Next in Tech | Ep. 214: KubeCon and Cloud Native

View Full Transcript

Eric Hanselman: Welcome to Next in Tech and S&P Global Market Intelligence Podcast where the world of emerging tech lives. I'm your host, Eric Hanselman, Chief Analyst for Industry and Company Data at S&P Global Market Intelligence. And today, we're going to be digging into all things cloud native. We'll be looking at what we saw at KubeCon and Cloud-NativeCon just closed up in London. And with me to discuss it are Jean Atelsek and William Fellows. Welcome back to the podcast to you both. 
 

William Fellows: Thank you, Eric. Thanks for having us back. 
 

Eric Hanselman: It's great to have you on. And we wanted to figure out what was interesting. There was an awful lot that was going on there from open-source capabilities and some interesting revelations. So, what was top of your list? 
 

William Fellows: Well, I guess, firstly, the setup now, the Linux Foundation is sort of expanding its remit and reaching to a number of other places across the industry. So, the context of KubeCon plus Cloud-NativeCon in London this year was situated in a couple of places with the addition at the front end of the FinOps Foundation Day in London and FinOps Foundation like CNCF and KubeCon is part of the Linux Foundation. And indeed, within days of the event as well, we also heard about, obviously, the Open Infra organization, what used to be OpenStack joining the Linux Foundation as well. And that was part of the show if you like. So, we can talk about that as well. I think the relevance of the FinOps Foundation event being the kind of front book end to CBC and cloud native C was planful and designed to really surface how optimization and especially cost is a first-class citizen when it comes to all things cloud native. And indeed, my contention would be that as part of any kind of developer-oriented stack, whether it's DevOps or platform engineering or SRE or anything else that, that optimization and cost visibility and management piece is now a part of that. So, I think that sent an important message to the audience in general. But the event itself, i.e., Con and cloud native Con in London for the return to Europe, the headline numbers on that were that there were, I think, 13,000 attendees, which was slightly up on Paris last year. And of course, it's always attracts a greater number of attendees because of the bigger catchment area than just the North American events for comparison. The most striking thing, though, was not that number itself, but the growth in the development of the "Day 0 event and program series, which is effectively the first day of Cuban event. And with 6,500 participants at that event, I mean, it really has become an event in itself. It was only a few years ago that it was much more of a forensic and nerdy set of events that were in different places at different times. And now it's become a pretty joined-up thing. Fully half of the attendees at the whole event participate in this, which I think is a testament to the maturity of cloud native. And well, Kubernetes is now entering its teenage years this year and to the whole practice and discipline of cloud-native development. Another benchmark by which we could sort of look at this is that there are now far fewer hoodies and died blue and purple hair than they used to be and more sports jacket, should we say? 
 

Eric Hanselman: So, we've got the hood to sports jacket ratio to contend with. But I think to your point, the other maturity piece, the fact that we've now got that next stage of maturity in infrastructure, which is a focus on cost. And so, the FinOps section of this, interesting to see that the Linux Foundation is expanding that. But maybe let our audience know what you're seeing at the FinOps side of this. Again, that's that piece where now it's widely enough adopted. It's starting to consume a large enough chunk of IT budgets that people are concerned about how much things cost. 
 

William Fellows: Sure. So well, to summarize quickly, the FinOps Foundation has overachieved, let's say, in terms of where it had anticipated it would be and it has really been catapulted to front and center with driven by enterprise interest in being able to see their cloud spend, first of all, and then bring it under control because this is not really about reclaiming money to go and put back in the bank. It's really about freeing up economic resources to then spend on additional services. This is not the hyperscalers losing out because people are choosing to find ways to spend less. I mean it's just more reason for them to spend more on more stuff, I suppose. But putting some guardrails into place about how that happens over what period of time, at what pace who's involved and so on. The FinOps Foundation already has its framework now for the discipline of FinOps in place, which is available. It also has also produced what it calls its focus specification, which is a way for the cloud billing data from the major hyperscalers to be normalized in such a way that it can be consumed passed, analyzed and so forth in a consistent format. So instead of having to deal with each of those separate and different billing formats. And now FinOps Foundation is reaching into and for other areas of IT and technology spend. And the next iteration of focus, for example, will include management when it comes to SaaS spending. And it's also encouraging organizations to try and define or bring FinOps discipline to other areas such as on-premises equipment to data centers, to license management, to sustainability, to AI and so on. And these are not things that the foundation itself is going to specify or indeed recommend more it's about organizations deciding how they can use the FinOps disciplines that they've learned from the foundation in other areas. But obviously, the foundation is hoping to get its arms around a bigger amount or to have a bigger universe of spend that its discipline is used over. And the subtext here is that it is and will increasingly but up against the aspirations of the more incumbent approaches that are collectively thought about as technology business management or IT financial management, where there are obviously other legacy players. So, it's going to be interesting to see how these sort of worlds come together, so to speak. So that covers off the FinOps foundation piece. We should get back to the actual event itself, and I give Gene a chance to maybe. 
 

Jean Atelsek: So, I'd say the most striking thing for me was that at recent cons, basically, participants were sort of highlighting the ability of Kubernetes, the -- not just the ability, but the suitability of Kubernetes for training AI models and the sort of compute-intensive pooling of resources that Kubernetes allows. And it has become a default for new applications, but what we saw this time was much more effort to highlight how it can help in modernizing brownfield applications. And so that showed in a lot more sort of edge to cloud to on-prem projects focusing on solving issues with respect to that. And then also the -- a lot of vendors spoke about kind of the VMware, the actions that Broadcom has taken with respect to VMware being a motivator for companies not just to move away from VMware, but also to modernize their environments and to use open source. The open-source paradigm has sort of proved itself as being capable of doing that, although there are still some concerns. 
 

Eric Hanselman: Well, it's the V movement catalyzed -- we talked about this on previous episodes. But just as a quick refresher, with Broadcom's acquisition of VMware, there has been what seems to be for most customers, a significant increase in licensing costs. So, there's a focus on looking at alternatives. And there was a very lively conversation at KubeCon about how open source was the path and the way. But to your point, there's that question of what's it going to take to actually make that migration happen. But certainly, a lot of hopeful conversations that the Cube community could be that path. So interested to see how that starts to shake out. 
 

Jean Atelsek: Yes. And it feels relatively low risk, I mean, for small projects and so on to sort of dip your toes in the open-source pool and see what's going on there versus some of these restrictive licensing regimes that are getting more restrictive from -- not just from VMware, but from other vendors as well. So... 
 

Eric Hanselman: The beauty of open source and of course, the foundation of all this, the flexibility of containers and everything that sort of sits around that ecosystem. I'm curious, you're talking also about some of the broader views about AI because, of course, we've got to talk about AI. And there also seem to be an awful lot of discussion about container environments being a great way to be able to marshal the computing resources necessary and aspects of that. That seemed to be a big part of what was going on there as well. 
 

Jean Atelsek: Yes, exactly. And one stumbling block for a lot of enterprises in particular, has been the security of open-source software. And the Linux Foundation, which is the parent of the host of the conference, the Cloud Native Computing Foundation, repeatedly is creating projects and tools to really reassure enterprises that they can trust open-source software that it's not -- there aren't going to be any back doors and so on. A couple of things that they did at this Con or that they mentioned at this CPO was the first-person project because they made a note that on a developer platform, you can be just a name and you could be a bit, you could be a person. And so, they're finding ways to basically authenticate developers behind the projects as real people so that enterprises will feel reassured by that. And also, again, with the kind of making things more enterprise friendly, they are looking at the -- they're taking a census of the most used open-source products and targeting those first for ensuring that they are secure, that the software bill of materials is clean and that enterprises can be comfortable deploying them in production? 
 

Eric Hanselman: Well, it's being able to get to a point at which you've got some assurance that the code is exactly what you expect it to be that was written by the people you expect. And I think back to the old cartoon, on the Internet, nobody knows you're a dog. 
 

Jean Atelsek: Exactly. 
 

Eric Hanselman: But there are real risks and in fact, have been real incidents of people joining open-source projects and contributing malicious code. So, they're clearly something that's a valuable aspect of this and their ongoing efforts to manage vulnerabilities and reflect back content with the software bill of materials pieces. 
 

William Fellows: One of the things that we did successfully road test at the event was the notion that there's any kind of a sense of the shine or the bloom being off platform engineering as a discipline and a concept because there's been a lot of discussion out there in the community. But with an entire day 0 track devoted to platform engineering in which there was really standing room only at all of the components. I mean I think it's safe to say that platform engineering is still healthy and growing. And of course, platform engineering is really about and it's very basic is about operations and development coming together really and collaborating around a sort of curated set of tools and processes. And obviously, there's a lot of devil in the detail in terms of how that happens. But the -- there were a wide range of large enterprise users who talked about their experience of doing this. I mean, the consensus seemed to be that you should be doing it. And from our point of view, in addition to obviously, Kubernetes and the latest cloud-native technology tooling, things such as Backstage, GitOps process, also a FinOps element on the requisite in a platform engineering approach, which in some way can be also conceived of as the kind of an enhanced version of DevOps, if you like, as well rather than a successor or a replacement. So, it was interesting that just the wealth of interest in platform engineering that seems to be now really driving with some momentum. 
 

Eric Hanselman: It seems like DevOps patterns were sort of that first stage. And then you had the proliferation of tools. And of course, for the CNCF, there is that iconic slide of all of the various projects that is just literally. 
 

Jean Atelsek: Yes, over 200. 
 

Eric Hanselman: Yes... 
 

Jean Atelsek: There are over 200 projects now. 
 

Eric Hanselman: It gets ever denser. -- so, platform engineering is that point at which we're trying to make some sense of at least for an individual company's environment, what does that complete set of tools look like? How do we start to get that to a point at which it's manageable and not something where every development team has its own development pipeline and hopefully has some standardization. But as you pointed out, also the -- that it's integrated with a set of tools like FinOps capabilities with observability with all of the open telemetry kinds of pieces so that we've got both the upfront pieces of cost management, the operational pieces with all the telemetry, but making sure that, that is one more cohesive whole, but still giving the agility that a lot of those original DevOps ideas really started with. 
 

Jean Atelsek: Right. And I would say the key challenge at the platform engineering sessions that I attended one of the key challenges is getting developers to actually use the platforms that the platform engineers give them. And that's a cultural change that is slowly happening. And one consequence of not having everyone on board this consistent governed platform is that configurations for Kubernetes environments and clusters can start to drift. Addressing an immediate issue may fix it, but then it's no longer aligned with the representation of the environment that the Kubernetes is trying to maintain. And that -- there are a bunch of risks to that. And one thing that we saw at this event was a lot of companies addressing that particular problem with what they're calling architectural observability or maybe it's an evolution of configuration management, but basically something that notifies the platform teams or actually prevents configurations from being tweaked out of alignment and really threatening the security and the resiliency of the environment. 
 

Eric Hanselman: Well, this is the thing that I found that was really interesting, and that's a fascinating point that you bring up, which is that the whole idea about containers was that they were ephemeral this was this whole concept of you were treating your workload, you're changing the way you're treating your workloads from treating them like pets where you groom them and cared for them and kept them for life to treating them like cattle that if they got sick and died, you just got another one. And the idea that you had -- you just simply spawn new containers. If you needed to change it, if you needed to do something else, you'd simply destroy the existing container, fire up a new one with all the appropriate fixes and off you go. Well, now we're drifting back into that old legacy thinking of, well, we'll just tweak it, we'll fix it, we'll change the configuration a little bit. And we've now got longer-lived containers, and we have drifted in many ways from the original design ideas. 
 

Jean Atelsek: Right. The exact sort of difficult to maintain applications that Kubernetes was supposed to prevent. 
 

Eric Hanselman: Old habits die hard. We get in these old patterns. We keep going back to all this stuff and doing all the bad things we used to do because, wow, it's kind of easy, but -- and I guess maybe that was one of the other things that I thought was interesting about this particular KubeCon is that it did seem -- it was well attended, but it did seem that there -- I heard a couple of comments that the conversations had shifted from being the vanguard of those folks who are really very focused, very -- had deep understanding of what the capabilities were and how to roll it out. We're passionate about it and we're advocates of this new way of building infrastructure. to now moving back to this broader enterprise and a set of folks who were not part of the Vanguard and now thinking that, hey, this is probably a pretty good thing but are also still working their way up the ladder in terms of understanding and operational maturity. 
 

Jean Atelsek: Right. And some of the projects that we heard about are dedicated to helping developers not have to learn the nitty-gritty of Kubernetes, giving them improving the user experience for developers building on Kubernetes as like an infrastructure platform. Yes. 
 

William Fellows: So -- and to your point, Eric, I think that part of the challenge of both the event, the organizer and that community in general is that it is -- it's not the new kid on the block anymore. And hence, my reference earlier to the fact that Kubernetes is now entering its teenage years. And one of the challenges is the extent to which it is able to bring that whole kind of AI front within the context of cloud native and whether it can land some of those things and develop its own processes and so on to accommodate that. And I think that's what we're sort of going through exactly at the moment and the sort of jury is out on that entirely, I think. 
 

Eric Hanselman: We'll have to see where this evolves to -- so any thoughts about what we should be looking for, keep kind of heading to Atlanta as we start getting towards the fall. What are your thoughts about where this is headed? Is this an overall maturing of approach or a broadening of the user base that's out there or maybe a little bit of both? 
 

William Fellows: Well, again, we touched on the V thing just slightly, but I mean, it seems to -- from what we understand, at least in the enterprise, most containers still run on VMs as well. So, there's a lot of old as well as the new, let's say. I can't see that there's going to be anything materially different between now and the Atlanta event apart from the overall maturing of some of these things. And well, hopefully, the embedding of more AI expertise and tooling into cloud-native constructs themselves. We obviously have heard about the progress of the MCP, the model context protocol. That will be something to look out for. And I know both yourself and Jan have also got some views on that as well, right? 
 

Eric Hanselman: We do indeed. And actually, I think we need to do a follow-up on that. We could dedicate a whole episode to it. 
 

Jean Atelsek: Probably --... 
 

Jean Atelsek: Yes, talk about the new kid on the block. I mean everyone is making integrations with the model context protocol, so... 
 

William Fellows: And we should say this is something that's developed by the large language model builder Anthropic that's designed to address the challenges of connecting large language models to the data itself and to the tools they need to operate. Some of these MCP servers are already available from cloud providers, ISVs and other market participants as well. But the interesting thing, I guess, for us is that it seems to have momentum. It's got some of the major players in the market are now behind it. So, we expect to see that sort of grow and accelerate and bottom over the next few months. And then beyond Atlanta, the Linux Foundation did actually give a pipeline for their next set of events. So KubeCon returns to Amsterdam in the EU event in spring 2026 before heading to L.A. in the fall. and then to Barcelona the following spring of '27. So, we'll be looking forward to all of those. 
 

Eric Hanselman: Wow, maturing in their planning process even. wow, this is great stuff. I guess we certainly need to double down on MCP and a lot of what's going on there, but we'll save that for a future episode. Thank you both for being back. 
 

William Fellows: Thanks for having us. 
 

Jean Atelsek: Yes. 
 

Eric Hanselman: That is, it for this episode of N. Thanks to our audience for staying with us, and thanks to our production team, including Sophie Carr, Feranmi Adeoshun, Kyra Smith and Adam Kovalsky on the marketing and events team. I hope you'll join us for our next episode where we're going to be going into a preview of the RSA Security Conference that will be coming up. I'll be talking to the team to see what they anticipate what will be happening, talking a little bit about what will be going on at our annual Wednesday morning breakfast. So, I hope you'll join us then because there's always something Next in Tech.