#039 – Banking on Kubernetes: From Fintech Frontier to Regulatory Reality with Kasper Nissen (Dash0)

Itiel Shwartz: Hello everyone and welcome to another episodes of the Kubernetes for humans podcast. My name is Itiel Shwartz and today with me in the show we have Kasper. Kasper, happy to meet.

Kasper Nissen: Happy to be here. Thank you for the invite.

Itiel Shwartz: Ah like sure sure and like let’s like let’s start by you telling us a bit about yourself like who are you what do you do and what’s like your favorite Kubernetes project.

Kasper Nissen: All right. yeah so so my name is Kasper. I’m currently in in a developer advocate role at Dash0. I recently joined like a month Argo. So I before that I was uh at a company called Lunarr which is a Nordic challenger bank. spent almost eight years there building platforms uh on Kubernetes um you know just going with the the sort of the evolution of the CNCF and all the projects coming out and then us adopting it right after almost and uh basically building a bank on CNCF projects which was uh very interesting also on the bleeding edge uh sometimes it hurts a little bit um but uh yeah so so that’s sort of the really really short introduction to what I do as uh in in my day job. I besides all of this, I’ve been very much involved in the in the community. I’m a CNCF ambassador. I’m currently a co-chair for KubeCon. KubeCon London will be my last as as co-chair. So, uh looking forward to

Itiel Shwartz: Well, you get you get free uh free KubeCons as a as a co-chair.

Kasper Nissen: and I already did Paris and Salt Lake City. So, this will be my my last um it’s also a lot of work. So, it’s it’s okay. but it’s it’s really fun and rewarding uh to to you know to get all the insights into the community and uh yeah besides this I’m also organizing local events KCDs here in Denmark and um founded Cloud Native Nordics a meetup alliance across the Nordics and uh last thing I want to mention is that I’m trying to reach the Golden Kubestronaut. I’m currently um out for exams. So, uh hopefully that will be uh be done by by KubeCon. But let’s see.

Itiel Shwartz: Oh, a lot of things. A lot of things like a lot to pick from. Maybe maybe we’ll go to like uh the bank, you know, like you said both bank and bleeding edge, which usually are like, you know, it’s not like the first thing when I think about a bank technology. So, if you can first of all, like what led you to work in the bank? like I’ll be honest like I don’t really know almost anything about that. So if you can share like why did you go there? Why did you chose to be that bleeding edge and you know share a bit about the story and what happened there.

Kasper Nissen: Yeah so um as as mentioned the company called was called is or is called lunar they still existing today. I joined the team back in 2016. more or less uh I spent like 5 months uh as a consultant after my uh my master thesis which I also wrote about Kubernetes and then basically they needed someone to um to rip apart their infrastructure. They were starting to do microservices um and they were sort of missing the infrastructure part a little bit. they were doing like manual orchestration through some Jenkins pipeline and stuff uh with you know terraform and all kinds of with stuff that were was basically building an environment and that caused a lot of issues between teams because now they were scaling and microservices were being added all the time.

Itiel Shwartz: what was the size of the company and like what per like experience did you brought like you know when you joined like so if you can

Kasper Nissen: yeah so so when I joined there were about 25 or something like that and um three or four four four development teams as far as I remember um but there were at at that scale already starting to see sort of the issues between the different teams of because when you bake everything into one big go and you release everything at one like in a single point of time then you have a lot dependencies between teams on are you ready to or is this master branch is is that ready to go out. so there was a lot of coordination and one of the goals of my employment was really to rip that apart and make sure that the teams were able to basically independently deploy their services into uh to some kind of platform and that platform then ended up becoming a Kubernetes. yeah.

Itiel Shwartz: Okay. So you know you mentioned that you guys use quite a lot of like open source CNCF projects maybe a bit about like the philosophy of Lunar or like you know like I’m I’m talking with quite a lot of banks some very big banks are also like commodore customers and again it’s not like the bank thing to do to go bleeding edge new technologies and so on like it’s like that’s not that’s not how things are working. So maybe share like why and how did the management got like so you know like it can’t be easy. So yeah tell us a bit.

Kasper Nissen: So yes so so we were a challenger bank or like a a new bank as as it was also called at that time. So we were very much so the initial philosophy was really that we you shouldn’t be a bank. we should leverage other banks in the different countries that we wanted to be present in and then really scale like to like global scale like like the big vision area take over the world more or less. but but and and that basically that that required us to to be able to scale quite rapidly and Kubernetes was also like one of the reasons for or this was also one of the reasons for for Kubernetes as that would allow us to scale quite fast uh while you know taking over the world. so so we were in the first couple of years we wasn’t we wasn’t a bank. we were just a fintech startup trying to do banking stuff. so everything was like from a a technical perspective. And then in I think it was 19 we uh we got a banking license. and actually became like a real certified bank and all that. which also changed the requirement quite a lot. uh that that really put a lot of additional work on us uh in ensuring that the infrastructure was secure, compliant, we were following all the rules also for how we were you know delivering software into production for for our developers uh required you know for principles and all kinds of uh different compliance rules that needed to be implemented and and in place. So we were already building on on this platform from like the fintech you know time to to then becoming a bank and then we needed to figure out how to actually make you know Kubernetes uh work in a in a banking environment in yeah in in 2019 and also at this time we were starting to ramp up um very much in the Nordics. we were going we changed sort of the philosophy of the company from from being this global uh scale to actually focus on diving deep into the Nordic countries of of Europe in Scandinavia because the infrastructure is is rather complex and uh it’s it’s stitched together of all kinds of different vendors. So going deep was a better business case for Lunarr um instead of you know trying to to build a bank in or build like an API layer a nice user experience on top of something existing. so that’s really what we we tried to do there. and that worked out pretty well to be honest. Kubernetes is a it’s a great platform for for building these kinds of things and and the community also embraced this and and I also other banks especially not as much in in in Denmark at least but in in the rest of uh of Europe and we saw companies like Monzo and and other banks doing similar things. So we knew that this was really uh this was the right way of of doing this. It really gave us an edge compared to like the competitors that we were going up against in the Nordic countries which were built on like old really really old technology and really old compliance rules as well. We’re also challenging how what what is actually necessary. How how are we how are we actually going to do this? We we need to rethink how we do banking.

Itiel Shwartz: So we you know like when when I think about like a bank JP Morgan whatever it’s both security compliance and stability I know like that like those will be like the first like three three things that come to mind. Kubernetes like five years Argo wasn’t that good in each one of those pillars like it wasn’t that secured. It wasn’t that like I will say like robust or a real production ready for banks. So like maybe if you can share some you know like a bad story and a good story from your time D like what didn’t work out to you for you guys in in Kubernetes or you had to like reinvent right like because it was like very bleeding edge.

Kasper Nissen: well of course initially we spend a lot of time in in trying to secure everything building everything and like so we were also cloud-based which was also completely new in in the Nordics at least that uh we were actually building a bank in the cloud. so we did a lot of work uh together with some of the vendors cloud vendors the providers that we were using to actually get some some regulations changed. But in terms of Kubernetes, I think that we we actually spent quite a lot of time in understanding like the security model, all the different ways of configuring following the best practices of uh the CIS Kubernetes Benchmark and and all of that. and and it was it took quite a lot of time uh to be honest to to actually you know first of all to get the knowledge and then secondly actually implement that knowledge uh into the platform. I don’t think I actually saw something where we wasn’t able to use Kubernetes or had a like a case where it didn’t really make sense. we were very much believing in Kubernetes as the foundation of of everything that we were building on. but it took quite a lot of effort to actually make sure that everything is uh is compliant and and of course also having like external pen tests and and all that done just to make sure that everything is uh is following best practices

Itiel Shwartz: like what what did you do from like a stateful set like perspective like databases and so on was it also on top of Kubernetes or like AWS or something else yeah that was actually one of the places where we didn’t use Kubernetes. We were really well for some for some uh for some services like our monitoring solutions we were using Kubernetes. but we really had a goal of keeping our production clusters uh as stateless as possible. so databases uh we were on AWS we were using RDS for for that and and really trying to push all the stateful uh stuff to to whatever uh managed solution we could find because we were also like a fairly small team this uh at this let’s say in 19 when we started to scale up we were three people in the platform team um so we didn’t really have like the full resources to go and manage databases and all that in in a cluster ourselves. So um so you know buy stuff where it makes sense to buy stuff um and then you know bid on on other stuff where you think that that is what will give you an edge going forward.

Itiel Shwartz: Okay. Well that that makes total total sense. so like you were there for quite a lot of time right like uh in the end of the day.

Kasper Nissen: Yes. Yeah.

Itiel Shwartz: What you know even it’s like the Kubernetes for Humans podcast what other like CNCF project were you guys heavily using or you like specifically really liked because you know there are like hundreds of projects right oh yes so what if you can you know share a bit on like what what project you were like like hardcore like adopter of

Kasper Nissen: yeah so so basically yeah Kubernetes was definitely the first one we took in uh secondly was uh Prometheus uh Prometheus has been was was really a great tool for monitoring uh in early days uh still is. but I right now uh as I’m also wearing an OpenTelemetry t-shirt, I really think that sort of the next wave of uh observability is like not looking at observability as these three different things like having logs, having metrics, having traces in in separate tools but but actually combining them to to get valuable information uh and and be able to correlate between these different signals. I think that is uh at least right now also it also makes sense in my current role at Dash0 an observability platform. but I really see some similarities between what we saw early days with Kubernetes. Kubernetes was really a platform for building platforms, right? I remember that quote from Brian L from San Diego KubeCon. uh that that really resonates a lot with me um because that was what we were doing at Lunar at that time building everything on Kubernetes and building a platform that was sort of tailored to to Lunar’s needs for for running software and compliance and security but we also I think that OpenTelemetry is like a project that is also for building platforms now now we’re just talking about like this is a metric collections platform but there’s so much to be gained from from that correlation and and and that insight that now is possible uh for for vendors to to to basically build new cool things on top of for that hopefully in the end will be a big benefit to all developers uh and platform engineers and and whoever else needs this information right that we get better insight into uh to what is what is basically going on in the systems that we are responsible for for running every day. we saw similar uh issues with with Lunarr. We we were we were using Prometheus. Prometheus was a great project. We we got some some great signals from that ride. But we we struggled with tracing. Tracing was hard for us. I think it was many struggled when when figuring out what was the actual value of of tracing where where do I need it? We did have it have like Jaeger running for a couple of years but didn’t really see developers actually like picking it up using it for something. and then logs. we were very much focused on logs. because that is really what developers is is is familiar with. It’s it’s they write like a log info and they see that log line come out in the uh whatever log management tool you have. and and we really doubled down on on implementing a really really good solution for logging for our developers at Lunarr because that we really wanted them to to take on this responsibility of of running these things in production and and providing them an interface that would allow them to to do that very easily was was very important to us. So we actually bought a product back then uh that focused a lot on on on that. yeah so I think that that was just some of the projects. OpenTelemetry is definitely one of the projects I would like to highlight right now. but also earlier projects that I’ve been very much active involved with especially in my Lunar days has been Linkerd as a service mesh love the simplicity love the Linux phil philosophy flux more or less the same thing also the same philosophy. yeah, so many great projects to choose from. and and and now I also think that a project like Perses for dashboarding going forward is is something that’s going to be like a game change game changer instead of having like all these different vendors inventing their own like dashboarding formats. Now we can actually standardize on how what what is a dashboard, what what kind of panels exist. so I really see some value in in in that going forward as well and and basically building on these standards and then have like vendors compete in in like a new area instead of uh the old way of of competing, right? So now we’re building on these standards that allow us to do even more things um and make it easier for for the consumers of these products to uh to now they only have to have tele own telemetry running. They need to understand what a Perses dashboard look like. they can you know commit that using GitOps and flux and Argo whatever and then uh get insight from from whatever vendor solution they they want in the end.

Itiel Shwartz: Okay. Sounds sounds very interesting but you know share a bit about you know you talked about the Dash0 you left and it became a de so like what happened why did it happen we also had Miracle here as a guest at the show but if you can do like talk a bit about like what do you guys do and like why you particular like like did this transformation right.

Kasper Nissen: Sure. yeah, so as mentioned earlier, um during my time at Lunarr, I spent a lot of time in in communities. I really enjoyed sharing experiences of what we were doing at Lunarr. So I’ve been doing a lot of the different conference talks, meetups, and and stuff like that. And that’s always been one of the things that I I knew that when when I needed to to find something new to do, I wanted to try this out full-time being like a DevRel kind of role. so so that was sort of the first thing. So so then why why desk zero? um as as mentioned I really see these similar similarities between the platform of Kubernetes and and then OpenTelemetry as sort of the next wave of uh of observability and and I really really think that is going to be like a game changer for for companies going forward that they don’t need to have all of these vendor like agents running in the clusters and all of that that we can like sort of start from from the beginning at OpenTelemetry there we are all agreeing like this is the standard we and get all the sprawl away with you know different naming of fields and stuff like that the semantic conventions and and all that. So, so we are sort of like raising the bar to to a new standard and and I think that uh that Dash0 is on the right track in this space because they they are building like an OpenTelemetry native platform that are util utilizing all of these uh open source projects and open standards to to build like the next uh generation of of of observability systems, right? But we are actually able to deep dive and correlate uh the different signals that we get. and and then again I really saw this big change um in the log management system that we took in back in the day that having like a developer focused a developer centric tool is something that really can bring some value and I think that is really where Dash0 is is also doubling down on is really providing like the easy experience for developers. it’s very also very much relates to like platform engineering, developer experience. All that we’ve been building these internal platforms has been about how do we make the life easier for developers? How do we provide like the right signals to them? so I really think they on a a the right track in in that space and and I really want to be part of that. So that’s sort of the reason why I I transitioned into Devrell and in the observability space. But I really see some some big potential in the space.

Itiel Shwartz: So, so like why why DevRel like you could done like you know things that are a bit more like hands-on right so why doing like this transition as well it’s like a double transition you did right yeah well I I really enjoy learning and and one of the things that I saw in my in in a role as de is also playing around with different technology uh learning all the different projects try you know pick pick a project from from a shelf take it down figure out how do I monitor that how do how do I get the most value out of this product and how can I or this project and how can I make it available to developers and make it as easy for developers to to consume. and also in in in this role is also very much about like uh getting feedback, getting insights, trying to you know help shape the product uh so that it fits whatever I hear from from the community as well, right? So I think that’s uh very very interesting uh to me. I before being like a platform engineer, I also spent some time building iOS apps. So u web web apps as well. I kind of like that product. uh the user experience that that feel and and that understanding as well is something that I haven’t been been doing much being deep dive into Linux and Kubernetes for the past eight years but it’s also it’s always something that’s been very interesting to me as well and I sort of like get a lot of that as well um yeah and uh I just needed to try it out um especially also given my role in the community I think this is a is a very interesting place to be And I want to be able to uh you know to get more involved with the different projects especially in the observability space um and help them as much as possible get time on like the big stage if possible I’ll I’ll be talking and giving a keynote at cubecon as well around OpenTelemetry observability where we going uh some of the bridging between platform engineering and observability I think the project is now providing like the kubernetes operator for OpenTelemetry why you get can get auto instrumentation and stuff like that is is very much again back to this making it easy for developers as possible so they don’t have to you know understand what is underneath this thing um yeah so I’m not sure if that answered the question but uh it did and you know like you are so recruited inside the community then I must ask you like where are we going right as a community inside the CNCF like like share a bit both predict prediction maybe on like technologies but also from like a cultural standpoint like the CNCF is like booming right like uh in the last five years I don’t know like I don’t have the numbers but I feel it’s like exponentially growing so if you can share a bit about the community aspects of the CNCF but also the technologies like where are we going as a like industry yeah I I really love this community it’s it’s growing and and I think we breaking records all the time for KubeCon attendance as well and and I think we still see like a lot of newcomers join like conferences like KubeCon and I think it’s it’s it’s actually above 50% all the time uh so far at least of new attendees or first-time attendees at a at a KubeCon um which is telling us at least that there’s still a lot of people that are joining this uh like rocket ship uh whatever we call it so I I’m only seeing this community grow much much bigger. people are still, you know, trying to figure out how to to get on this rocket ship and and by I think for many of the projects and and for us as a community, there’s still work to do in in lowering that bar and make it as easy as possible to to get on the rocket ship. I think that’s definitely one of the things that we will we will see going forward. I’ve I’ve a project like backstage uh I see some value in in many of the projects making it easy for so if if you as a newcomer to the space you you want the UI you want everything up and running but but providing like plugins for all projects would would sort of lower that bar for for many people to to start utilizing uh some of the projects. So I think there’s some some really interesting perspectives in figuring out how we as a community also uh lower the bar, make it as easy as possible like like a may more or less a one-click experience in in some cases to to get new people on boarded. and I guess we can’t talk about the future without talking about AI as well. AI has been like a like a big topic on at least on the last three or four KubeCons. yeah, we’re definitely going to see more in that space. I think we will be seeing AI being used in many of the different projects to some extent. we will see it like getting embedded into all kinds of things. but we also see again figuring the community is also figuring out how how we’re actually building like these platforms that would that allows developers to utilize these features as again as a company. That was actually one of the things I discussed in my keynote in in Salt Lake City for KubeCon was how do we actually bridge the gap between AI and platform engineering and and make it as as easy for developers to consume AI as possible but but still in like a secure and regulated way uh a compliant way. but there’s oh there’s so many so many topics I think observability is gonna going to have it its next wave now. um AI. Yeah, I have to say ability AI but but also just you know tailoring Kubernetes to to to these new kind of workloads has also been like a big big topic. developer experience uh we just added the App Development track to KubeCon this time. I think we’re going to hopefully going to see some more uh projects that will focus on the easiness of use for developers as well. um providing the right insights uh at the right time uh like integration into IDEs could also be from an observability perspective as well getting signals directly into where the source code was written and where the issue might be. there’s all kinds of interesting uh perspectives and and things will that will happen in in sort of the next let’s say 10 years now. Now we are celebrating 10 years of last year we celebrated 10 years of Kubernetes and now we’re celebrating 10 years of CNCF right in this year. So, it’s that’s going to be a lot of exciting things happening within the next 10 years as well.

Itiel Shwartz: Okay. Super cool. Any last remarks or like things that you want to to say to our audience?

Kasper Nissen: No, I I really if you’re not part of the community yet, I advise you to strongly join. Go to KubeCon, talk to people. People are super friendly. just, you know, be there, be part of it. uh try try and yeah make some new friends. It’s a it’s it’s an awesome community and uh that’s a lot of opportunities in in the space still.

Itiel Shwartz: Okay. Thank you very much uh Kasper. Pleasure having you and see you at KubeCon London where you will be like your last KubeCon as a co-chair.

Kasper Nissen: Yeah. Thank you for having me. Bye-bye.

[Music] Kubernetes for Humans.

This is an AI generated transcript of the conversation