Practice

How we implemented synthetic monitoring across Paris, Amsterdam, and Warsaw using Grafana Operator CRDs and GitOps principles, open source and independent. Proudly European.

Most containers request 2GB, use 200MB. Prometheus queries show actual vs requested. Right-sizing saves money, improves density, reduces environmental impact.

Step-by-step: choose pioneer workload, set up OpenTofu, use Flux, encrypt secrets with SOPS, deploy to Kubernetes. Real pattern from our experience.

Claude isn't writing all our code—it's helping us understand complex systems faster. OpenTofu modules, Kubernetes manifests, debugging GitOps issues.

Management cluster queries metrics from test/prod clusters. Secured with mTLS via Gateway API. The implementation: separate Gateways, client certificates, ReferenceGrants.

Our disaster recovery isn't theoretical—we've tested it. Fresh cluster from Git: OpenTofu provisions infrastructure, Flux bootstraps, under an hour to working cluster. Drift is the enemy.

Secrets encrypted in Git, per-environment keys, Flux automatic decryption. Our actual implementation: kubeconfig files in project root, SOPS .enc.yaml files, age keys in GitHub secrets.

True cloud independence requires solving software distribution—making complex applications as easy to deploy as mobile apps are to install. This is where Europe can leapfrog current cloud providers.

The term "public cloud" has become synonymous with American hyperscalers, but this naming convention obscures a critical issue: these aren't truly public infrastructures but private platforms controlled by foreign corporations.