DevOps Paradox

#344: Kubernetes is boring now. That's the whole point. KubeCon EU 2026 in Amsterdam -- likely the biggest KubeCon ever at more than 13,000 attendees -- made one thing extremely clear: the container orchestrator is done being interesting on its own. Every keynote, every new sandbox project, every vendor announcement pointed the same direction. AI. Inference. Agents. NVIDIA donated a DRA driver for GPUs to CNCF. Google open-sourced their cluster autoscaler and shipped a DRA driver for TPUs. Red Hat brought LLM-D for disaggregated inference. NVIDIA contributed the KAI Scheduler for AI workloads. The Gateway API now has an inference extension in beta -- model routing baked directly into the Kubernetes networking layer. And here's the thing Whitney pointed out that should make everyone pause: you can't even run inference workloads in containers. They can escape. You need micro VMs. So the container orchestrator is orchestrating things that aren't containers. The platform engineering conversation shifted too. The bottleneck isn't technology anymore -- it's culture. Getting teams to work together differently. And if your company can't trust its own employees to make decisions, good luck trusting agents. Viktor's take on the determinism objection was blunt: agents aren't deterministic, but neither are you. You just think you are. One thread that kept surfacing: agents as first-class platform users. Not agents doing agent things -- agents as the users your platform serves. Viktor sees it in real time -- pull requests created by agents, reviewed by his Claude, responses written by the submitter's agent. Humans aren't even in the conversation anymore. The new CNCF sandbox projects tell the story too. LLM-D, KAI Scheduler, Higress (AI-native gateway). And then Velero -- the Kubernetes backup tool that everyone assumed was already CNCF -- finally donated by Broadcom. Which raises a fair question: is CNCF becoming a dumping ground for projects companies don't want to maintain? Probably some of both. Viktor compared the current state to the first five years of Kubernetes -- everyone focused on low-level components, trying to figure out how to combine 57 different tools. The next wave will be higher-level platforms that bundle all of it. And somewhere underneath it all, the mainframe keeps running. Viktor's bet: it'll outlive AI. YouTube channel: https://youtube.com/devopsparadox Review the podcast on Apple Podcasts: https://www.devopsparadox.com/review-podcast/ Slack: https://www.devopsparadox.com/slack/ Connect with us at: https://www.devopsparadox.com/contact

#343: Here's the thing about your company's APIs -- they were built for your own engineers to use inside your own software. Nobody designed them to be the front door. But that's exactly what's happening. Matt DeBergalis, CEO of Apollo GraphQL, makes a pretty compelling case that AI agents are turning internal APIs into the actual interface between companies and customers. Not the website. The APIs themselves. And most of them aren't ready for that. At all. Think about what happens when you point a model at a typical REST API. GitHub's API returns hundreds of fields for a single repository object. Fine when another service is calling it. But a model? All those extra fields are context you're paying for, and they make the model hallucinate. Matt says you need something between the model and all those backend services -- an orchestration layer that takes one request and handles the mess underneath. That's where GraphQL comes in. He draws a parallel that'll land immediately if you've been in this space a while. APIs right now are pets -- handwritten, named, carefully managed. But AI-generated code is about to produce way more microservices, which means way more APIs. They're going to become cattle. And just like containers needed Kubernetes, APIs are going to need declarative infrastructure to manage them at scale. The conversation takes an interesting turn when Darin pushes back on the idea that developers are becoming architects. His take: we're becoming product managers. Matt says both. Viktor throws in code reviewers. Matt's own story backs it up -- he codes more as CEO than he did as CTO, because AI handles the parts he never had time to learn. He doesn't know modern React. Doesn't need to. One more thing that should make any tech company uncomfortable: if AI agents are how customers find you now, what happens to your docs-page-driven acquisition funnel? Apollo's already made the shift -- their first audience for documentation is the models, not the humans. Matt's contact information: LinkedIn: https://www.linkedin.com/in/debergalis/ YouTube channel: https://youtube.com/devopsparadox Review the podcast on Apple Podcasts: https://www.devopsparadox.com/review-podcast/ Slack: https://www.devopsparadox.com/slack/ Connect with us at: https://www.devopsparadox.com/contact/

#342: Most companies have plenty of documentation. The problem is almost none of it is findable, current, or true. Between what's documented, what's actually true, and what people actually do, there are gaps wide enough to kill any AI initiative before it starts. Viktor makes a distinction that reframes the whole problem: there are two types of documentation. Why something was done -- that's eternal. How something works -- that's outdated the moment someone changes a config and forgets to update the wiki. The information about that change probably exists somewhere -- in a Zoom recording, a Slack thread, somebody's head -- but it's not where anyone would think to look for it. The running system itself is the most accurate documentation any company has. Your Kubernetes cluster tells you how many pods are running right now. Git tells you how many you wished you had. Those aren't the same thing, and pretending Git is the source of truth is a comfortable lie most teams tell themselves daily. RAG won't save this. Not the way most people imagine it -- point an agent at your docs and let it answer questions. That fails for the same reason Google's old enterprise search appliance failed. What could work is a continuous process that watches every information source, extracts what matters, and updates a central location intelligently. We have the pieces for this. Nobody's built it yet. The practical path forward: audit what you have before building anything new. Instrument your documentation the way you instrument applications -- find out what people search for and can't find. Design for retrieval, not storage. Build feedback loops. And stop treating documentation as a project with an end date. The companies that treat this as a strategic advantage instead of a chore are the ones that will actually make AI work for them. YouTube channel: https://youtube.com/devopsparadox Review the podcast on Apple Podcasts: https://www.devopsparadox.com/review-podcast/ Slack: https://www.devopsparadox.com/slack/ Connect with us at: https://www.devopsparadox.com/contact/

#341: Nobody's arguing about whether you need feature flags in 2026. That debate ended years ago. But the code flowing through those flags? That's a different story. AI is writing more of it than ever, review times are climbing, and delivery throughput has actually declined. Trevor Stuart, co-founder of Split.io and now running Feature Management & Experimentation at Harness, calls it the six-lane highway ending in a two-lane bridge. The bottleneck didn't disappear. It moved. Coding got faster, but everything downstream -- reviews, security scans, delivery pipelines -- stayed the same width. Viktor points out this is the exact same pattern from the early agile days: his team shipped every two weeks, but testing still took six months. Different era, same structural problem. Feature flags are part of the fix, but not the way most people use them. Teams are now stuffing prompts, token limits, and temperature settings inside feature flag configurations and running A/B tests on AI agents in production. That's a long way from changing button colors on a marketing page, which is where experimentation started 15 years ago. The culture problem is harder than the tooling problem. Trevor has watched teams run one experiment, see it fail, and quit experimenting entirely. The fear of admitting failure kills more experimentation programs than bad data ever will. Meanwhile, the companies getting real results -- a fast food chain generating millions from kiosk experiments, a global bank driving hundreds of millions in customer acquisition -- are the ones treating experimentation as a permanent operating model, not a one-off project. The conversation also covers Trevor's path from co-founding Split to running it inside Harness post-acquisition. He stayed -- which doesn't happen as often as you'd think. Harness runs what he calls a 'startup within a startup' model, and he breaks down what that actually looks like from the inside, what was hardest to let go of, and why finding your 'why' matters more than any exit. Trevor's contact information: LinkedIn: https://www.linkedin.com/in/trevorbstuart/ YouTube channel: https://youtube.com/devopsparadox Review the podcast on Apple Podcasts: https://www.devopsparadox.com/review-podcast/ Slack: https://www.devopsparadox.com/slack/ Connect with us at: https://www.devopsparadox.com/contact/

#340: The smartest ops people are often the most likely to resist new technology -- and they're not wrong. If you don't change anything, nothing breaks, and nobody blames you. That's a completely rational choice. It's also the one that guarantees you fall behind. Bare metal to VMs, VMs to cloud, cloud to Kubernetes -- every time, the teams that played it safe ended up scrambling to catch up two years later. The safe bet isn't safe. It just feels that way. It gets worse when you look at where the tools come from. Kubernetes? Built by developers. Terraform? Developers. Containers? Developers. The tools ops teams depend on were made by a different tribe. So the pushback isn't really about whether the tech is ready or whether the risk is too high. It's about identity. 'Not my people' is a harder objection to overcome than 'not ready yet,' because no amount of documentation or proof-of-concepts answers it. And about proof -- everyone wants it before they'll move. But the proof already exists. It's the tool someone on your team has been running in shadow IT for a year without any official support. If it survived that long on its own, that's stronger evidence than any pilot program. That's your roadmap. And the way in is small chunks, not grand plans. Move one service. Learn something. Adjust. Repeat. AI in ops follows the exact same pattern. A tool that gets you 50% of the way there for free means you can focus your expertise on the other 50%. That's a win. But the people waiting for AI to be perfect before they'll touch it? They're making the same mistake as the teams that waited for perfect proof before migrating to the cloud. Different decade, same trap. YouTube channel: https://youtube.com/devopsparadox Review the podcast on Apple Podcasts: https://www.devopsparadox.com/review-podcast/ Slack: https://www.devopsparadox.com/slack/ Connect with us at: https://www.devopsparadox.com/contact/

#339: DNS has been around since the 1980s. Nobody's writing blog posts about how it changed their life. But every single thing on the internet depends on it -- including all those AI tools everyone's excited about. Anthony Eden has been in the DNS business since the late nineties, when he was CTO of one of the first seven domain registrars after the .com deregulation. In 2010 he started DNSimple, and he did it without a dime of venture capital. Sixteen years later, his 20-person team runs a global DNS infrastructure with 14 edge nodes and 9 origin servers spread across multiple continents. The conversation covers the mistakes companies make with their domains -- running production DNS on a registrar that was never built for it, sharing logins with no access control, zero documentation on why records exist. Anthony breaks down how DNS actually works at scale (unicast vs anycast, the onion layers of resolvers), why your email deliverability problems are probably a DNS problem, and what the www vs no-www debate looks like in 2026. On AI tools, Anthony's take is practical. They're giving his engineers more time to think about problems instead of typing out solutions. But he's not buying the vibe coding hype -- when you run critical internet infrastructure, everyone on the team needs to understand the systems they're building. And for AI startups hoping to cash out? Most will fail. The twist you put on somebody else's model won't be a moat. It'll just become a feature for something bigger. Anthony's contact information: X: https://x.com/aeden Bluesky: https://bsky.app/profile/anthonyeden.bsky.social LinkedIn: https://www.linkedin.com/in/aeden/ YouTube channel: https://youtube.com/devopsparadox Review the podcast on Apple Podcasts: https://www.devopsparadox.com/review-podcast/ Slack: https://www.devopsparadox.com/slack/ Connect with us at: https://www.devopsparadox.com/contact/

#338: Every company adding AI coding tools runs into the same wall. Developers produce more code, but features don't ship any faster. The bottleneck just slides downstream -- to QA, to security, to legal, to whoever comes next in the pipeline. And the team that got faster? They don't even realize the people upstream could be feeding them more work. Viktor's take: the fastest possible setup is one person carrying a feature from idea to production. Not one person doing everything alone -- a system designed so nobody waits. Tests run in CI. Deployments happen through Argo CD. Security scanning is automated. There's a real difference between wiring up a light switch and hiring a butler to flip it for you. None of this is new. The same thing happened with punch cards, client-server, cloud, Kubernetes. One group adopts the new thing, everyone else says it doesn't apply to them, and the market eventually forces their hand. Meanwhile, every team in every company says they'd love to change if only the rest of the organization would get on board. Every team says this. So who's actually blocked? YouTube channel: https://youtube.com/devopsparadox Review the podcast on Apple Podcasts: https://www.devopsparadox.com/review-podcast/ Slack: https://www.devopsparadox.com/slack/ Connect with us at: https://www.devopsparadox.com/contact/

#337: Time series databases have become essential infrastructure for the physical AI revolution. As automation extends into manufacturing, autonomous vehicles, and robotics, the demand for high-resolution, low-latency data has shifted from milliseconds to nanoseconds. The difference between a general-purpose database and a specialized time series solution is the difference between a minivan and an F1 car - both will get around the track, but only one is built for the demands of real-time operational workloads. The open source business model continues to evolve in unexpected ways. While companies like Elastic and Redis have seen hyperscalers fork their projects, a new partnership paradigm is emerging. Amazon Web Services now pays to license InfluxDB and offers it as a managed service, signaling a shift toward collaboration rather than competition. This approach benefits everyone: vendors maintain development velocity, cloud providers get workloads on their platforms, and customers receive better-supported products. Evan Kaplan, CEO of InfluxData, joins Darin and Viktor to discuss the trajectory from observability metrics to physical world instrumentation, why deterministic models matter more than probabilistic ones when your robot might run over your cat, and what it takes to build a sustainable open source company over a decade-plus journey. Evan's contact information: X: https://x.com/evankaplan LinkedIn: https://www.linkedin.com/in/kaplanevan/ YouTube channel: https://youtube.com/devopsparadox Review the podcast on Apple Podcasts: https://www.devopsparadox.com/review-podcast/ Slack: https://www.devopsparadox.com/slack/ Connect with us at: https://www.devopsparadox.com/contact/

#336: The workplace is on the verge of a transformation as significant as the Industrial Revolution. Just as Bring Your Own Device policies emerged after the iPhone disrupted corporate mobile standards, we are now entering an era where employees may arrive with their own AI teams in tow. The question is no longer whether AI will change hiring and employment - it is how quickly companies will adapt before being left behind by competitors who embrace this shift. Current AI productivity gains remain largely individual rather than organizational. Writing code twice as fast means nothing if the deployment pipeline stays the same speed. But within five to ten years, entire industries face disruption - from primary care physicians to transportation to knowledge work. Companies clinging to restrictive AI policies today risk driving away top talent who have already integrated these tools into their workflows. The intellectual property implications alone - who owns an AI stack trained on company processes when an employee leaves - will require entirely new frameworks for employment law. Darin and Viktor explore these scenarios through the lens of a hypothetical job interview where a candidate brings their own team of AI agents. The conversation surfaces uncomfortable questions about compensation models, corporate governance, and whether we are witnessing the emergence of a new kind of talent that blends human expertise with digital capabilities. YouTube channel: https://youtube.com/devopsparadox Review the podcast on Apple Podcasts: https://www.devopsparadox.com/review-podcast/ Slack: https://www.devopsparadox.com/slack/ Connect with us at: https://www.devopsparadox.com/contact/

#335: Observability tools have exploded in recent years, but most come with a familiar tradeoff: either pay steep cloud vendor markups or spend weeks building custom dashboards from scratch. Coroot takes a different path as a self-hosted, open source observability platform that prioritizes simplicity over flexibility. Using eBPF technology, Coroot automatically instruments applications without requiring code changes or complex configuration, delivering what co-founder Peter Zaitsev calls opinionated observability—a philosophy of less is more that aims to reduce cognitive overload rather than drowning users in endless metrics and dashboards. The conversation explores how Coroot differentiates itself in a crowded market with over a hundred observability vendors. Rather than competing head-to-head with cloud giants like Datadog and Dynatrace, Coroot focuses on developers who need answers fast without building elaborate monitoring systems. The platform combines systematic root cause analysis with AI-powered recommendations, using deterministic methods to trace how errors propagate through microservices before handing off to LLMs for actionable fix suggestions. Darin and Viktor dig into Coroot's business model with Peter, examining why the company chose Apache 2.0 licensing instead of more restrictive options, and how staying bootstrapped with minimal angel funding allows them to play the long game without pressure to chase every hype cycle. Peter's contact information: X: https://x.com/PeterZaitsev Bluesky: https://bsky.app/profile/peterzaitsev.bsky.social LinkedIn: https://www.linkedin.com/in/peterzaitsev/ YouTube channel: https://youtube.com/devopsparadox Review the podcast on Apple Podcasts: https://www.devopsparadox.com/review-podcast/ Slack: https://www.devopsparadox.com/slack/ Connect with us at: https://www.devopsparadox.com/contact/