Skip to content

Dedicated Tenancy

Dedicated Tenancy gives each deployment its own infrastructure stack (VMs). In a Compose-based plan, select OMNISTRATE_DEDICATED_TENANCY for dedicated VMs and attached storage, then configure a broader network or Kubernetes cluster boundary when required.

The boundary is per deployment, not per tenant: one customer can have several deployments, each with dedicated VMs. Your control plane, generated with Omnistrate, manages provisioning, scaling, upgrades, backups, and lifecycle operations.

Choose the Dedicated Tenancy Level

Level What is dedicated How to configure it What may still be shared
Dedicated VMs VMs and attached storage for each deployment Set OMNISTRATE_DEDICATED_TENANCY. Kubernetes host cluster, network, and cloud account.
Dedicated network A customer network and the workloads placed in it For Hosted SaaS, enable CUSTOM_NETWORKS and select or create a customer network. For an existing customer-owned network, use BYOC with BYO-VPC. Deployments using the same customer network can share the network and host cluster; the account is not dedicated by this feature.
Dedicated Kubernetes cluster A host cluster for each deployment Combine Dedicated Tenancy with CUSTOM_DEPLOYMENT_CELL_PLACEMENT and maximumDeploymentsPerCell: 1. The surrounding cloud account and network unless separately isolated.
Dedicated account A customer-owned cloud account boundary Use BYOC and onboard a separate customer account for each required account boundary. Deployments placed in the same customer account share that account boundary.

The VM, network, and Kubernetes cluster levels are suboptions of Dedicated Tenancy, not separate tenancyType values. Account-level isolation uses the BYOC deployment model, which can be combined with Dedicated Tenancy. A dedicated VM does not automatically imply a dedicated network, cluster, or account; OMNISTRATE_DEDICATED_TENANCY alone does not provision accounts.

For Helm, Kustomize, Terraform, and OpenTofu, use Custom Tenancy and define the dedicated infrastructure in your deployment artifacts.

See the dedicated VM architecture for the same application components deployed on separate VMs, compared with pod-level sharing on shared VMs.

When to use Dedicated Tenancy

Choose Dedicated Tenancy when a deployment needs machine-level or broader infrastructure isolation that separate pods on shared VMs cannot provide. Base the decision on the workload's requirements and the customer's infrastructure policies, rather than treating all enterprise customers as having the same needs.

  • Avoid competition for shared VM capacity. Latency-sensitive databases, sustained data ingestion, and compute-heavy processing may need their own machines so that other customer deployments do not compete for resources on the same workload VMs. Dedicated VMs separate that capacity, but you must still size the deployment and validate its performance under peak load.
  • Meet an explicit compute-isolation requirement. A customer's infrastructure policy or security review may require workload VMs and attached storage to be exclusive to their deployment. Use dedicated VMs for that boundary; do not assume that this also dedicates the network or Kubernetes cluster.
  • Provide customer-specific network boundaries. Customers may need separate routing, private connectivity, or network-level access policies. Configure the dedicated network level when those requirements call for a separate network, rather than relying on VM isolation alone.
  • Separate Kubernetes host clusters. A customer may require a separate host cluster instead of sharing the cluster-level operational boundary with other deployments. Choose the dedicated Kubernetes cluster level for that requirement; dedicated VMs alone can still use a shared host cluster.

Account for the cost of the boundary you choose. Plan for dedicated compute and storage per deployment, including idle capacity, scaling headroom, and provisioning time. One customer can create multiple deployments, each with its own dedicated VMs. Reflect that capacity model in your plan pricing rather than treating dedicated infrastructure as a single allocation per customer.

Use shared infrastructure when it meets the requirements. If separate pods, resource requests and limits, and application-level access controls provide sufficient separation, Infrastructure Multi-Tenancy can use shared VM capacity more efficiently. Customer-specific credentials, endpoints, or application configuration alone do not require dedicated machines.