Comparing Kubernetes on European cloud providers
Timo Wevelsiep•Updated: 22.07.2026Editorial note: Versions, commands and prices may change. Please verify critical steps independently before production use. This guide does not replace individual consulting.
Planning Kubernetes on German or European infrastructure? WZ-IT compares providers and operating models against your applications, data, failure objectives and ownership requirements, then implements the selected platform. Explore European Kubernetes infrastructure
European Kubernetes services differ primarily in their operating model. Hetzner provides cloud and bare-metal infrastructure with Kubernetes integrations. OVHcloud, IONOS, STACKIT, Open Telekom Cloud and Scaleway each offer managed Kubernetes. Behind the same product label are different regions, platform integrations, service levels and responsibility boundaries.
The assessment follows operating model, infrastructure path, regions, platform integrations, service levels and responsibility boundaries. These criteria show which provider fits the target architecture.
Contents
- Provider overview
- Hetzner
- OVHcloud
- IONOS Cloud
- STACKIT
- Open Telekom Cloud
- Scaleway
- Core selection criteria
- Why an EU region does not automatically mean sovereignty
- Decision matrix by starting point
Provider overview
Status: 22 July 2026. Product capabilities, regions and prices change. Review current service descriptions before making a decision.
| Provider | Kubernetes model | Infrastructure path | Pay particular attention to |
|---|---|---|---|
| Hetzner | Self-managed or partner-operated | Cloud VMs, dedicated servers, optionally Proxmox | Customer-owned control plane, nodes, storage and platform operations |
| OVHcloud | Managed Kubernetes Service or self-managed | Public Cloud, vRack, dedicated servers | Service tier, regions, shared responsibility and supporting services |
| IONOS Cloud | Managed Kubernetes | IONOS Cloud node pools and storage | Regions, node types, networking and service integrations |
| STACKIT | STACKIT Kubernetes Engine | STACKIT Cloud | Regional availability, platform services, support and responsibility boundaries |
| Open Telekom Cloud | Cloud Container Engine, optionally additional managed services | OTC compute, networking and storage | Difference between the CCE product and additional Managed Containers operations |
| Scaleway | Kapsule or Kosmos | Scaleway Cloud and, with Kosmos, external nodes | Shared responsibility, Kapsule/Kosmos differences and external node pools |
This table is not a ranking. A managed service removes specific tasks but may create a stronger dependency on its cloud ecosystem. A self-managed cluster offers greater design control while moving lifecycle and incident handling to the internal team or operations partner.
Hetzner
Hetzner supports Kubernetes through its cloud and dedicated infrastructure and official integrations such as its Cloud Controller Manager and CSI driver. A native Hetzner managed Kubernetes product is not the basis of this model. Clusters are built and operated by customers or service providers.
This supports several architectures:
- entirely on cloud VMs,
- control plane on cloud VMs and workers on dedicated servers,
- directly on bare metal,
- as VMs inside a Proxmox private cloud.
Hetzner belongs on the shortlist when API-driven cloud resources or dedicated performance should be combined with a customer-defined Kubernetes stack. The complete operational scope must then be designed. See Kubernetes on Hetzner for details.
OVHcloud
OVHcloud Managed Kubernetes Service provides a managed control plane, node pools, load-balancer and storage integration, and private networking within the OVHcloud ecosystem. OVHcloud publishes an explicit responsibility boundary: it maintains the control plane while customers remain responsible for areas including workers, Pods, containers, sizing and data protection.
Self-managed clusters can additionally run on Public Cloud VMs, dedicated servers or Proxmox. OVHcloud therefore supports both native managed and self-managed architectures. Selection depends on the exact region, service tier and required connectivity across Public Cloud, vRack and dedicated resources. See Kubernetes on OVHcloud.
IONOS Cloud
IONOS Cloud Managed Kubernetes operates the Kubernetes control plane and integrates node pools with IONOS Compute Engine. IONOS lists automated upgrades, autoscaling, CSI-based storage and a geographically redundant control plane as service capabilities.
The service is relevant for organizations seeking Kubernetes as a German or European cloud service with integrated compute, storage and networking. As with every managed service, workload security, GitOps, observability, backup and application operations still require a separate assessment.
STACKIT
STACKIT Kubernetes Engine is a managed, CNCF-conformant Kubernetes service within STACKIT Cloud. Its documentation positions SKE for standard Kubernetes applications and containerized workloads, including the migration of existing applications into its European cloud.
STACKIT belongs on a shortlist when a German cloud provider, managed Kubernetes and the STACKIT service ecosystem should be assessed together. Review available regions and zones, required platform services, network paths, support models and ownership above the control plane.
Open Telekom Cloud
Open Telekom Cloud provides the Cloud Container Engine, a Kubernetes-based cluster service. This should be distinguished from Telekom's additional Managed Containers scope, which can cover consulting, implementation and CCE cluster operations plus further platform tasks.
That distinction matters in service comparisons: the cloud product and an additional managed service are not the same offering. Contractual, certification and support characteristics can be relevant for public-sector or regulated organizations, but suitability still depends on the specific requirements.
Scaleway
Scaleway has two models. Kapsule is a managed Kubernetes service on Scaleway infrastructure. Kosmos combines a managed control plane with nodes from other providers or private servers.
The official shared-responsibility documentation details provider and customer duties. Kapsule fits a Scaleway-native platform, while Kosmos addresses multi-cloud or external node pools. Distributed nodes require particular attention to networking, latency, failure domains and the reduced automation available for external resources.
Core selection criteria
1. Responsibility boundary
The word "managed" is not sufficient. Document ownership for:
- data centre, hardware and hypervisor,
- control plane and etcd,
- worker operating system and node lifecycle,
- CNI, CSI, ingress and platform add-ons,
- GitOps, registry and delivery,
- workloads, databases and data,
- monitoring, backup, recovery and incident response.
2. Location and failure boundaries
A region can contain one or several availability zones. Determine whether the control plane, workers, storage, load balancers and backups use the same or separate failure boundaries. Capacity and service levels can also differ by region.
3. Networking
Relevant questions include:
- Are private node networks and private load balancers available?
- How are the API, management services and internal applications reached?
- Can cloud, bare metal and on-premises networks connect?
- Which egress charges and stable outbound addresses apply?
- Which CNI and Network Policy options are supported?
4. Storage and databases
CSI integration alone does not define the data architecture. Review performance, zone binding, snapshots, encryption, replication and recovery. For stateful applications, a managed database may fit better than running the database inside the cluster.
5. Lifecycle and automation
Review supported Kubernetes versions, upgrade deadlines, maintenance windows, node replacement, Terraform or OpenTofu support, Cluster API, GitOps and auditability. Automated creation is only one lifecycle step; upgrades and recovery matter as well.
6. Total cost
TCO includes:
- control plane,
- worker compute and spare capacity,
- storage and snapshots,
- load balancers and public IPs,
- outbound traffic,
- registry, logging and monitoring,
- backups and a second location,
- support,
- internal or external operational labour.
Why an EU region does not automatically mean sovereignty
A data centre in Germany or the EU is relevant but not a complete sovereignty or GDPR assessment. Also review:
- contracting entity and applicable law,
- subprocessors and support locations,
- administrative access paths,
- key management and encryption,
- telemetry and external platform dependencies,
- application data flows,
- portability of images, data and Infrastructure as Code,
- deletion and exit processes.
A European provider can still be misconfigured. Conversely, self-hosting alone does not create data control when backups, identities or SaaS dependencies are overlooked.
Decision matrix by starting point
| Starting point | Provider or model direction |
|---|---|
| Cost-effective cloud or bare-metal foundation with a customer-defined operating model | Hetzner self-managed |
| Native managed platform plus Public Cloud and dedicated connectivity | OVHcloud MKS or OVHcloud infrastructure |
| Managed Kubernetes in a German cloud with integrated node pools | Compare IONOS Cloud and STACKIT against detailed requirements |
| Existing relationship with Telekom cloud and additional managed-service needs | Review Open Telekom Cloud CCE plus exact operating scope |
| European managed platform or external multi-cloud nodes | Scaleway Kapsule or Kosmos |
| Own hardware and classic VMs alongside Kubernetes | Assess Proxmox plus Kubernetes as a joint target architecture |
WZ-IT derives a shortlist from the requirements, validates responsibility boundaries and then builds the platform on the selected model. Existing cloud estates can also include EKS, AKS and GKE in the assessment.
Sources
Rather have it operated?
You'd rather not run Kubernetes yourself? WZ-IT handles setup, operations and maintenance - GDPR-compliant from Germany.
Frequently Asked Questions
Answers to the most important questions
There is no universally best provider. Hetzner suits self-managed clusters on cloud or bare metal, while OVHcloud, IONOS, STACKIT, Open Telekom Cloud and Scaleway provide managed Kubernetes services with different integrations, regions and responsibility boundaries. The choice depends on workloads, operating model, failure objectives and compliance.
No. Provider and data-centre location are only part of the assessment. Data flows, subprocessors, support access, encryption, identities, retention, contracts and application configuration must also be reviewed.
With managed Kubernetes, the provider operates at least the control plane and often parts of the node lifecycle. Customers typically remain responsible for cluster configuration, workloads, data and backup. With self-managed Kubernetes, control-plane and node lifecycle also belong to the customer or operations partner.
Yes. Providers such as Hetzner and OVHcloud offer dedicated servers on which Kubernetes can run directly or inside Proxmox. Load balancing, storage, cluster lifecycle, upgrades and recovery then require a customer-managed design.
Compare control-plane charges, load balancers, block and object storage, public IPs, egress, backups, registry, logs, support and operational labour. VM prices alone do not represent total platform cost.
Yes. WZ-IT creates a requirement-based shortlist, documents responsibility boundaries and designs the target platform. Delivery can use managed Kubernetes, cloud VMs, bare metal or a Proxmox private cloud.
More on Kubernetes
- What is Kubernetes?
- Do we need Kubernetes?
- Kubernetes vs Docker Compose vs Coolify
- Kubernetes on Proxmox
- Kubernetes on Hetzner
- Kubernetes on OVHcloud
- European Kubernetes providers compared
- k3s vs. RKE2 vs. Talos
- Longhorn vs. Rook-Ceph
- Kubernetes backup: the four layers
- Access Kubernetes privately with NetBird
- Gateway API vs. Ingress
- MetalLB vs. kube-vip
- Migrate from ingress-nginx to the Gateway API
- Migrate from Helm 3 to Helm 4
- Taking over a Kubernetes cluster: audit checklist






