What is vCluster?
vCluster creates fully functional virtual Kubernetes clusters that run inside namespaces of a host cluster. Each virtual cluster has its own API server, runs on shared or dedicated infrastructure, and gives you flexible tenancy options—from simple namespaces to fully dedicated clusters.Trusted by the industry: Over 40 million virtual clusters have been deployed by companies like Adobe, CoreWeave, Atlan, and NVIDIA.
Why Virtual Clusters?
Traditional Kubernetes multi-tenancy relies on namespaces, which have significant limitations. Virtual clusters solve these problems by providing:Cluster-Scoped Resources
Use CRDs, namespaces, cluster roles, and other cluster-scoped resources without host cluster conflicts
True Isolation
Complete separation with dedicated control planes and API servers for each tenant
Cost Efficiency
Much cheaper than full clusters—runs as a single pod with shared host resources
Ease of Use
Created in seconds with a single command, works in any Kubernetes cluster
Cluster-Wide Permissions
Grant users admin access inside their vCluster while limiting host cluster permissions
Version Flexibility
Test different Kubernetes versions inside a single host cluster
Key Concepts
Control Plane Isolation
Each vCluster runs its own Kubernetes control plane components:- API Server: Dedicated API endpoint for complete API isolation
- Controller Manager: Manages Kubernetes controllers within the virtual cluster
- Data Store: Embedded etcd or external database (PostgreSQL, MySQL, RDS)
Resource Syncing
vCluster uses bidirectional resource synchronization to connect virtual and host clusters:Virtual Cluster Creation
Users create resources (pods, services, secrets) in the virtual cluster through its API server
Syncing to Host
The syncer component watches for resources and syncs them to the host cluster namespace
Physical Execution
Workloads run on the host cluster’s nodes, using its CNI, CSI, and infrastructure
Shared vs. Isolated Infrastructure
vCluster supports multiple deployment architectures with different levels of isolation:| Architecture | Node Isolation | CNI/CSI Isolation | Best For |
|---|---|---|---|
| Shared Nodes | ❌ | ❌ | Development, testing, maximum density |
| Dedicated Nodes | ✅ | ❌ | Production workloads, compliance |
| Private Nodes | ✅ | ✅ | Full isolation, GPU workloads |
| Standalone | ✅ | ✅ | Bare metal, edge, no host cluster |
Architecture Overview
vCluster follows a lightweight architecture that maximizes efficiency: The vCluster control plane runs as a single StatefulSet in a host namespace. Users interact with the virtual cluster’s API server, completely isolated from the host. The syncer ensures workloads execute on host infrastructure.For more detailed architecture information, see the Architecture Overview section.
Main Benefits
Multi-Tenancy Without Compromise
Provide teams with full Kubernetes clusters instead of limited namespace access:- Admin privileges inside their vCluster
- Ability to install cluster-scoped operators and CRDs
- Complete isolation from other tenants
- Minimal permissions required on the host cluster
Rapid Environment Provisioning
Create isolated Kubernetes environments in seconds:- CI/CD pipelines requiring clean Kubernetes environments
- Development and testing workflows
- Preview environments for pull requests
- Temporary sandbox environments
Cost Optimization
Consolidate multiple teams or projects onto fewer clusters:- 70% cost reduction reported by Fortune 500 companies
- Share expensive infrastructure (GPU nodes, storage, load balancers)
- Sleep mode for inactive clusters to save resources
- Higher density than traditional cluster-per-tenant approaches
GPU and AI/ML Workloads
Optimize GPU utilization across teams:- Maximize GPU utilization without sacrificing isolation
- Support multiple schedulers (Karpenter, Volcano, YuniKorn)
- Private nodes with full CNI/CSI control for AI workloads
- Auto-scaling with Auto Nodes feature
Common Use Cases
GPU Cloud Providers
Launch managed Kubernetes for GPU customers with isolated, production-grade environments
Internal GPU Platform
Self-service GPU access for AI/ML teams with maximum utilization
Multi-Tenancy
Give teams full cluster access while maintaining security boundaries
Development & Testing
Rapid provisioning of isolated test environments for CI/CD
Bare Metal Kubernetes
Run Kubernetes on bare metal with zero VMs and strong isolation
Cost Savings
Consolidate clusters and reduce Kubernetes infrastructure costs by 70%
How vCluster Compares
vCluster offers a unique approach compared to other multi-tenancy solutions:| Feature | Namespaces | vCluster | Separate Clusters |
|---|---|---|---|
| Isolation | Limited | Strong | Complete |
| Cluster-scoped resources | ❌ | ✅ | ✅ |
| Admin permissions | ❌ | ✅ | ✅ |
| Cost | Low | Low | High |
| Provisioning time | Seconds | Seconds | Minutes/Hours |
| Operational overhead | Low | Low | High |
| Kubernetes version | Same as host | Flexible | Any |
What’s New
vCluster continues to evolve with cutting-edge features:v0.30 - VPN & Network Isolation
v0.30 - VPN & Network Isolation
Tailscale-powered overlay networks and automated network isolation for hybrid infrastructures with vCluster VPN and Netris integration.
v0.29 - Standalone Mode
v0.29 - Standalone Mode
Run vCluster without a host cluster—deploy the control plane directly on bare metal or VMs for the highest level of isolation.
v0.28 - Auto Nodes
v0.28 - Auto Nodes
Karpenter-powered dynamic autoscaling for private nodes with automatic provisioning and deprovisioning.
v0.27 - Private Nodes
v0.27 - Private Nodes
External nodes with full CNI/CSI isolation join the virtual cluster directly with their own networking stack.
v0.26 - Hybrid Scheduling
v0.26 - Hybrid Scheduling
Multiple scheduler support for AI/ML workloads and fine-grained namespace synchronization.
Getting Started
Ready to create your first virtual cluster? Here’s what comes next:Quickstart
Get your first vCluster running in under 5 minutes
Installation
Detailed installation instructions for all platforms
Architecture Deep Dive
Understand how vCluster works under the hood
Configuration
Learn how to configure vCluster for your needs
Community and Support
Join thousands of vCluster users:- GitHub: Star the project and contribute
- Slack: Join 5K+ community members
- Documentation: Comprehensive guides and references
- Blog: Latest updates and best practices