On a managed Kubernetes service such as Kapsule or Kosmos, Scaleway is responsible for overseeing the Kubernetes control plane.
The control plane comprises multiple components that play a crucial role in cluster management, application scheduling, and cluster state maintenance.
These components encompass the control plane’s core elements, namely etcd, API server, scheduler, cloud controller, and controller manager.
Scaleway offers a range of Kubernetes control planes to cater to a wide range of use cases and to provide you with maximum flexibility.
Whether you are seeking a mutualized environment or a dedicated control plane, we provide a flexible offer to fit your infrastructure needs, which can be a differentiating advantage from other Kubernetes engine providers.
| Control plane type / Features | Mutualized | Dedicated 4 | Dedicated 8 | Dedicated 16 |
|---|---|---|---|---|
| Memory | up to 4 GB1 | 4 GB dedicated RAM | 8 GB dedicated RAM | 16 GB dedicated RAM |
| CPU | up to 1vCPU | 2vCPU | 2vCPU | 4vCPU |
| API server Availability | 1 resilient replica | 2 replicas for HA | 2 replicas for HA | 2 replicas for HA |
| etcd Availability | 3 replicas in multi-AZ | 3 replicas in multi-AZ | 3 replicas in multi-AZ | 3 replicas in multi-AZ |
| SLA | N/A | 99.5% uptime | 99.5% uptime | 99.5% uptime |
| Audit logs | N/A | Enabled | Enabled | Enabled |
| Max cluster size | 150 nodes | 250 nodes | 500 nodes | 500 nodes |
You have the option of choosing the most suitable control plane type for your workload when setting up a Kubernetes Kapsule or Kubernetes Kosmos cluster.
The cost of the control plane varies depending on whether it is a shared or dedicated environment, and by the specific dedicated resources that you choose. Although indicated that the control planes are billed monthly in the Scaleway console, the billing for each control plane option is calculated on an hourly basis. Scaleway chooses to display a monthly billing for simplicity purposes.
The control plane within the Kubernetes architecture goes beyond being just another component; it acts as the central command that orchestrates the cluster’s activities. Comprising elements like the API server, etcd, scheduler, and controller manager, the control plane oversees critical operations including resource distribution, scheduling, health checks, and scaling. Making frequent modifications to the control plane can result in unintended repercussions, compatibility conflicts, and potential downtimes, directly impacting the overall dependability of the whole cluster. Therefore, it is imperative to handle the control plane with the utmost care and prudence to sustain a smoothly operating Kubernetes cluster.
When selecting a dedicated environment, you are committing to this plan for 30 calendar days. While the cluster is active, you retain the flexibility to shift their control plane among different types to better align with their workloads. However, certain conditions govern this migration process:
Migrating to a lower tier will have some impact on key features. The allocated RAM resources will be reduced, the maximum number of nodes will be limited, and in the case of a mutualized control plane, the SLA (Service Level Agreement) will be revoked.
If a dedicated cluster is deleted, the commitment is automatically canceled, and there will be no further billing for this dedicated service.
During the lifecycle of a Kubernetes cluster, you have the flexibility to transition its control plane to a different type, whether it involves scaling up or scaling down, to align with their specific workloads.
Refer to our dedicated documentation for more information regarding the choice of control plane offers for clusters, as well as instructions on how to manage a control plane through the console or API.