Managed Kubernetes hosting in France
Your infrastructure is slowing you down. Our managed Kubernetes hosting will change that.
Kubernetes orchestrates the deployment, scaling and availability of your containerised applications. But between risky manual deployments, traffic peaks that aren’t handled well and reliance on the knowledge of one or two people, most technical teams aren’t yet getting the full benefit. With our managed Kubernetes hosting, we run your Kubernetes infrastructure from our own data centres in France, with our engineers on call 24/7.
No obligation · A direct conversation with an engineer.
(S(AI-generated diagram) The compute nodes are hosted on the local virtualization platforms at DC1 and DC2. The control plane is distributed across three locations—DC1, DC2, and our stretched cluster platform—so that the loss of a datacenter leaves the cluster manageable and allows workloads to resume on the remaining site.
They use our Kubernetes hosting
We have been hosting and securing the information systems of media groups, financial institutions and public bodies since 1985
> 99,95 %
Measured uptime on our high-availability infrastructure
24/7
On-call service provided by our own system engineers
200+
Managed services clients
Dozens
of Kubernetes clusters in production, in our data centres in France
Since 1985
Designing and operating infrastructure
6 common situations… 1 answer.
Kubernetes doesn’t solve abstract problems. It addresses situations most technical teams face every day.
Incidents that wake up the team
A component fails? Kubernetes isolates it, restarts it and rebalances the load automatically. For everything else, whether it’s a failed node, a storage incident or a saturated cluster, our engineers are on call 24/7. Your teams are no longer on the front line.
Releases that make everyone nervous
CI/CD pipelines make your deployments repeatable and reversible. They no longer depend on one person’s memory or a Friday-evening slot: a rollback returns the application to its previous state with no manual intervention.
Sovereignty that’s hard to guarantee
Your infrastructure is hosted in our own data centres in France, operated by French teams, out of reach of the US Cloud Act. ISO 27001 certified, and HDS V2 certified for health data.
Permanently oversized capacity
With dedicated servers per application, each one is sized for its annual peak: you pay for unused capacity all year round. On a cluster, your applications share a single pool, sized as a whole and adjustable up or down, with no migration.
Traffic peaks that aren’t handled well
Kubernetes detects rising load and starts new instances of your application within seconds. Beyond the provisioned capacity, there are two options: we expand the cluster with you, or you enable the automatic scaling option.
Environments that don’t match
Containerisation ensures that development, staging and production run exactly the same application image, from the developer’s machine all the way to production. Bugs that can’t be reproduced and “it worked on my machine” disappear, along with the time lost chasing them.
Kubernetes with us, or Kubernetes with a hyperscaler?
Kubernetes features are the same everywhere.
The difference lies in the legal framework governing your data, the kind of support you get, and who picks up the phone at 3 a.m.
| Criterion | Hyperscaler Kubernetes (EKS / GKE / AKS) | SdV Kubernetes |
|---|---|---|
| Kubernetes features | = Full | = Full |
| Legal framework for data | ✗ Subject to the US Cloud Act | ✓ French and European law only |
| Where operations take place | European region, international teams | ✓ Data centres and engineers in France |
| Incident support | Tickets, support tiers billed separately | ✓ 24/7 on-call by the engineers who run your cluster |
| Architecture design | Rarely included as standard, negotiated separately | ✓ Support from audit to design and operations |
| HDS certification | Depends on the offering and scope chosen | ✓ HDS-certified hosting |
| Billing clarity | ✗ Variable cost, egress and add-on services billed | ✓ Sizing and price agreed in advance |
| Reversibility | Dependence on proprietary managed services | ✓ Standard Kubernetes, no proprietary service imposed |
A gradual migration, not an abrupt switchover.
We build a step-by-step plan with you, tailored to your context and your teams’ pace. Each step is usable in production before moving on to the next.
STEP 01
Free DevOps audit
Review of your infrastructure, identification of bottlenecks, costed action plan. No obligation.
STEP 02
Test cluster
A Kubernetes test environment, first applications deployed, and your teams upskilled in containerisation and continuous integration.
STEP 03
Migration one application at a time
No service interruption. Your teams become more autonomous with every application migrated.
STEP 04
Production and operations
Production cluster run by our teams: continuous monitoring, 24/7 on-call, DevOps support, capacity changes as your needs evolve.
Case study: from ageing infrastructure to a production cluster.
« We’ve significantly reduced the risks linked to obsolescence and gained the ability to evolve our services with much more peace of mind.. »
Aurélie Cros – CTO, Dans nos cœurs
Starting point: infrastructure designed in the early 2010s, manual deployments, a small in-house team.
The journey: audit, test cluster, team upskilling, then migration one application at a time through to production, at their own pace, with no service interruption.
The outcome: more reliable deployments, less reliance on individuals, and a renewed ability to evolve
Who runs what?
“Managed” means nothing without a clear split of responsibilities. Here’s ours, set out in the service agreement. The foundation is the same for everyone; some responsibilities (observability, CI/CD pipeline) are split differently depending on whether you keep control of them or delegate them to us.
Indicative split. Where the line falls for observability and the CI/CD pipeline depends on the level of support you choose.
| Scope | SdV | shared | you |
|---|---|---|---|
| Physical infrastructure, network, data centres | sdv | ||
| Kubernetes cluster: installation, control plane, etcd | sdv | ||
| Kubernetes version and system component upgrades | sdv | ||
| Cluster networking, ingress, TLS certificates | sdv | ||
| Storage classes and persistent volumes | sdv | ||
| Backup of volumes and manifests | sdv | ||
| Platform monitoring, 24/7 on-call | sdv | ||
| Observability stack: metrics, logs, alerting | partagé | ||
| Cluster sizing and capacity decisions | partagé | ||
| Autoscaling policy: cluster autoscaler limits, application thresholds | partagé | ||
| Image registry and continuous integration pipeline | partagé | ||
| Application secrets management | vous | ||
| Manifest design and application packaging | vous | ||
| Application code, images, tests | vous | ||
| Triggering production releases | vous | ||
| Functional monitoring of applications | vous |
What your cluster runs on
Because an architect wants to know before signing.
Distribution and lifecycle
Standard Kubernetes, with no proprietary layer: you’re not locked into an in-house distribution that could one day make switching provider harder. Minor version upgrade policy defined with you, carried out by our teams.
Cluster layout
Your worker nodes are hosted across our two data centres. The control plane is spread over three locations, across our two data centres and our stretched cluster platform, so that neither quorum nor control is lost if a site goes down.
Capacity reserve
Each site is sized to absorb the other’s workloads. That’s what makes automatic rescheduling effective rather than theoretical: a pod can only restart elsewhere if there’s room elsewhere.
Storage and backup
Replicated storage for persistent volumes, S3-compatible object storage, off-site immutable backups: even if ransomware has encrypted your cluster, a clean, unalterable copy remains available for restoration.
Data services
Managed databases on high-availability infrastructure, alongside or outside the cluster depending on your architecture. Replication, backups and version upgrades are handled by our teams, not yours.
Security
Anti-DDoS protection, network filtering, VPN access, environment segregation, regular penetration testing. All within the scope of our ISO 27001-certified ISMS.
Health data
For organisations subject to the HDS requirement, hosting falls within the scope of our certification, with the same hosting, access and traceability requirements as for your other environments.
What we add when your architecture calls for it
The foundation covers the vast majority of needs. These options address specific constraints: they’re decided during the audit, not by default.
Cluster autoscaler
Automatic addition and removal of worker nodes based on load, going beyond pod scaling alone. Relevant for highly variable workloads: editorial peaks, campaigns, batch processing. Minimum and maximum limits are defined with you, so elasticity doesn’t turn into budget drift.
Nodes on the stretched cluster platform
By default, your nodes are local to one data centre and fail over to the other site if an incident occurs. For components that can’t handle that restart (stateful services, strict continuity requirements), we place them on our stretched cluster platform, where failover is handled by the virtualisation layer, with no application restart.
Managed data services
Managed databases on high-availability infrastructure, alongside the cluster, so you don’t have to run stateful engines inside Kubernetes yourself. This is relevant if your architecture relies on a critical relational database rather than fully stateless services.
HDS scope
Your hosting is brought within the scope of our HDS certification, with the associated documentary and contractual commitments, without requiring dedicated infrastructure separate from your Kubernetes foundation.
A question about your architecture?
Your questions about Kubernetes
Yes, as long as you don’t start with the cluster. For an SME, the value of Kubernetes rarely comes from scalability: it comes from more reliable deployments and less reliance on one or two people. We size the cluster to the real scale of your projects and run it for you, so you don’t have to recruit the DevOps profile the market is so short of.
No, and we never recommend it. The approach is always step by step: audit, test cluster, migration one application at a time, then production. Your teams get to grips with the tools gradually, with no service interruption.
Kubernetes features are comparable. Three differences matter: the legal framework governing your data, the kind of incident support you get, and reversibility. Our comparison table sets out the details criterion by criterion.
Two separate mechanisms come into play, and they shouldn’t be confused.
First, cluster control: the three control plane nodes are spread over three locations, our two data centres and our stretched cluster virtualisation platform. Losing one data centre therefore leaves two members out of three, quorum is maintained and the cluster remains manageable.
Second, your applications: worker nodes are hosted on each data centre’s local virtualisation platform. Those on the lost site disappear, and Kubernetes automatically reschedules their workloads onto the remaining site, which is sized to absorb them. For components that can’t handle that restart, we offer the option of placing the relevant nodes on the stretched cluster platform.
Automatic scaling of your applications is: Kubernetes adjusts the number of instances according to load, within the limits of your cluster’s capacity. Automatic scaling of the cluster itself, meaning adding and removing worker nodes, is an option: the cluster autoscaler. We recommend it for highly variable workloads, with minimum and maximum limits defined together. For a steady workload, it adds complexity with no benefit, and we’ll tell you so.
The cost depends on cluster capacity, the number of applications and the level of support. We don’t publish a price list, but we give you a ballpark figure from the first conversation, before any audit, so you quickly know whether the offering fits your budget.
Yes. For organisations subject to the health data hosting requirement, such as healthcare facilities, medical software publishers and laboratories, hosting falls within the scope of our HDS certification. The exact scope applicable to your project is provided, along with the certificate, when we qualify your needs.
Our support includes knowledge transfer on containerisation and continuous integration, delivered by the engineers who run your platform.
It depends on the number of applications and how complex they are. The initial audit and test cluster usually take a few weeks. The migration, one application at a time, then proceeds at your pace: we don’t set a deadline, we move at the speed your teams can absorb the change.
Your cluster runs on standard Kubernetes, with no proprietary service imposed. Your manifests and images work with any other Kubernetes provider. We don’t build technical dependencies that would lock you in with us.
Let’s talk about your Kubernetes project
Describe your situation in a few lines. One of our engineers will call you back to understand your context and tell you, frankly, whether Kubernetes is the right answer for you.
