ai-cloud
Tenant Isolation for GPU Clouds
Build tenant isolation into your GPU cloud. vMetal provisions infrastructure, while vCluster separates tenant APIs and assigns dedicated production workers through Private Nodes.
Trusted by the fastest-growing AI cloud providers
Problem
The GPU Cloud Isolation Trap
Isolating GPU customers from each other means drawing a boundary around both the Kubernetes API and the machines underneath it.
Namespace Isolation Is Too Weak
Namespace-only designs share control-plane and node-level components between tenants.
Physical Clusters Are Too Expensive
A separate management stack for every tenant increases infrastructure and operational work.
DIY Isolation Takes Ongoing Engineering
Custom isolation layers take ongoing engineering effort to maintain across cluster, network, and machine lifecycles.
Solution
GPU Tenant Boundaries with Dedicated Private Nodes
vMetal orchestrates machines beneath independent tenant control planes. Private Nodes dedicate GPU workers with each tenant's own networking and storage configuration, while Platform coordinates templates, access and fleet operations. vCluster Labs software powers 100K+ GPUs.
Layered Tenant Isolation for GPU Clouds
vCluster covers the tenant cluster, worker node, runtime, and optional physical network boundaries needed for GPU cloud tenant isolation.
Worker Isolation
Dedicated GPU Workers Per Tenant
Private Nodes dedicate worker capacity to one tenant cluster, with its own CNI and storage configuration. For GPU cloud tenant isolation, select physical servers or VMs according to the required infrastructure boundary.
- Dedicated worker nodes per tenant
- Tenant-owned CNI and storage configuration
- Physical servers or supported VMs
Control Plane
Isolated Control Planes Per Tenant
Every tenant cluster gets its own virtualized control plane, API server, and RBAC boundary on the control plane cluster.
- Independent tenant API and data store
- Tenant-scoped scheduling on Private Nodes
- No dedicated control-plane servers per tenant
Workload Security
Kernel-Native Isolation Without VM Overhead
vNode uses Linux user namespaces and seccomp to restrict workload privileges and system calls. It adds runtime hardening for GPU cloud tenant isolation without a guest kernel or hypervisor, complementing dedicated workers rather than allocating GPUs.
- Linux user namespaces and seccomp
- No guest kernel or hypervisor
- Helps limit workload breakout
Network Isolation
Hardware-Enforced Per-Tenant Network Boundaries
Tenant-owned CNI configuration and supported Netris integrations provide separate network controls for GPU cloud tenant isolation. Ethernet segmentation can use VLANs, VRFs and ACLs; InfiniBand isolation uses fabric partitions rather than Ethernet constructs.
- Tenant-owned worker network configuration
- Netris integration for supported fabrics
- Explicit segmentation and access policies
Tenant Experience
Self-Service Cloud Experience for Tenants
Tenants can create approved cluster products through the built-in UI and CLI or through a custom portal backed by vCluster APIs.
- Approved templates and access policies
- Dedicated production tenant workers
- Readiness depends on available capacity
Why vCluster
This isn’t a side project. Behind every vCluster deployment is 5+ years of deep K8s engineering, security hardening, and battle-tested infrastructure work at massive scale.
100K+
GPUs Powered
50+
GPU Clouds & F500s
<45
Days to Launch
30K
GitHub Stars
Get Started in 3 Steps
1
Schedule a Demo
Talk to our team about your stack
2
Deploy vCluster
Deploy vCluster on your infra in minutes
3
Onboard Your Tenants
Go live with a hyperscaler-grade tenant experience in days
FAQs
What is tenant isolation in a GPU cloud context?
vCluster gives each tenant an independent Kubernetes API, data store and RBAC scope. Private Nodes dedicate worker capacity to that tenant, including its own CNI and storage configuration. vMetal orchestrates the machines underneath. This separates tenant administration and worker placement; shared infrastructure, application permissions and network policies still require an explicit security design.
How is vCluster Platform different from namespace-level isolation?
Namespaces divide resources within a Kubernetes cluster, but share its API and cluster-wide administrative scope. RBAC, quotas and network policies remain useful controls. vCluster adds a separate tenant API and data store, with Private Nodes dedicating production workers. This reduces dependencies between tenant environments while platform administrators still manage the underlying infrastructure and its failure domains.
Do tenants need to share physical GPU nodes with other tenants?
Private Nodes are worker nodes joined to one tenant cluster, with that tenant's own CNI and storage configuration. They can be physical machines or supported VMs. Use dedicated bare metal when the service requires exclusive physical servers. vMetal can provision the underlying machines, while platform policies define which capacity a tenant may request and how it is released.
How quickly can a new tenant cluster be provisioned?
The virtualized control plane comes up once its control plane and workers are ready, and QumulusAI spins up isolated Kubernetes environments in about 1 minute. For bare metal capacity, vMetal provisions a Machine and joins it to the target tenant cluster once the infrastructure driver finishes.
What GPU cloud providers use vCluster for tenant isolation?
vCluster Labs software powers 100K+ GPUs and 1M+ CPUs, serving 50+ GPU clouds & Fortune 500s combined. In production, Lintasarta operates 170+ tenant clusters, while Boost Run completed its managed Kubernetes launch in under 45 days from the decision. These named deployments provide scale and launch examples for teams evaluating the platform for their own infrastructure.
Does vCluster Platform support compliance and air-gapped deployments?
Supported plans include air-gapped deployment and FIPS features. Private Nodes, tenant-scoped access and optional network segmentation provide controls you can include in a regulated deployment. Select supported images and configurations, and validate the full environment against your requirements. These capabilities do not establish certification of the customer's service, application, hardware or operational procedures.
Launch Your Tenant Isolation GPU Cloud
See how GPU clouds deliver hardware-level tenant isolation with Private Nodes as the recommended production model.
