composable? Can people just swap out certain components for things they want or, from experience, as you said earlier, did you realize people don't really care too much?

JOHNNY HORVI: The way it's set up is that we don't have a lot of different options within each aspect. You can't select from three different login tools. So we provide a standard option that we feel covers most of the use cases. But since you have access to your own namespace, you have full access, basically. So you can create the kinds of resources that you want if you have other requirements that go outside of what's offered through the platform, so that's entirely possible.

But like I mentioned earlier, it's very rare that this actually happens. And after having run the platform now for several years, we have covered at least a lot of the use cases, and most of the functionality is there. But it's still an option to set up things yourself.

FRODE SUNDBY: If we see that there are a lot of teams who are spending a lot of time figuring out how to do a specific thing and they are all trying to do the same thing and solve the same problem, then it's when we as a platform team come into the picture and say, OK, we've got X amount of teams trying to solve Redis, for example.

That is when we will add an operator and expand the application manifest to say, OK, you can now, say, have a Redis and build the tools for them to have that out of the box as well. So when it makes sense for us to build it as a central or a platform feature, that is when we will do it. If it is just one or two teams, we're not building platform functionality for a small amount of special use cases.

ABDEL SGHIOUAR: I see. Cool. I like this, how the conversation started off. Over time, you realize most people just want standard things, like bread and butter. So people don't really care too much about, "I just want to write code, solve my problem and just move on with my day," right?


ABDEL SGHIOUAR: So I want to switch gears a little bit because we started talking about NAV, NAV, NAV. What is NAV? Can you explain to the audience what does your organization do?

FRODE SUNDBY: [LAUGHS] Yeah, that makes sense to actually clarify. NAV is the Norwegian Welfare Administration, I think would be the English term for it.

JOHNNY HORVI: Labour and Welfare Administration.

FRODE SUNDBY: Labour and Welfare, yeah. So for anyone outside of Scandinavia, I think the concept is perhaps a little bit foreign. But we've got one central government agency which is responsible for every service that the government hands out to the public from their born, where you get birth money and you get [NON-ENGLISH SPEECH]-- I don't quite know what that would be in English, but where you get a monthly salary from the government for having a baby in practice.

ABDEL SGHIOUAR: Like a parental allowance, essentially, or a parental pay.



JOHNNY HORVI: Sickness benefits, pension and retirement benefits, helping people back to work if you're unemployed, and unemployment benefits.

FRODE SUNDBY: All the way to the pension when you are retired. All the pensions in Norway are paid out by NAV as well. So a third of the national budget in Norway is funneled through NAV going out, again, into paying back to the public.

ABDEL SGHIOUAR: I think that this is pretty unique to Norway. I think most other countries, they pretty much don't have anything like this, like public pension or like parental allowance or parental money, or it's handled typically by different institutions. But Norway is a relatively small country, so having one entity doing this makes, I guess, a lot of sense, right?

JOHNNY HORVI: As we can view, NAV is kind of the-- trying to oil the Norwegian machinery and support Norwegian citizens throughout life in their different phases. It's like copilot but for life.



ABDEL SGHIOUAR: I like that comparison. So I was telling you earlier when we were preparing for this that I have been-- in 2019, I attended the meetup in Oslo where there was public institutions like NAV and institutions similar to NAV, like the Tax Agency, et cetera, together in the same room, presenting what they have been doing in the last few months, what they have learned, what pain points are they facing. And all of them are using different cloud platforms, and they seem to be very independent from each other. But it sounded like a forum for sharing knowledge and experience.

And what was unique about it is that everybody who was present in the room was a public institution. I think there was also-- police were there. I don't remember who were the other ones. But is this a common thing across those public institutions in Norway? Is it, like, cultural? What's the story there?

FRODE SUNDBY: It is a common thing now. It was unheard of in 2016 when we started with it. I think it was NAV and Tax administration who started this public forum. And it makes sense because all of the public agencies, they are trying to solve pretty much the same problem. And we have the same challenges. And we most often find the similar solutions to the same problems.

So it makes sense that we should talk together and share the knowledge that we've learnt because we are paid by the same government. And it's the same government who is spending money in 40 different public agencies solving the same problem. So it's up to us to make it easier, save money, help each other.


JOHNNY HORVI: I would say that it's quite a lot of different solutions that springs out of this. What we see is that, like you said, we are solving basically the same problem, but we come to different conclusions. And it's very nice to see what other people are doing and what kind of directions they are moving their platforms, and so it's very helpful.

ABDEL SGHIOUAR: Is open-sourcing tools like NAIS, for example, also coming across other agencies, or you are like the only one who have a similar thing?

JOHNNY HORVI: I think some of the other agencies have things that they open source. NAIS is basically just a collection of a lot of different functionality packaged as maybe a Kubernetes operator or some internal APIs and so forth, so it's not like one meaningful product in and of itself. It's a lot of different services that is stitched together and provide more like an experience, I would say. We try to be open about different operators and stuff that we create. As a standalone thing, it's not necessarily super meaningful and it's to consumers, basically.

FRODE SUNDBY: But we still want to do everything we build. We want to build it in the open because, as we said, we are paid by the Norwegian government, which in turn is paid for by the Norwegian people. So the things that we build, do it in the open. We're not trying to hide what we're doing. This is what we are doing with the Norwegian money.

ABDEL SGHIOUAR: Yeah, I think that's very unique to Scandinavia in general but I guess to particularly in Norway and NAV. I remember having a very good experience from that meetup. The next question is, what's going on with NAIS? What does the future look like for you?

JOHNNY HORVI: So it's quite a lot of things happening right now. The current primary focus is based around cost and providing our developers with insight as to how what their workloads are driving, of course, how much is the compute, how much is the database and the buckets that you're running and try to surface this in a good way to the developers. We do this through the NAIS console that Frode mentioned earlier.

One thing is to provide the awareness of the actual costs because it's not free, and also in terms of insight into how good their utilization is. Because in the application spec, you define how much resources does my application need or you think it needs, but this is kind of a hard problem because it's not easy to know a prompt and it varies. And trying to provide, at least as a first step, the teams with tools in order to see how well it utilizes the resources that they've allocated is our start there so that they can see that. Yeah, I can probably use 1/10 of what I've actually requested, for instance.

FRODE SUNDBY: Yeah, that's the cost aspect of the console. We also want to use the console to improve the developer experience. As it is now, we've been given a lot the functionality for free. We've built operators saying that, OK, just give us an application specification and we will make the Kubernetes resources for you.

But then day two operation, you will have to use kubectl and you would have to go into the cluster and suddenly, you as a developer need to understand a lot of how Kubernetes works. Oh, suddenly there's a deployment here. What is a deployment? What is a pod? How do I check the logs? Ah, OK, so there's suddenly this whole new landscape which they need to figure out.

And we've got around 400 developers. And giving 400 developers kubectl and teaching them Kubernetes. Yes, we could do that and we have been doing that, but they don't necessarily need to understand all of this because we can take the relevant information for them and just display it in the console. We're not taking away kubectl from the developers, but we're just taking the relevant parts and making it even more accessible so that they don't have to.

ABDEL SGHIOUAR: Yeah, just making it easier basically for people to figure out what's going on.

FRODE SUNDBY: Yep, absolutely.

JOHNNY HORVI: Try to fade the implementation details into the background, allowing GCP and Kubernetes to fade a bit in the background. Yeah, wipe that.


FRODE SUNDBY: Yeah, just letting them because it turns out what they care most about is their application and not where the application runs, so, OK, let's give them what they need then. This is application-focused for them. Here is exactly what you are interested in about your application.

ABDEL SGHIOUAR: Nice. Any plans to have NAIS being part of the CNCF ecosystem, or is it not even like an ongoing discussion?

JOHNNY HORVI: Not really a discussion we've had. Like I mentioned, it's like a collection of different operators and tools that is sewn together to provide a certain experience for our developers. And I'm not sure how it would kind of fit into the CNCF. I'm not sure if that makes sense in the collection of the tools.

FRODE SUNDBY: It's not exactly a one-click install, nice, you've got your own platform running.

ABDEL SGHIOUAR: I see. I mean, you'd be surprised how many projects that are part of the CNCF are. Because it feels to me like NAIS is actually not a specialized tool because it has multiple functionalities. And the whole ecosystem of CNCF is full of tools that does one thing. So I guess even if NAIS is not really, as you are describing it, like a one product, it's a combination of multiple products. If it's presented to people as further described like the golden path and according to golden path, what are the tools you need, that could actually be a very cool, in my opinion, developer story just the way I see it.


JOHNNY HORVI: Yeah, it's entirely possible to maybe imagine a way that it could be stitched more together and be provided as a whole product.

FRODE SUNDBY: As you said, NAV is a general tool, and that is actually quite true. It's a general tool for solving general problems. And when we started building this in 2016, we saw that, yeah, we are trying to solve the problems of the developers at NAV. And being then in the public platform forums, we've seen that all the other public agencies are trying to solve the same problems using the same-ish tools.

So in, I think it was 2021 perhaps, we saw that, OK, what we've built here is not unique to NAV. It can work for everyone else as well. So this is when we started to say that, OK, here is NAIS. We can provide it for other public agencies as well. We built some tools to make sure that we could set it up in other public agencies Google organization and just terraform it up and they've got their own NAIS asset service.

JOHNNY HORVI: Yeah, I was actually looking what you were talking of looking at the website. I like the URL. It's naas.nais.io, which I guess stands for NAIS as a service.

FRODE SUNDBY: Yeah, that's great.

ABDEL SGHIOUAR: [LAUGHS] So, yeah, it might be a little bit confusing in this whole conversation because we're sometimes saying NAIS too much on the platform or just nice to say nice things. So I think people will have to figure out which one we're mentioning.

But for the reference, the website is NAIS, nais.io. The documentation is there. You can check it out. I looked at the quickstart guide. It looked pretty straightforward. So if you are listening to this episode and you want to try it out, knock yourself down, I guess. All right, well, thank you very much, Johnny and Frode.

FRODE SUNDBY: Yeah, thank you.

JOHNNY HORVI: Thank you.

ABDEL SGHIOUAR: Thank you for the time. I appreciate you talking to me. And I hope you enjoyed the episode. And until next time.


KASLIN FIELDS: Thank you for that interview, Abdel. That was really interesting.

ABDEL SGHIOUAR: Yeah, it was indeed. I got to know these people when-- actually, I don't know if you remember this, but the Twitter account of NAIS reached out to our Twitter account when we got the show.

KASLIN FIELDS: Oh, yeah, yeah.



ABDEL SGHIOUAR: And they were like, hey, it would be awesome to come on the show and talk about what we do.


ABDEL SGHIOUAR: And it only took us a year to do it.

KASLIN FIELDS: It only took us a year. We got around to it eventually. We're figuring stuff out.


KASLIN FIELDS: They have such a great story, though. I feel like it's like what I want most stories to be generally in this space where, like, we have a problem and we figured out how we're going to solve it with Kubernetes, and our focus is on making developers' lives easier, and they just did that. And then I think it makes it really cool that they have all of these different government agencies and they've created this one solution that solves the 80% problem, like you all were saying. And then just all of the government agencies in Norway seem to just use it now. That's crazy to me.

ABDEL SGHIOUAR: Yes. And they are using public cloud and they are doing everything in the public as a government agency, which is like an extra level of cool, I guess.

KASLIN FIELDS: Yeah. So many problems with that for folks who work in other government agencies, I'm sure. I'm sure a lot of them are jealous right now. [LAUGHS]

ABDEL SGHIOUAR: Yeah. I guess also during the question about "what does NAV do," I don't remember which one of-- it was Johnny or Frode who was talking about the fact that it is a welfare agency sounds like a deranged kind of concept. But then it's that and then they're doing stuff in the cloud is like an extra layer of, what's going on.


ABDEL SGHIOUAR: But, yeah, it was pretty interesting to see like, as you said, they came from the JBoss and web servers old-school way and then they straight into Kubernetes after meeting Kelsey. And then they realized-- I guess one of the-- part of the conversation was, like, most of the developers need the same thing. Like, no one has unique use cases. Everybody pretty much just needs to run an app and write data somewhere.

KASLIN FIELDS: Yeah. And something that I really loved about your conversation is that I feel like it speaks to one of the core tenants of security, actually, which is that-- I mean, essentially, everything in security is trying to keep people from doing things they shouldn't be doing, which you do by making those things as hard as you possibly can, basically. Right?

ABDEL SGHIOUAR: Or the other way of looking at it is abstracting away the complexity from people.

KASLIN FIELDS: For the positive path, for the path that you want them to go on.


KASLIN FIELDS: Which is what they're doing here is they're taking that concept and applying it to say, well, we want people to do this, we made it as easy as possible for them to do it and that's what they do. So that's a really good story.

ABDEL SGHIOUAR: Yeah, it's pretty cool. Yeah, and then we met again-- so the story of how this whole interview came to be is they reached out to us on Twitter. They were somewhere in our backlog. We didn't really get time to get to it. And it took me all the way until I went to JavaZone, which is a conference in Norway in September last year, and I met the people from NAV because they were a sponsor for that conference. And then I saw people wearing these T-shirts that says "NAV," and I was like, oh, yeah, I heard about you guys, what's going on?

KASLIN FIELDS: Yeah, let's talk. [LAUGHS]

ABDEL SGHIOUAR: Let's talk, yeah. And then we started chatting and then I realized, oh, we actually do have a relationship with them because they're a customer. And, I mean, that's not the reason why we brought them on the show just because they happen to be a customer. It doesn't really matter. And then I ended up talking to these two folks, which was cool.

KASLIN FIELDS: Yeah. And I'm excited to do more of that hopefully here in 2024. Looking forward to conferences already. So hoping that we can meet a lot of you out there and you can come up to us and tell us about your story so that we can learn more.

ABDEL SGHIOUAR: Yeah, that's actually a good call for action. If you are doing something cool with Kubernetes, come talk to us.

KASLIN FIELDS: Yeah, especially KubeCon EU, we're doing all sorts of gearing up for that right now. So if you're going to be there, make sure you come and find us. We'll be around.

ABDEL SGHIOUAR: Yeah, we have stickers, by the way.

KASLIN FIELDS: Yeah, we have stickers.

ABDEL SGHIOUAR: We have, like, 3,000 of them.

KASLIN FIELDS: We have so many stickers.


ABDEL SGHIOUAR: That'll be nice.



KASLIN FIELDS: Yeah. So one other thing I guess I'll mention about this is that one thing you all talked about is that it's basically just a collection of operators is what NAIS is.

ABDEL SGHIOUAR: Pretty much.

KASLIN FIELDS: And so it's not like one thing that you just deploy, that was like a whole conversation that you all had. But with the whole story that they told about how this makes their processes so much easier, I'm really curious about how they do the database bit, like, how they integrate their operators into the whole workflow and everything. So I'm probably going to go look around at those operators a little bit more. So if you are interested in joining me in looking at those operators, we'll have some links in the show notes to all of the resources.

ABDEL SGHIOUAR: Yes, we'll make sure to include all of the resources there.

KASLIN FIELDS: Awesome. Thank you so much, Abdel.



KASLIN FIELDS: That brings us to the end of another episode. If you enjoyed this show, please help us spread the word and tell a friend. If you have any feedback for us, you can find us on social media at KubernetesPod, or reach us by email <Kubernetespodcast@google.com>.

You can also check out the website at Kubernetespodcast.com, where you'll find transcripts, show notes, and links to subscribe. Please consider rating us in your podcast player so we can help more people find and enjoy the show. Thanks for listening, and we'll see you next time.