-
-
GPU Utilization Is Becoming the New Cloud Waste Crisis
Enterprises are now paying premium-market prices for infrastructure that spends most of its life waiting. The number that frames this era: average GPU utilization across enterprise Kubernetes clusters sits at 5%, according to Cast AI’s 2026 State of Kubernetes Optimization Report — drawn from measured production telemetry across 23,000 clusters, not a survey. That figure…
-
-
Your CI-CD Pipeline Is Your Real Infrastructure Control Plane
Terraform defines desired state. Kubernetes reconciles workload state. Cloud consoles expose state. None of those systems decides whether infrastructure state is allowed to change. The ci-cd control plane does — or more precisely, the CI-CD pipeline is the only system in most environments that can hold that authority. That distinction — between storing state and…
-
PersistentVolumes vs StorageClasses: When You Actually Need Each
The PersistentVolume vs StorageClass confusion is not a syntax problem. It is an architectural model problem. Teams get confused because they compare the factory to the disk and forget the claim is what the workload actually touches. PersistentVolume and StorageClass are not alternatives. They operate at different layers of the same provisioning stack — and…
-
Google Just Moved the Control Plane Boundary
The control plane boundary just moved. Most platform architectures were not built for that assumption — and most teams have not noticed yet. For a decade, the Kubernetes scaling playbook had one move: add another cluster. Need more capacity? Add a cluster. Need workload isolation? Add a cluster. Need regional separation? Add a cluster. Need…
-
GPU Scheduling in Kubernetes: Start Before the Scheduler
Most teams think gpu scheduling starts with the scheduler. It starts with demand modeling. By the time Volcano, Kueue, or KEDA enters the conversation, the expensive mistake has usually already been made. The cluster was provisioned against a theoretical peak that rarely materializes. The demand curve was never drawn. The concurrency profile was assumed rather…
-
etcd Is Your Kubernetes Database: What It Does, What Breaks, and What to Watch
etcd kubernetes is the only component in your control plane that holds state — and most teams don’t think about that until the cluster starts behaving in ways they can’t explain. Kubernetes doesn’t store state in your pods. It doesn’t store state in your nodes, your scheduler, or your API server. It stores state in…
-
Operating Gateway API in Production: What the Migration Guides Don’t Cover
>_ Kubernetes Ingress Architecture Series >_ Part 0 The Decision Layer: Four Paths, Four Failure Modes How to evaluate the ingress-nginx retirement before picking a direction >_ Part 1 Gateway API Is the Direction. Your Controller Choice Is the Risk. The architectural shift and what annotation sprawl costs over time >_ Part 1.5 The Control…
-
Kubernetes Ingress to Gateway API Migration: How to Move Without Breaking Production
>_ Kubernetes Ingress Architecture Series >_ Part 0 The Decision Layer: Four Paths, Four Failure Modes How to evaluate the ingress-nginx retirement before picking a direction >_ Part 1 Gateway API Is the Direction. Your Controller Choice Is the Risk. The architectural shift and what annotation sprawl costs over time >_ Part 1.5 The Control…
