Managed - FPT Kubernetes Engine — Overview
Managed - FPT Kubernetes Engine (M-FKE) runs and maintains the Kubernetes control plane for you. You create worker groups, deploy workloads, and let FPT Cloud handle the control plane operations underneath.
Key features
| Feature | Description |
|---|---|
| Managed control plane | FPT Cloud operates and maintains the Kubernetes control plane. You manage Worker Groups and workloads. |
| Multiple Worker Groups per cluster | Mix Worker Groups with different flavors (CPU, RAM, disk) in the same cluster — up to 100 Worker Groups, each with up to 100 Worker Nodes. |
| Native Kubernetes | Runs upstream Kubernetes, compatible with clusters on other clouds or on-premises — no lock-in for your manifests or tooling. |
| Cluster autoscaling | Scale Worker Nodes automatically with demand, including scale-in to a specific node. |
| Hibernate / wake up | Pause a cluster's Worker Nodes on a schedule to cut cost during idle periods, then wake it back up. |
| Persistent storage | Attach Block Storage volumes to Pods, with scheduled snapshot and restore for PVCs. |
| Load balancing | Expose applications with a Service of type LoadBalancer, including Layer 7 and internal-subnet options. |
| Security enhancements | Cluster benchmark, runtime security, workload scanning, and audit logging for compliance needs. |
| GPU support | Add GPU Worker Groups to a cluster — see FPT Kubernetes Engine with GPU for details. |
Which FPT Kubernetes Engine should you use?
FPT Cloud offers three Kubernetes options under FPT Container. They share the same underlying Kubernetes engine and differ mainly in who operates the control plane:
| Control plane | Best for | |
|---|---|---|
| Managed - FPT Kubernetes Engine (M-FKE) | Operated by FPT Cloud | Teams that want to run Kubernetes without managing control-plane nodes |
| Dedicated - FPT Kubernetes Engine | Dedicated Master Nodes for your cluster (HA or non-HA) | Workloads needing isolated, dedicated control-plane resources |
| FPT Kubernetes Engine with GPU | Same as M-FKE, plus NVIDIA GPU Worker Groups | AI/ML and other GPU-accelerated workloads |
You can also add a GPU Worker Group to a Managed FKE cluster — GPU support is not exclusive to a separate offering.
Supported regions
M-FKE is available in all 4 FPT Cloud regions: HAN (Hanoi), SGN (Saigon/HCM), HAN2 (Hoa Lac), and JPN01 (Japan). A cluster runs in a single region — for cross-region resilience, create one cluster per region and configure it for your BC&DR strategy.
Not sure which option fits your workload? Start with Managed - FPT Kubernetes Engine — you can add GPU Worker Groups or move to Dedicated FKE later as your requirements change.
Get started
- Complete the Initial Setup — account access, a Subnet with Static Pool, and FKE activation/quota.
- Create a new Kubernetes cluster.
- Get cluster access information and connect with
kubectl.