The cloud-native landscape continues to evolve at breakneck speed, but what’s actually happening in production environments at major enterprises? At KubeCon London, Swapnil Bhartiya sat down with Brian Douglas, Director of Ecosystem at CNCF, to get the inside scoop on how end users are implementing Kubernetes and cloud-native technologies in the real world.

Platform Engineering: The New Standard for Cloud-Native Adoption

One of the most significant trends emerging from Douglas’s conversations with end users is the shift toward platform engineering as a standardization mechanism. “A lot of earlier companies have different processes for different engineering teams. We have different series as well, and then we have platform engineering,” Douglas explains.

Bloomberg serves as a prime example of this transformation. “Bloomberg was recently on a panel with me, discussing how they standardized their practices around platform engineering within the organization,” Douglas notes. “It’s no longer, ‘I build this way, we deploy that way.’ It’s now, ‘As an organization, this is how we interact with the cloud.'”

This standardization isn’t just about consistency — it’s about scalability and talent acquisition. “The pattern we’re seeing is that they’re able to scale and do more with less,” Douglas observes. “I know it’s hard to find really talented engineers, and it’s also hard to find folks who are standardized on technologies like Kubernetes and other cloud-native products.”

The 99.9% Uptime Imperative

When it comes to reliability, the bar has been set high—and it’s non-negotiable. “Right now, 99.9% uptime is the standard. You don’t expect anything less,” Douglas emphasizes. “Any downtime or server outage means losing money for the business.”

The key to achieving this level of reliability lies in standardization around observability tools, particularly OpenTelemetry. “A lot of these end users—their business is not about worrying about the cloud. So things are really standardized around OpenTelemetry, which is a great project within the CNCF for collecting telemetry from servers. That way, if something happens, everyone knows the next step in the playbook to mitigate risk,” Douglas explains.

The AI Challenge: GPU Utilization and Cost Control

As organizations prepare for AI integration, a significant inefficiency has emerged. Douglas reveals a startling statistic: “I saw a metric in a paper showing around 5% utilization of all accessible GPUs.”

This inefficiency stems from organizations preemptively acquiring resources for AI workloads. “I think we’re specifically preparing for the next wave of AI and how we integrate AI within our companies. So we’re preemptively hoarding GPUs,” Douglas notes.

However, he sees this as an opportunity for the cloud-native community to provide solutions: “However, there’s now an opportunity to mitigate that risk—allowing us to spin resources up and down as needed, so we’re not running servers and spending money unnecessarily.”

Security: A Community-First Approach

Security in the cloud-native ecosystem has evolved from an afterthought to a community-driven priority. “Security is top of mind. We haven’t had many incidents that are public. We do receive CVEs, but they are mitigated.”

The open-source nature of cloud-native technologies creates a unique security advantage: “That’s the beauty of open source. One of the most secure platforms in the world, when it comes to technology, is open source—because if my company and your company are both using the same technology, and I discover something and share it through an open CVE or issue, it gives you the opportunity to catch up as quickly as possible.”

Resources for the Cloud-Native Journey

For organizations looking to accelerate their cloud-native adoption, CNCF provides multiple pathways for engagement and learning. Douglas highlights several key resources, “When it comes to resources, LF Training offers a really strong program to help you level up with Kubernetes. We also have reference architectures—recently, we received one from Adobe—where folks are sharing how they’re building their infrastructure and cloud-native environments. Additionally, we’re always looking for case studies.”

Looking Ahead: Staying Current in a Fast-Moving Ecosystem

The pace of innovation in cloud-native technologies presents ongoing challenges for practitioners. Douglas recommends active community participation as the key to staying current. For organizations just beginning their cloud-native journey, the CNCF’s sandbox projects offer glimpses into emerging technologies, including AI-focused tools like CVT for infrastructure understanding and analysis.

As the cloud-native ecosystem continues to mature, the insights from end users like Bloomberg, Intuit, and NatWest Bank provide valuable guidance for organizations at any stage of their journey. The key themes—standardization through platform engineering, unwavering focus on reliability, and community-driven security—will likely define the next phase of cloud-native adoption.

The message is clear: success in cloud-native isn’t just about technology adoption; it’s about building sustainable practices, leveraging community resources, and preparing for the AI-driven future of infrastructure.

Edited Transcript

Swapnil Bhartiya: Hi, this is Swapnil Bhartiya, and we are here at KubeCon CloudNativeCon in London. Today we have with us Brian Douglas, Director of Ecosystem at the Linux Foundation. Brian, it’s great to have you on the show.

Brian Douglas: Thank you so much for having me. It’s my pleasure.

Swapnil Bhartiya: First of all, I would love to know about this position, Director of Ecosystem.

Brian Douglas: My role is Head of Ecosystem at the CNCF, specifically within the Linux Foundation. I work with end users. So end users are folks who leverage the cloud for their technology businesses. These are not technically vendors. Vendors are kind of obvious, like Google or Microsoft, but these are folks like Intuit or NatWest Bank. We’ve also got Bloomberg as end users, and I help them to engage in cloud native. The benefit of that is that they have their own way of doing cloud native, and they share in a non-competitive way on how to do it.

Swapnil Bhartiya: Which means, do you also have a good pulse of how folks are using Kubernetes? First of all, I’m not even going to ask the question of who is using it, because almost everybody is—like the Linux kernel or Kubernetes. Talk a bit about when you talk to these customers and users, because everybody’s use case is a bit different, right? So talk a bit about how people are using it and what benefits they see. That’s why they’re using it.

Brian Douglas: Yeah, I mean, a lot of times companies start smaller. A lot of earlier companies will have different processes for different engineering teams. So we have different series, but then we have platform engineering. So now you can standardize on how you interact with the cloud. So for example, Bloomberg recently was on a panel with me talking about how they standardized their practices around platform engineering within an organization. So it’s no longer “I build this way, we deploy that way.” It’s now “us as an organization, this is how we interact with the cloud.”

Swapnil Bhartiya: You just shared one example—Bloomberg—but talk a bit about what are the trends and patterns you’re seeing? Because we cannot go through every use case, but what patterns are you seeing? What benefits are people seeing? That’s why they use Kubernetes.

Brian Douglas: Yeah, correct. The patterns we’re seeing is that they’re able to scale and do more with less. Essentially, I know that it’s hard to find really talented engineers, and it’s hard to find folks who are standardized on technologies like Kubernetes and all the other cloud native products. So what we’re seeing now is a trend where everyone can go to a KubeCon like this, and folks—we don’t advocate for everyone to go find your new job—but there is recruiting that happens here, so that way you know what you’re getting when you have folks coming through, because we now have training through LF Education and the CNCF Kubernetes certifications as well. So you can have expectations that if your platform is built this way, you could bring on new engineers to be ready to work within your system on day one.

Swapnil Bhartiya: And as folks start embracing Kubernetes—Kubernetes is also not something easy. We all agree on that; it’s complicated, it’s costly. Also, as they start working from small scale—and I’m not talking about folks like Bloomberg; they have all the resources—as they go on this journey, do they also share the lessons they have learned, which can benefit others?

Brian Douglas: Yeah, correct. So that’s what I run—the end user program. So we have a Technical Advisory Board, we’ve got a Technical Oversight Committee as well, and the Governing Board for the CNCF. These are ways for us to engage with these end users to share their case studies. So the idea is, on the CNCF website, we have case studies where folks can talk about their journey and expanding their cloud native ecosystem, but also just being ready for what the next wave is, which—we get into it—it’s AI.

Swapnil Bhartiya: I talk to a lot of folks, and uptime becomes, especially when you look at interconnected cloud native systems—there are tools, there are processes, and there’s a whole culture there. So when it comes to uptime, what strategies do they deploy to ensure uptime?

Brian Douglas: Yeah, I mean, the uptime right now is like 99.9% is standard. You don’t expect anything less than that. The time that you’re out or your server goes down, you’re losing money for the business. And a lot of these end users, their business is not worrying about the cloud. So it is really standardized around OpenTelemetry, which is a great offshoot inside the CNCF, for how you get telemetry within your servers. So if you see something happen, everyone knows the next step in the playbook to go mitigate their risk. So now we have perfect standards, and a lot of the sponsors here at KubeCon provide really good tools out of the box. But the real best thing about this is OpenTelemetry. The standard is open source, so no matter what vendor you’re using or what tool you’re using in-house, that standard is established. So expectations around downtime are clear and cut and dry.

Swapnil Bhartiya: The flip side of uptime is downtime, right? And downtime has so many causes—something goes wrong. What kind of incident response strategies are organizations embracing to keep the downtime low and uptime high?

Brian Douglas: Yeah, it’s all about communication. So now we don’t have pagers anymore. Back in the day we had pagers—like, as recent as 10 years ago, we had pagers. Now we have PagerDuty, which is a sponsor of KubeCon. The idea there is that it’s all about communication. So there are playbooks on what to do next and what runbooks to run when you have to mitigate risk. So being able to have multi-cluster environments, to be able to spin up an environment, spin down environments—very, very popular topic here at KubeCon. So the idea of having your data sovereign within those clusters helps you mitigate those risks. So when you have, unfortunately, downtime—downtime happens—but you can now be in this position where you don’t even have to notice, where your end users, your customers, don’t have to notice when your stuff goes down.

Swapnil Bhartiya: And if you look at the whole—we talked about uptime, we talked about downtime, challenges, cost. Do you also hear from customers saying, “This is still missing in the stack”? And if yes, what is that?

Brian Douglas: Yeah, I mean, the funny thing is the stack continues to evolve. Now we’re really, really hammering down on what GPU usage is. And things that are missing are around utilization. I saw a metric in a paper around 5% utilization of all accessible GPUs. And I think we’re specifically preparing for the next wave of AI and how we integrate AI within our companies. So we’re preemptively hoarding GPUs, but there’s an opportunity for us to now mitigate that risk so we can spin up stuff and spin stuff down preemptively, so we’re not spinning servers and spending all that money for no reason. So there’s a lot of cost controls around GPU usage that are going to be coming out in the next couple of months. I believe no one’s really hinted at that for me specifically, but I just see that’s where the industry is going.

Swapnil Bhartiya: Since you have thrown the AI hand grenade in here, I do have to ask: what are the emerging tools, frameworks, technologies that people are either dipping their toes in, or they are mature enough, like Kubernetes, that they are using in production?

Brian Douglas: Yeah, I encourage everyone to check out sandbox projects within the CNCF. There’s one that’s CVT, which is good for understanding your server architecture infrastructure. So as you’re trying to mitigate risk and you’re asking questions about the knowledge of your server setup, that’s something actually coming up pretty soon. Currently it’s in sandbox, so open to contributions from anybody in the ecosystem. We definitely need more case studies and stories around people who are now working in this problem set.

Swapnil Bhartiya: We have a good vendor ecosystem here, but what kind of resources are available for users? So not only can they share their stories, but they can also learn from other stories. And what role is CNCF and Linux Foundation playing in enabling that?

Brian Douglas: Yeah, first of all, to say that for resources, LF Training—we’ve got a really good program around leveling yourself with Kubernetes. But the other thing I’ll mention is we have reference architectures. We got one from Adobe recently, so folks are sharing how they’re building their infrastructure and their cloud native environments. The other thing is case studies. If anybody’s got a story around how they build in cloud native, we’re all ears—we want to share those stories publicly at the next KubeCon. So we would love folks to find me and share those stories.

Swapnil Bhartiya: We are here—I’ve been covering this for years, and the pace of innovation as a journalist is—it’s very hard for me to keep up with that. Think about those practitioners who have to actually use it. So how are organizations keeping up with this pace of innovation? Because first of all, they have to keep dipping their toes in it—they don’t want to be left out. But then they don’t have technical expertise, so how do they cope with that? And then how is CNCF and Linux Foundation helping with that?

Brian Douglas: Yeah, so the way we help is really through—one, I mentioned training a couple of times, but membership is another opportunity. So we have the web membership through silver and gold tier and platinum. But we also have a bottom tier membership through the supporter network. And this is an opportunity for folks—if you’re just getting started, you want to keep up to date, you want to be on our newsletter, our mailing list—this is an opportunity for you now to stay in touch. So that way, when it comes time for someone to share their case study or share a story before it’s even ready, we have standing meetings with the TOC on a bi-weekly basis, public and private ones. Public is when anybody can come through and find out what’s happening within the ecosystem. You have to be a member to be part of the private ones, and that’s our encouragement to join and get some extra information on what’s happening in the industry. So we mentioned AI—that’s very fast moving, fast paced. It’s very hard to keep up, but if you can now chat with another end user in a non-competitive fashion to say, “Hey, this is what we’re thinking about, looking for other reference architectures and use cases on how we can solve these problems internally.” For us, it’s an open forum for folks to collaborate and have those conversations.

Swapnil Bhartiya: I do remember the early days—security used to be someone else’s problem, but with cloud, cloud native, even in the cloud native space, then folks started to have security days, and the ecosystem is there. How are folks looking at security in the cloud native ecosystem?

Brian Douglas: Yeah, it’s a good question. Security is top of mind. We’ve had a lot of—well, actually, we don’t really have a lot of incidents that are as public. We do get CVEs that are mitigated. So again, being a part of a newsletter, being part of the community, you can be up to speed around when things happen. For big incidents, the community really comes together and supports each other. It’s not a competitive environment where we’re all keeping these sorts of ideas to ourselves, especially when it comes to security. So the idea is really just being an active participant and consuming the information that’s out there, but also, when you see something, say something. That’s the beauty of open source—one of the most secure platforms in the entire world, when it comes to technology, is open source, because if my company and your company are both using the same technology, and I find something and I go share it within an open CVE or an issue, that’s an opportunity for you to catch up as fast as possible. So it really comes down to paying attention, being involved in committees, showing up at events like this, coming to the keynotes. Tomorrow’s keynote is going to be exceptional when it comes to talking about how folks are using cloud native in the wild, and that’s everyone’s opportunity to catch up.

Swapnil Bhartiya: Brian, thank you so much for joining me today, and of course, for sharing the trends in the user space. We talk about vendors, but we don’t talk enough about users, so thanks for sharing all those trends, patterns, mistakes, lessons learned. And you know, that’s how the whole ecosystem grows—it also has a vendor ecosystem as well. Thanks for sharing all those insights. And as usual, I would love to have you back on the show.

Brian Douglas: Yeah, my pleasure. Thank you.