Why single node control plane?
- Monitoring and alerting are simple and on point. It reduce the number of false positive alerts.
- Setup and maintenance are quick and straightforward. Less complex install process lead to more robust setup.
- Disaster recovery and recovery documentation are more clear and shorter.
- Application will continue to work even if Kubernetes control plane is down.
- Multiple worker nodes and multiple deployment replicas will provide necessary high availability for your applications.
- Downtime of control plane node make it impossible to change any Kubernetes object. For example to schedule new deployments, update application configuration or to add/remove worker nodes.
- If worker node goes down during control plane downtime when it will not be able to re-join the cluster after recovery.
- If you have a heavy load on Kubernetes API like frequent deployments from many teams then you might consider to use multi control plane setup.
- If changes to Kubernetes objects are infrequent and your team can tolerate a bit of downtime when single control plane Kubernetes cluster can be great choice.