Tag archive
Platform engineering notes on cluster design, workload operations, and the systems around running Kubernetes well.
A practical Renovate Kubernetes guide: keep Helm charts, Argo CD manifests, image tags, OpenTofu providers, and GitHub Actions current without the PR flood.
Helm chart debt comes from the templating model, not the packaging. What it costs, what the alternatives change, and how we keep authoring charts anyway.
Kubernetes cost allocation with OpenCost: install it with Helm, query per-namespace showback from the Allocation API, and account for idle spend honestly.
Learn how to use Argo Rollouts on Kubernetes for canary and blue-green deployments, automated rollback, and GitOps-based progressive delivery.
Karpenter changes EKS node autoscaling by launching right-sized EC2 nodes from pending pods, then consolidating and replacing them as the fleet drifts.
Use KEDA on Kubernetes to scale from queue depth, lag, and schedules, and understand when scale to zero is worth the cold-start trade-off.
Cilium changes Kubernetes networking by replacing more of kube-proxy with eBPF, enforcing identity-aware policy, and adding flow visibility through Hubble.
Use Kueue to queue Kubernetes batch workloads before they hit the scheduler, so shared cluster quota is admitted deliberately instead of all at once.
Use Crossplane to make Kubernetes your infrastructure control plane — a self-service platform API for cloud resources, reconciled continuously.