Skip to main content

Managed K8s with Metal Cloud Overview

Managed K8s with Metal Cloud gives you a Kubernetes cluster with worker nodes running on dedicated bare metal GPU servers — no shared virtualization layer between your workload and the GPU. FPT Cloud manages the control plane; you manage the Worker Nodes and what runs on them, so you can focus on your AI/ML applications instead of operating the cluster.

What to consider before creating a cluster:

  • Region: the physical location of the cluster affects network latency to your users or data sources. Choose the region closest to where your traffic actually originates.
  • Node count and sizing: your FPT Cloud account has quotas for RAM, CPU, Storage, IP, and similar resources. Work out how many GPU servers you need and the limits you expect to hit before you start, so FPT Cloud can provision the right quota for you.

If this is your first time using Managed GPU Cluster, complete the tasks in Initial setup before you create a cluster. See the FAQ for common questions about regions, quotas, and cluster limits.