Managed Kubernetes Alternatives in Europe: EKS/GKE Alternatives for GDPR-Sensitive Workloads (2026)

Kubernetes is a fantastic abstraction for running containerised workloads at scale. But if you're a European company — or an international company running workloads that touch EU personal data — the default answer of "just use EKS" is getting harder to justify in 2026.

Data residency, Schrems II, egress fees, DORA, NIS2, the AI Act, and an ever-growing list of sector-specific rules all push European teams to look for alternatives closer to home. And with US cloud providers confirming that data processed by their US parent entities is subject to US law regardless of where the servers sit, "the data is in Frankfurt" is no longer the silver bullet it used to be.

This guide walks through what managed Kubernetes actually gives you, why Europeans increasingly look beyond the hyperscalers, and nine credible alternatives — ranked, priced, and compared. If all you need is K8s-style autoscaling without wanting to run K8s yourself, we'll also show how a Knative-based serverless platform can replace a cluster entirely for 10–20% of the cost.

What Does "Managed Kubernetes" Actually Buy You?

Before comparing providers, it's worth being precise about what a managed Kubernetes service actually includes. The term gets thrown around loosely, and the differences between "managed" offerings matter a lot.

A typical managed Kubernetes service (EKS, GKE, AKS, Kapsule, DOKS, OVH, SKS, etc.) manages:

• The control plane — the kube-apiserver, etcd, controller-manager, and scheduler. You don't SSH into these, you don't back up etcd, you don't patch them. The provider does.
• Control plane HA and upgrades — new Kubernetes minor versions, CVE patches, and control plane scaling are handled for you (usually with a click or a Terraform apply).
• Integrations with the provider's networking — CNI that integrates with cloud VPCs, LoadBalancer services backed by the provider's L4/L7 load balancers, and often a managed ingress controller option.
• CSI storage drivers — so PersistentVolumeClaims map to the provider's block storage (EBS, Scaleway Block, OVH Cloud Disks, etc.).
• Auto-scaling hooks — cluster-autoscaler integrations so node groups grow and shrink with load.
• Optional add-ons — metrics, logging, identity integration, sometimes a service mesh.

What you still own, even on a "fully managed" service:

• The worker nodes. You choose the instance types, pay for them as VMs, and own the OS-level patching in many cases (though providers increasingly offer "node auto-repair" and "auto-upgrade").
• Your workloads. Deployments, StatefulSets, ConfigMaps, Secrets, Jobs, CronJobs — all on you.
• Cluster-level policies. RBAC, NetworkPolicy, PodSecurity, OPA/Gatekeeper or Kyverno rules.
• Day-2 operations. Version pinning, in-place upgrades vs. blue/green, certificate rotation (cert-manager), DNS (external-dns), secret management (Vault / ExternalSecrets), observability stack.

The short version: managed Kubernetes removes etcd nightmares and makes version upgrades less painful. It does not remove Kubernetes itself — it still takes expertise and engineering hours to run well.

Why Europeans Are Looking Beyond EKS, GKE, and AKS

US hyperscalers work great technically. The pressure to look elsewhere is economic, regulatory, and — increasingly — political.

1. The Schrems II / CLOUD Act problem hasn't gone away

The Schrems II ruling invalidated the EU–US Privacy Shield in 2020. The EU–U.S. Data Privacy Framework replaced it in 2023, but a growing body of legal opinion in Germany, France, and the Netherlands treats the framework as fragile — especially after changes to the U.S. Privacy and Civil Liberties Oversight Board. For public-sector and regulated workloads, many CISOs now require European-owned providers, not just European data centers.

2. Control-plane fees add up fast

Amazon EKS charges $0.10 per hour per cluster for the control plane. That's $72/month before you've launched a single pod. Run dev, staging, and production? That's $216/month just for empty control planes. GKE charges similarly for "Standard" clusters ($0.10/hr), though it offers a free tier for one zonal cluster per billing account. AKS is free — which explains a lot about its market share.

3. Egress pricing is punitive

AWS charges roughly $0.09/GB for the first 10TB of egress out to the internet. A workload pushing 5TB/month of outbound traffic — easily reachable for a busy API or a content-heavy site — pays $450/month in egress alone. European providers like Hetzner, Scaleway, and OVHcloud typically include multiple terabytes per server, or charge a fraction of a cent per GB overage. This is one of the biggest hidden drivers of hyperscaler bills and a top reason teams repatriate workloads.

4. Sovereignty and certifications

Regulated industries in France (healthcare HDS), Germany (BSI C5), and the EU financial sector increasingly require providers that hold specific sovereignty certifications: SecNumCloud (France), BSI C5 (Germany), EUCS High (EU-wide, still finalising). Hyperscalers partner with local companies for sovereign offerings, but these are usually more expensive, slower to get new features, and rarely fully independent from the US parent.

5. Ecosystem lock-in

EKS plays best with IAM, VPC, ALB, RDS, S3, and CloudWatch. GKE plays best with IAM, VPC, Cloud SQL, Cloud Storage, and Cloud Logging. The more deeply you integrate, the harder it is to leave. European providers tend to ship more "standard Kubernetes" with fewer proprietary hooks — which is a feature if portability matters.

The Comparison Table: 3-Node 4vCPU / 8GB Cluster, Monthly Cost

Prices are list prices in April 2026, exclusive of VAT, for a realistic starter production cluster: 3 worker nodes, 4 vCPU / 8 GB RAM each, ~80GB storage per node, a single LoadBalancer, and ~1TB/month egress. Where a provider doesn't charge for the control plane, that's reflected below. Actual bills vary by region, reserved instance pricing, committed-use discounts, and special offers.

Note: these are "realistic starter cluster" prices. Real EKS/GKE bills often land 2–3x higher once NAT gateways, egress, managed logging, ALBs, and ECR/Artifact Registry are added. European providers generally don't meter NAT, logs, or image registry as aggressively.

The Alternatives, Ranked

Ranking is subjective. We've weighted by a mix of: feature parity, EU sovereignty, price, ecosystem maturity, and fit for SMBs and mid-market teams — the audiences most likely to be shopping for alternatives in the first place.

1. OVHcloud Managed Kubernetes

OVHcloud is the largest European cloud provider by revenue and the most credible hyperscaler alternative for European sovereignty. Their managed Kubernetes offering runs on top of Public Cloud instances in France, Germany, Poland, and the UK.

• Control plane: free, managed, HA.
• Nodes: B3, D2, R2, and dedicated "GPU" instance types — good range from burstable to compute-optimised.
• Storage: OVH's Cloud Disks via CSI driver; additional integration with their NAS-HA for ReadWriteMany.
• LoadBalancer: their own load balancer as a Service.
• Sovereignty: Standard Public Cloud is not SecNumCloud-qualified, but OVH sells a separate SecNumCloud-qualified offering for highly regulated workloads — the strongest formal sovereignty story in Europe, alongside Bleu, Orange, and Thales.
• Ecosystem: mature Terraform provider, native ArgoCD/GitLab CI hooks, Harbor-compatible private registry.

Choose OVHcloud if: you want a one-to-one EKS replacement with formal sovereignty options and you're comfortable with a provider that has a deeper ops heritage than a developer-UX heritage.

2. Scaleway Kapsule (and Kosmos)

Scaleway is the French answer to DigitalOcean — a developer-first cloud with clean pricing and a strong Kubernetes story. Kapsule is their standard managed Kubernetes. Kosmos is a multi-cloud variant that lets you attach on-prem or other-cloud nodes to a Scaleway-hosted control plane (useful for hybrid deployments).

• Control plane: free for Kapsule, with an optional "dedicated" tier for strict isolation.
• Nodes: their "Play2", "Pro2", "GP1", and "POP2" instance lines cover most shapes.
• Storage: Block Storage CSI with snapshot support. Their Object Storage is S3-compatible.
• LoadBalancer: managed L4 LB, integrated with Kapsule.
• Sovereignty: 100% French-owned (iliad group), data in Paris/Amsterdam/Warsaw. Popular with French public-sector partners.
• Ecosystem: arguably the best developer experience of any European cloud — clean dashboard, solid Terraform/Pulumi providers, managed Postgres and Redis, multi-AZ support, and Easy Deploy for quick app hosting.

Choose Scaleway if: you want a DigitalOcean-tier experience but from a European provider, and you need things like managed Postgres and object storage co-located with the cluster.

3. Hetzner + Talos / k3s (self-managed on dedicated or cloud)

Hetzner doesn't sell managed Kubernetes. But their raw economics (€5–€45 for the 4vCPU/8GB CCX/CX range, or dedicated servers starting around €40 for Ryzen-class machines with NVMe) make them the de-facto substrate for cost-conscious teams building their own K8s. You'll usually see one of two stacks on top:

• Talos Linux + kubeadm — an immutable, API-driven OS designed specifically for running Kubernetes. No SSH, no package manager, all config via the Talos API. Great for teams that want a "real" upstream Kubernetes cluster without the OS-patching headache.
• k3s — Rancher's lightweight Kubernetes distribution. Single-binary, ~50MB, ships with Traefik, Flannel, local-path provisioner. Great for small teams, edge deployments, and development clusters. Production-viable with external etcd or HA-embedded etcd (3+ control-plane nodes).

Add-ons you'll typically wire in yourself: hcloud-cloud-controller-manager for LoadBalancer services, hcloud-csi-driver for block storage, cert-manager for TLS, external-dns for DNS, and a metrics stack (kube-prometheus-stack + Loki + Grafana, or a hosted alternative).

Choose Hetzner + Talos/k3s if: you have the engineering bandwidth to run your own control plane, you want the lowest possible hosting bill in Europe, and you're comfortable owning upgrades and etcd backups yourself. (This is also what DanubeData runs under the hood — so in a sense, every workload on our Serverless Containers is already benefiting from this pattern.)

4. Exoscale SKS (Scalable Kubernetes Service)

Exoscale is a Swiss-owned cloud with deep roots in European regulated industries (banks, health). Their managed Kubernetes — SKS — has two flavours: Starter (single-node control plane, free) and Pro (HA control plane, ~€70/mo).

• Control plane: Starter is free and fine for dev/staging; Pro gives HA.
• Nodes: Standard and CPU-optimised flavours, sensible pricing.
• Storage: Exoscale Block Storage via CSI. Object Storage is S3-compatible.
• LoadBalancer: their own L4 NLB.
• Sovereignty: Swiss-owned (not EU, but often considered equivalent or better by sovereignty-conscious buyers — Switzerland has strict data protection law and is outside CLOUD Act reach).
• Ecosystem: strong for finance/health/public sector; less developer-toy-friendly than Scaleway but more enterprise-serious.

Choose Exoscale if: you need a Swiss or Alpine-region provider, or you're in a sector where Swiss data protection is a selling point.

5. DigitalOcean DOKS (EU regions)

DigitalOcean is US-owned, which breaks the sovereignty story for some buyers. But DOKS (DigitalOcean Kubernetes) is one of the most pleasant managed K8s experiences on the market, and they have real EU regions (Frankfurt FRA1, Amsterdam AMS3, London LON1).

• Control plane: free, managed, HA.
• Nodes: Droplets, including "Premium CPU-Optimised" and AMD variants.
• Storage: Block Storage via CSI; Spaces is S3-compatible.
• LoadBalancer: well-integrated L4 LB.
• Sovereignty: US-owned. If your buyer asks about the CLOUD Act, DigitalOcean is subject to it.
• Ecosystem: great docs, strong community tutorials, the marketplace has 1-click apps. Managed Postgres/MySQL/Redis/Kafka are solid.

Choose DOKS if: you don't have strict sovereignty requirements and you want the easiest managed K8s experience with EU-resident data.

6. Civo

Civo bet early on k3s as their managed K8s distribution. That means extremely fast cluster provisioning (under 90 seconds) and low overhead, which has made them popular for dev/CI/ephemeral-cluster workloads.

• Control plane: free.
• Nodes: small, medium, large; prices are among the lowest in this list.
• Storage: their block volumes.
• LoadBalancer: L4 LB included.
• Sovereignty: UK-owned. Post-Brexit, the UK is a third country for EU GDPR purposes, though the UK has an "adequacy decision" from the EU (as of 2026).
• Ecosystem: lean. Marketplace has Kubeflow, Crossplane, Argo CD templates. Limited managed DB options compared to Scaleway/DO.

Choose Civo if: you want fast, cheap managed K8s and you're comfortable with a k3s-based control plane and a UK-jurisdiction provider.

7. DanubeData — the "no-K8s-at-all" angle

DanubeData is not a managed Kubernetes provider. We run Kubernetes internally (k3s on Hetzner dedicated servers, with KubeVirt for VPS and Knative for serverless) — but we deliberately don't expose a kubectl-level interface to customers. That's the whole point.

If your workload is a stateless container with occasional database and cache calls — which is most SaaS backends, APIs, and web apps — you probably don't need a K8s cluster. You need:

• Scale-to-zero and autoscaling — which is exactly what Knative gives you.
• Managed Postgres / MySQL / MariaDB — our managed databases start at €19.99/mo.
• Managed Redis / Valkey / Dragonfly — our managed caches start at €4.99/mo.
• S3-compatible object storage — €3.99/mo including 1TB storage and 1TB traffic.
• A custom domain with TLS — automatic, included.
• EU data residency — everything runs in Falkenstein, Germany, on European-owned infrastructure.

A typical DanubeData deployment for a stateless app with a DB and a cache:

• Serverless Container for the app: ~€5–€15/mo depending on traffic.
• Managed Postgres: €19.99/mo.
• Managed Redis: €4.99/mo.
• Object storage: €3.99/mo.
• Total: ~€35/mo, including EU data residency, TLS, HA, and 20TB of traffic on the VPS backing the platform.

Compare that to a similar stack on EKS: control plane (€72) + 3 nodes (€200) + RDS (€50+) + ElastiCache (€30+) + LB (€20) + egress. You're well over €400 before you've done anything interesting.

Choose DanubeData Serverless if: your workload fits the stateless-container model, you want K8s-style autoscaling without the operational burden, and you want to stay in Germany.

8. STACKIT (Deutsche Telekom)

STACKIT is Deutsche Telekom's sovereign cloud play, launched in 2021 and marketed primarily at the DACH enterprise and public sector. Their managed Kubernetes (STACKIT Kubernetes Engine, SKE) runs in Frankfurt and Berlin.

• Control plane: included.
• Nodes: a range of flavours including GPU options; pricing is higher than OVH/Scaleway because they're targeting enterprise, not startups.
• Storage: block storage with CSI; object storage S3-compatible.
• LoadBalancer: managed LB.
• Sovereignty: BSI C5 Type 2 attestation, C5:2020, DE-owned by Deutsche Telekom's Schwarz Group joint venture. Used by German federal and state government, hospitals, and large corporates.
• Ecosystem: smaller than AWS/GCP, but includes managed Postgres, MongoDB Atlas partnership, Redis, and an S3-compatible object store. Integration with SAP BTP is a selling point for DACH enterprises.

Choose STACKIT if: you need a formal BSI C5 / GDPR sovereign cloud story, you're selling to German federal/state government or regulated enterprises, and budget isn't the primary constraint.

9. IONOS Managed Kubernetes

IONOS (the cloud arm of United Internet, the German group behind 1&1) is one of the oldest European cloud providers. Their managed Kubernetes offering is solid, predictable, and priced below AWS.

• Control plane: free, managed, HA.
• Nodes: configurable vCPU/RAM/storage combinations.
• Storage: block storage CSI.
• LoadBalancer: managed LB.
• Sovereignty: DE-owned, data in DE/ES regions.
• Ecosystem: business-friendly pricing and contracts, solid enterprise support in German; ecosystem is smaller than Scaleway/OVH but competent.

Choose IONOS if: you're a DACH business with predictable workloads, you value clear German-language support and invoicing, and you don't need a huge ecosystem of managed add-ons.

When to Choose What: A Decision Guide

You're a small team (1–5 engineers) with one or two production apps

You almost certainly don't need a Kubernetes cluster. You need a way to deploy containers, a database, and a cache, and forget about infrastructure for weeks at a time.

• First pick: DanubeData Serverless Containers + Managed Postgres + Managed Redis. €30–€50/mo total. Zero cluster operations.
• Second pick: Scaleway Serverless Containers or Civo's small managed K8s if you truly need kubectl.
• Avoid: EKS/GKE. You'll burn a week setting up IAM, IRSA, ALB, ACM, and cert-manager, and then pay €200+/mo for the privilege.

You're a mid-size team (5–20 engineers) with a real microservices architecture

This is where managed Kubernetes starts to earn its keep. You have enough services that the abstraction pays off, and enough engineers to absorb the operational overhead.

• First pick: Scaleway Kapsule or OVHcloud Managed K8s. Both give you a proper managed control plane, real LoadBalancer integration, and priced-to-win compared to the US hyperscalers.
• Second pick: IONOS or Exoscale SKS if you have regional or regulatory ties.

Sovereignty is a hard requirement (public sector, regulated industries)

You're not shopping on price. You need formal certifications and a provider that's demonstrably outside US jurisdiction.

• First pick (FR): OVHcloud SecNumCloud or Outscale (Dassault).
• First pick (DE): STACKIT (BSI C5).
• First pick (EU-wide): Orange Business Cloud Avenue, Bleu (SAP/Orange/Capgemini JV), T-Systems Open Telekom Cloud.
• Note: DanubeData is EU-owned and DE-hosted, but we don't hold BSI C5 or SecNumCloud certifications — we're not the right pick for federal-government workloads, but we are a very cost-effective pick for EU SMBs.

Cost is the dominant constraint

You want a real Kubernetes cluster, you have the engineering bandwidth to run it, and you want to spend under €100/mo for three nodes.

• First pick: k3s on DanubeData VPS. Three DD Small VPS (2 vCPU / 4 GB) at €12.49/mo each = €37.47/mo for a 3-node k3s cluster with embedded HA etcd. Compare to ~€150 on DOKS or Scaleway.
• Second pick: k3s or Talos on Hetzner Cloud CCX22 instances (€11.90/mo each = €35.70/mo for three).
• Tradeoff: you own etcd backups, upgrades, CVE patching, and debugging. A week's engineering time per year is a reasonable expectation.

You're fleeing an EKS bill

The repatriation trend is real. If you're currently spending €2–€10k/month on EKS and you've been told to cut 30–50%, the playbook usually looks like:

1. Assess what the cluster actually runs. Stateless apps? Move to Knative/Serverless Containers. Databases? Move to managed DB (or self-hosted on VPS if you're brave). S3? Move to a European S3-compatible provider.
2. Repatriate egress-heavy workloads first. They have the fastest payback on European infrastructure.
3. Migrate the K8s cluster to OVH/Scaleway/Exoscale/STACKIT if you genuinely need K8s, or to DanubeData Serverless if you don't.
4. Use dual-running + DNS weight shifting for the cutover — don't big-bang it.

A Concrete Example: 3-Node k3s Cluster on DanubeData

If you want to run your own k3s cluster on DanubeData VPS, here's what a sensible starter setup looks like. This is the same pattern we use internally for our own platform.

Provisioning

# Create 3 VPS instances (DD Small: 2 vCPU, 4 GB RAM, 40 GB NVMe)
# via the DanubeData dashboard or Terraform provider

# Server nodes: k3s-01, k3s-02, k3s-03
# Ubuntu 24.04 LTS
# Static IPv4 + IPv6 (included)

Install k3s with embedded HA etcd

# On the first node (becomes the "server" and starts the etcd cluster)
curl -sfL https://get.k3s.io | K3S_TOKEN=YOUR_SHARED_TOKEN sh -s - server 
  --cluster-init 
  --tls-san k3s.yourdomain.com 
  --disable traefik 
  --disable servicelb 
  --node-label "node-role.danubedata/k3s=true"

# On the second and third nodes
curl -sfL https://get.k3s.io | K3S_TOKEN=YOUR_SHARED_TOKEN sh -s - server 
  --server https://IP_OF_NODE_1:6443 
  --tls-san k3s.yourdomain.com 
  --disable traefik 
  --disable servicelb

Install MetalLB for LoadBalancer services

# Install MetalLB in L2 mode using one of your VPS's public IPs
# (or use an additional IP from the DanubeData dashboard)

kubectl apply -f https://raw.githubusercontent.com/metallb/metallb/v0.14.8/config/manifests/metallb-native.yaml

# Then create an IPAddressPool + L2Advertisement targeting your public IP(s)

Install ingress-nginx, cert-manager, external-dns

helm upgrade --install ingress-nginx ingress-nginx/ingress-nginx 
  --namespace ingress-nginx --create-namespace 
  --set controller.service.type=LoadBalancer

helm upgrade --install cert-manager jetstack/cert-manager 
  --namespace cert-manager --create-namespace 
  --set crds.enabled=true

helm upgrade --install external-dns bitnami/external-dns 
  --namespace external-dns --create-namespace

Storage

k3s ships with local-path-provisioner by default, which uses host-local directories. For HA storage across nodes, you've got a few options:

• Longhorn: replicated block storage with snapshots and backups to S3. The standard choice for k3s in production.
• OpenEBS Mayastor / LocalPV-LVM: higher-performance local storage with snapshotting.
• External S3 for object storage: point your apps at DanubeData Object Storage via the S3 API — €3.99/mo for 1TB.

Total cost

• 3× DD Small VPS @ €12.49/mo = €37.47/mo
• 1× additional public IPv4 (if you want a dedicated LB IP) ~€1/mo
• DanubeData Object Storage for Longhorn backups = €3.99/mo
• Total: ~€42/mo for a 3-node HA k3s cluster with replicated storage and S3 backups.

That's one-fifth of the price of the cheapest managed European option, in exchange for owning control plane upgrades and etcd backups yourself. For a small team that's comfortable with Kubernetes, it's a no-brainer.

Frequently Asked Questions

When do I actually need Kubernetes?

You need Kubernetes when you have enough workloads that the abstractions (Deployments, Services, ConfigMaps, cron jobs, PDBs, HPAs, NetworkPolicies) start paying for themselves. A rough rule of thumb:

• < 5 services, 1–2 engineers: you don't need K8s. Use Serverless Containers or a simple "deploy a container + DB" platform.
• 5–30 services, 3–15 engineers: K8s starts to make sense, but consider whether a PaaS (like DanubeData Serverless + managed DB/cache) gets you 80% of the way there for 20% of the complexity.
• 30+ services, 15+ engineers: K8s is almost always the right choice. At this size you have the headcount to absorb the operational tax, and you probably have bespoke networking, security, or compliance requirements that benefit from being expressed as K8s resources.

Managed Kubernetes vs. self-managed k3s — how do I decide?

Three factors:

• Upgrade cadence: managed K8s handles minor version upgrades for you (usually one-click). Self-managed k3s requires you to drain nodes, run a k3s-upgrade, and verify. Budget 2–4 hours per cluster per minor upgrade; Kubernetes ships 3 minor versions per year.
• Control-plane HA: managed providers give you HA control planes automatically. Self-managed means you run 3+ server nodes with embedded etcd (or external etcd). Not hard, but you need to know what you're doing.
• Debugging depth: when something breaks at 2am, self-managed means you're looking at etcd logs, containerd logs, and kernel dmesg. Managed means you file a ticket. If your team can't do the former comfortably, pay for the latter.

What storage CSI options do I have on each provider?

• Scaleway: Block Storage (CSI), with snapshot support.
• OVHcloud: Cloud Disks (CSI), plus NAS-HA for RWX.
• Exoscale: Block Storage (CSI).
• Hetzner (self-managed): hcloud-csi-driver for block storage; Longhorn for replicated; Mayastor for high-performance.
• DigitalOcean DOKS: Block Storage via CSI; Spaces CSI for object-backed volumes.
• Civo: Civo Volumes (CSI).
• STACKIT: Block storage via CSI; integration with their S3-compatible object storage.
• IONOS: Block storage (CSI).
• DanubeData (k3s on VPS): local-path-provisioner + Longhorn + external S3.

Do I get an HA control plane by default?

• OVHcloud, Scaleway Kapsule, DOKS, Civo, STACKIT, IONOS: yes, HA by default.
• Exoscale SKS: only on the Pro tier; Starter is single-control-plane.
• Self-managed (Hetzner/DanubeData): only if you deploy 3+ server nodes with embedded etcd or external etcd. A single-node k3s with SQLite backend is fine for dev, not for production.

How do GDPR and sovereignty differ — do I need both?

GDPR is a law that applies to personal data processing anywhere, whether in the EU or abroad. Sovereignty is about jurisdictional control over the data: which government can legally compel the provider to hand it over.

• GDPR-compliant but not sovereign: AWS Frankfurt, GCP Frankfurt, Azure Germany. Good enough for many workloads, but subject to the US CLOUD Act.
• GDPR-compliant and sovereign: OVHcloud (especially SecNumCloud), Scaleway, Exoscale, STACKIT, IONOS, DanubeData. Immune to US legal compulsion.
• You need both if you handle health data (HDS), financial data (DORA, PSD3), public-sector data, or explicit "no US cloud" customer contracts.

What does an EKS-to-European-cloud migration actually look like?

A typical 2–6 week migration:

1. Week 1: inventory. Dump kubectl get all -A, list RDS/ElastiCache/S3 buckets, calculate current costs by namespace.
2. Week 2: spin up target cluster (OVH/Scaleway/DanubeData). Install cert-manager, external-dns, ingress, observability.
3. Week 3–4: migrate stateless workloads first (deployments, services, configmaps, secrets). Validate with smoke tests.
4. Week 4–5: migrate stateful workloads — databases (logical replication or dump+restore), Redis (snapshot+replay), S3 (rclone or provider-level migration tools).
5. Week 6: DNS weight shift, run both in parallel for 1–2 weeks, then decommission EKS.

Expected savings: 40–70% on hosting spend, plus a massive reduction in egress and NAT gateway costs.

Can I use DanubeData as a K8s node provider like I use Hetzner?

Yes. DanubeData VPS instances are KubeVirt-backed VMs on Hetzner dedicated servers. You get static IPv4 + IPv6, NVMe local storage, and 20TB of included outbound traffic per VPS. We don't expose a CSI driver for attached block storage today (that's on the roadmap), but for many k3s workloads, local-path + Longhorn on the VPS's local NVMe is actually better than cloud block storage — lower latency and no per-IOPS pricing. Create a VPS and install k3s on it the same way you would on Hetzner Cloud.

Why would I pick DanubeData Serverless Containers over a "real" managed Kubernetes?

Because 80% of the time, you don't actually need Kubernetes — you need what Kubernetes gives you. Specifically: a scheduler that runs your container somewhere, autoscaling (including scale-to-zero), a URL with TLS, and integration with your database and cache. Knative — which powers our Serverless Containers — gives you exactly that, without kubectl, without etcd, without node pools, without upgrade windows, and without a €72/mo control plane fee. The tradeoff: you can't run a StatefulSet, a DaemonSet, or your own custom operator. If you need those things, you need real K8s. If you don't, skip the complexity.

Summary

Managed Kubernetes in Europe in 2026 is a buyer's market. You have credible options at every price point and sovereignty level:

• Price-first, EU sovereign: Scaleway Kapsule (€150–€180/mo for 3 nodes).
• Enterprise + formal sovereignty: OVHcloud (SecNumCloud tier), STACKIT (BSI C5).
• Cheapest possible, willing to self-manage: k3s or Talos on Hetzner or DanubeData VPS (€40–€50/mo for 3 nodes).
• No K8s at all, but you want K8s-style scaling: DanubeData Serverless Containers + managed DB/cache (€5–€50/mo depending on scale).

The real question isn't "which managed K8s should I pick?" — it's "do I need K8s at all?" If you do, there are great European options. If you don't, you can ship faster and cheaper without it.

Ready to try the no-cluster approach? Sign up for DanubeData, get €50 in free credit, and deploy your first Serverless Container in under 2 minutes. If your workload outgrows that, you can always spin up a k3s cluster on our VPS — the migration is just a kubectl apply away.

Questions? Talk to our team — we run k3s ourselves and are happy to help you size a cluster or decide whether you need one at all.