Blog category
Articles on Cloud & Infrastructure from our project work
Everything devs group has written on Cloud & Infrastructure, newest first.
Infrastructure comes down to two questions at devs group: what does running this actually cost, and who still understands the system once the person who set it up has left? The articles in this category answer both with numbers rather than opinions — a fully worked Kubernetes cost model in Swiss francs across Swiss providers, and an upgrade path that carries a cluster through three minor versions without downtime.
Cloud & Infrastructure
Kubernetes v1.37: upgrading without downtime
Static pods break, old cAdvisor flags stop the kubelet, SELinuxMount is now on. What actually goes wrong upgrading to Kubernetes v1.37.
Cloud & Infrastructure
What Kubernetes really costs in Switzerland
A provider comparison in CHF, a fully worked cost model, and the item everyone underestimates: operations cost more than the infrastructure.
Cloud & Infrastructure
Kubernetes is not a goal
When a cluster earns its keep, when a single VM is the better architecture, and how to tell the difference.
Cloud & Infrastructure
Three dashboards instead of thirty
Observability rarely fails because of the tooling. It fails because nobody looks at it any more.
What this category covers
Alongside that sits the counter-position missing from vendor blogs: Kubernetes is not the goal. When a single VM is the better architecture, how to tell the difference, and why observability you actually read is worth more than thirty dashboards — all of it out of running real clusters, from the ERP rollout for GMG to onocoy's globally scaled platform.
For your own cluster, the Kubernetes and cloud infrastructure page describes how we build and operate platforms, on Google Cloud, at Swiss providers, or on-premise. And if you first want to know whether a cluster is worth it at all, a first conversation will get you an honest answer with the arithmetic attached.
Every figure in these articles comes from a public price list with the date it was read, or from our own projects, and every cost model states its assumptions. That makes the texts contestable, which is the point: a cost calculation you cannot recompute is advertising.
The category is maintained on the rhythm of Kubernetes releases. With every new minor version we revisit the upgrade article, and the price comparisons are refreshed when providers move their lists. The update date under the title shows where an article stands.
For your own infrastructure this means you do not have to rebuild any of it yourself. From the cost check through the upgrade path to running it day to day, we take on exactly the work described here — on your cluster rather than on our blog. The details are on the Kubernetes and cloud infrastructure page, and the way in is, as always, a short first conversation.
The best moment for a cost check, incidentally, is before you choose a provider, not after the first annual invoice.