Title: Kubernetes Complexity Fatigue and the Search for Simplicity

---

September 6th, 2021. The day I decided to confront my own Kubernetes complexity fatigue head-on.

I’ve been on this journey with Kubernetes since its early days—I’ve wrestled with it, loved it, and ultimately came to terms with how much we’ve come to rely on it. But let’s be real, by 2021, there was an undercurrent of dissatisfaction bubbling beneath the surface in our engineering teams.

It started like any other morning: I woke up with a nagging feeling that something wasn’t right. As I sat down to check my notifications, I found myself reading about a new search engine that favored text-heavy sites and punished modern web design. Huh? That’s like going back to stone tablets compared to the slickness of today’s apps. It made me think: why do we even need such complex systems?

Later in the day, while attending an internal tech conference, I overheard a conversation about SRE (Site Reliability Engineering) roles proliferating. The topic felt timely, but it also struck a chord because it reflected the broader trend of complexity and overhead that seemed to be growing within our platform.

That night, as I was browsing Hacker News, I stumbled upon an article titled “A search engine that favors text-heavy sites and punishes modern web design.” It hit me: we’re drowning in complexity. The ease with which developers can spin up a Kubernetes cluster now means they often end up with overly complex deployments—like choosing between a text-heavy website or a beautiful, modern design.

The irony wasn’t lost on me as I thought about our own platform. We were built on Kubernetes and had adopted Backstage for internal developer portals, but it felt like we had created yet another layer of abstraction that only added to the complexity.

In my personal project today, I was tasked with fixing a tricky bug in one of our microservices. It wasn’t the first time I’ve encountered such issues, but this one felt particularly frustrating because it seemed so unnecessary. The problem boiled down to a misconfigured Kubernetes deployment where we were trying to run multiple containers within a single pod. This is not only non-idiomatic but also makes debugging more complicated.

As I worked through the issue, I couldn’t help but think about eBPF (Extended Berkeley Packet Filter) and its potential impact on our infrastructure. It’s gaining attention for its ability to perform low-level operations without needing to modify kernel code, which could potentially simplify some of our deployment challenges. But it was just one more tool in a vast toolbox that already felt overwhelming.

Around the same time, I received an email about a developer who had been fired from Atlassian while caring for their wife battling cancer. This personal issue highlighted the human aspect behind all these technical decisions and made me reflect on the pressure we place on our teams to deliver complex solutions.

That evening, as I typed up my reflections, I realized that simplification is not just about reducing complexity in code or infrastructure but also in process and culture. We needed a way to cut through the noise and focus on what truly matters— delivering value to our users and customers.

Kubernetes will continue to evolve, but we need to ensure it evolves with us rather than against us. Maybe the next version should come with built-in simplicity tools or best practices to help developers avoid the pitfalls of over-complication. After all, as SREs have long preached, the simplest solution is often the right one.

As I hit save on my blog post, I felt a sense of clarity. Complexity fatigue may be a real issue, but it’s not insurmountable. By embracing simplicity and continuously refining our approach, we can make Kubernetes work for us rather than against us.

---

That was my reflection on that day in 2021. The journey to simpler infrastructure is ongoing, but I’m committed to making progress.