B(H)log Logo

B(H)log

Technology Stories & Life

About
Settings

Chats

Contact

GitHubLinkedInEmail
The Harsh Reality of DevOps for a Broke High School Student

The Harsh Reality of DevOps for a Broke High School Student

4 minutes
DevOpsK3sGrafanaFluxCDGitOpsPrometheusRaspberry Pi

1. The Beginning — A Reckless Challenge from a Broke High Schooler

One day, I made up my mind:

"Even though I have little money, I’ll build my own PaaS."


The monthly cost of a cloud server felt too heavy.
So I chose a Raspberry Pi 4B (2GB RAM) with a 64GB SD card.
In fact, it wasn’t brand-new — I bought it second-hand from a friend for 30,000 KRW (~$23).
This tiny computer became my local data center, where I installed K3s, a lightweight Kubernetes distribution.

curl -sfL https://get.k3s.io | sh -

I created namespaces and connected my GitHub repository for GitOps. At that point, I honestly thought: Finally! I have my own PaaS. 🤯

2. A Quick Detour — Kubernetes, K3s, and PaaS

While preparing for this project, I refreshed my understanding of Kubernetes fundamentals.

Kubernetes Core Concepts

  • Cluster — The entire environment where applications run, made up of multiple servers (nodes).
  • Master Node — The brain of the cluster, responsible for scheduling, the API server, and overall state management.
  • Worker Node — Where containers actually run, following the Master’s instructions.

Master and Worker Nodes can exist on the same machine or on separate ones. K3s is a special case where the Master and Worker are combined into a single instance.

Key Features of K3s

  • A lightweight Kubernetes distribution, ideal for IoT and edge computing.
  • Delivered as a single binary for extremely simple installation.
  • Supports combined Master + Worker node operation, enabling the full Kubernetes experience on a single device.
  • Perfect for local DevOps experiments and learning.

What is PaaS (Platform as a Service)?

PaaS is a service model that lets developers deploy and operate applications without worrying about infrastructure. It automates server setup, networking, and scaling so developers can focus entirely on writing code.

3. First Steps in Observability — Meeting Grafana

After finally setting up K3s, configuring Traefik for proper HTTPS, and confirming GitOps was working via FluxCD, I moved on to what I believe is the most important part of service operation: monitoring.

Observability is crucial in any production environment. So I installed Prometheus + Grafana.

helm repo add grafana https://grafana.github.io/helm-charts
helm install grafana grafana/grafana -n monitoring

When the Grafana dashboard loaded in my browser, I felt truly proud. Real-time graphs for CPU, memory, and network usage danced across the screen.

4. Reality Check — The Limits of a Raspberry Pi

It didn’t take long for reality to hit me.

Source: My Raspberry Pi 4B Grafana + Prometheus Monitoring
Source: My Raspberry Pi 4B Grafana + Prometheus Monitoring
  • CPU usage was always above 50%.
  • With only Grafana and Prometheus running, RAM usage hit 1.4GB out of 1.6GB.
  • The SD card I/O was constantly overwhelmed.
kubectl top nodes

The output from this command felt like my Raspberry Pi was crying out: “Please… stop…” 😭

5. The Decision — The Gap Between Ideals and Reality

At first, it looked amazing. Running a K3s cluster at home, with GitOps and a full monitoring stack. But reality was harsh.

  • For a student, cloud services were expensive.
  • Local hardware wasn’t powerful enough.

Without money, even DevOps feels out of reach… which is honestly depressing.

6. Lessons Learned

  1. Observability is essential — Without Grafana, I wouldn’t have known when to make these decisions.
  2. Resources are not infinite — RAM and I/O hit their limits faster than expected.
  3. Balancing ideals with reality — Finding the sweet spot between cost and performance is where DevOps really begins.

7. Closing Thoughts

I’ve always been interested in container technologies, but Kubernetes always felt like an unscalable wall. However, after experiencing a small part of Kubernetes through K3s, I feel like I’ve grown much closer to it.

Now, I’m more motivated than ever to study further. And I’ve made a big decision — I’m going to build a mini PC powerful enough to run Kubernetes smoothly! The purchase decision is already made in my heart. 🚀

P.S. One day, I’ll come back with a full K8s cluster and a rock-solid monitoring system. 😎 For now, my adorable Raspberry Pi 4B is happily running just a few containers. ^o^ Funny enough, after spending the whole weekend setting up K3s, I ended up going back to plain Docker…