Managed Kubernetes

Managed Kubernetes clusters for orchestrating containerised workloads.

What each cloud calls it

All four European providers in this table offer managed Kubernetes: OVHcloud Managed Kubernetes Service, Scaleway Kubernetes Kapsule, STACKIT Kubernetes Engine, and IONOS Managed Kubernetes. Hetzner does not offer a managed control plane; Kubernetes on Hetzner requires self-managing the control plane on top of their Cloud Servers, which many teams do using tools like k3s or Cluster API.
Control plane management (upgrades, etcd, API server availability) is handled by the provider on all four managed offerings. Node pool management — scaling, instance type selection, OS patching — remains a customer responsibility on all platforms.
Kubernetes version currency varies. AWS EKS, Azure AKS, and GCP GKE typically support the latest minor versions within a few weeks of upstream release. European providers follow at a slower cadence. Check your required version against each provider's current support matrix before committing.
Multi-zone node pools are available on OVHcloud and Scaleway. STACKIT and IONOS operate fewer availability zones per region, so high-availability node pool configurations may be limited to two zones or require multi-region design.
Autoscaling: Cluster Autoscaler is supported on OVHcloud, Scaleway, and STACKIT via standard annotations. IONOS supports node pool scaling but check current Cluster Autoscaler integration status, as support has varied between versions.

OVHcloud, Scaleway, and STACKIT do not charge for the managed control plane. You pay only for the underlying compute nodes. This matches the GCP GKE Autopilot model and is more favourable than EKS ($0.10/hour per cluster) or AKS (free control plane but charges for uptime SLA if enabled).
IONOS charges for the managed Kubernetes control plane; verify current pricing in their portal as it has changed.
Node instance pricing follows the same egress advantage as other compute: intra-cluster traffic within a Scaleway or OVHcloud region is free, which can reduce the networking overhead of east-west microservice traffic that AWS charges for across AZs.
Persistent volume pricing (block storage attached to pods) is generally cheaper on European providers than EBS on AWS, though IOPS guarantees differ. See Virtual Machines for a detailed block storage comparison.

All managed Kubernetes offerings produce a standard kubeconfig. kubectl, Helm, ArgoCD, Flux, and standard Kubernetes tooling work without modification. Provider-specific integration is limited to the control plane API and cloud-provider integrations (load balancers, storage classes).
Terraform providers exist for all four European managed Kubernetes offerings. The OVHcloud and Scaleway providers are the most mature. The STACKIT provider is actively maintained by STACKIT engineering. The IONOS provider covers the essentials but has a smaller surface area.
Load balancer integration: creating a Service of type LoadBalancer automatically provisions a cloud load balancer on OVHcloud, Scaleway, and STACKIT. IONOS support is available but check current documentation for API version compatibility.
Storage classes: each provider supplies a default storage class backed by their block storage. Dynamic provisioning with ReadWriteOnce (RWO) works on all four. ReadWriteMany (RWX) with NFS or object-backed volumes requires additional configuration or an external provisioner.
If your current workload uses EKS Blueprints, AWS Load Balancer Controller, or Karpenter, plan to replace these with provider-neutral equivalents (Cluster Autoscaler, the provider's native load balancer controller) on European clusters.

OVHcloud and Scaleway carry SecNumCloud qualification in addition to ISO 27001 and SOC 2, making them eligible for French public-sector workloads under ANSSI guidance. STACKIT holds BSI C5 and ISO 27001, making it well-suited for German regulated environments. IONOS holds ISO 27001, SOC 2, and BSI C5.
Control plane traffic (kubectl API calls) stays within the provider's European infrastructure. No control plane API calls route through US-based systems, which is the key difference from EKS in EU regions.
Pod-level network policies are supported via Calico or Cilium CNIs on all four providers. Encrypting pod-to-pod traffic requires enabling CNI-level encryption (WireGuard via Cilium) or using a service mesh (Istio, Linkerd), which is a customer responsibility.

Kubernetes workloads are highly portable. Deployments, StatefulSets, ConfigMaps, Secrets, and Ingress resources move with minimal changes. Provider-specific annotations on Services and Ingresses (EKS ALB annotations, GCP Backend Config) need to be removed and replaced with the target provider's equivalents.
Container image registry migration. ECR images need to be pushed to the target provider's registry (Scaleway Container Registry, OVHcloud Managed Private Registry) or a self-hosted registry. Update imagePullSecrets and image references in manifests.
Persistent volume migration. EBS volumes cannot be moved cross-provider. Data must be migrated using application-level backup and restore (Velero snapshots, database dumps, rsync). Plan migration downtime or a blue-green cutover for stateful workloads.
IAM and IRSA replacement. AWS IRSA (IAM Roles for Service Accounts) lets pods assume IAM roles. European providers use their own IAM systems; replace IRSA with provider-specific service account credentials or Vault-based secret injection.
Availability zone count differences. EKS clusters often spread across three or more AZs. European providers may offer only two AZs in a region. Pod anti-affinity rules that reference three zones will not schedule correctly; update topologySpreadConstraints and PodDisruptionBudgets to match the available zone count.