Sidero Labs Omni on Oxide
When is Oxide the right infrastructure for self-managed Sidero Labs Omni clusters?
Clusters run on Oxide’s integrated rack design
No separate hypervisor layer to provision (read our guide)
Manual setup
Hardware root of trust with trust-quorum key management built into the rack itself (see details)
Auditable, verifiable
Inherits server OEM’s firmware/BMC
No unified control
Capex
One-time purchase, no recurring fees or usage limits
Capex
Hardware purchase
Manual setup
Installer provisions on top of existing vSphere layer
Installer provisions resources directly
Clusters run on Oxide’s integrated rack design
No separate hypervisor layer to provision (read our guide)
Inherits server OEM’s firmware/BMC
No unified control
Inherits underlying hardware firmware
vSphere/NSX software-layer controls above
Provider-managed hardware root of trust
Audited but opaque
Hardware root of trust with trust-quorum key management built into the rack itself (see details)
Auditable, verifiable
Capex
Hardware purchase
Opex
VMware/Broadcom licensing
Opex
Consumption-based billing
Capex
One-time purchase, no recurring fees or usage limits
Talk to an expert to find out which of your workloads are best suited for Oxide.
Which clusters are ideal for Oxide?
Oxide is ideal for teams that want an immutable Kubernetes node OS and API-led control without needing to worry about whether their infrastructure is being routinely updated. In terms of workloads, it’s ideal for any that need locality, sovereignty, predictable infrastructure that you own, an air-gapped environment, or an on-prem target.
Regulated, high-assurance workloads that require an air-gapped environment (especially ones that can’t tolerate opaque BMC/firmware)
Stable-capacity, long-lived clusters (e.g., core platform services, internal PaaS, CI/CD backends, data platforms)
Workloads you want to move off VMware (with Oxide, you own your whole stack, so you can move away from the volatile costs of a subscription licensing model)
Multi-tenant private cloud workloads that need strong isolation guarantees
Why Oxide and Sidero Labs Omni are a good fit
The Oxide Cloud Computer is a rack-scale system secured by a hardware root of trust (with trust-quorum key management built into the rack itself) that ensures attestation across the whole stack.
The rack comes pre-configured with hardware, firmware, the host OS, storage, and a hypervisor that makes all compute, networking, and storage resources accessible through a REST API.
You own the system outright, so there are no recurring subscription fees.
You run self-managed clusters using familiar workflows, on infrastructure built with deep co-design between hardware and software to make sure you don’t have to manage anything below those clusters yourself.
How it works
Oxide exposes infrastructure as an API, Talos exposes the node OS as an API, and Omni represents clusters as declarative resources.
Three layers, one model
Oxide provides the cloud infrastructure, Talos Linux defines the Kubernetes node OS, and Sidero Labs Omni manages machines and cluster lifecycles.
Machine classes become resources
Oxide turns an Omni machine class into VM, disk, VPC, and subnet resources through the Oxide API, giving the three layers a consistent, declarative operating model.
Documented end to end
Oxide documents provider registration, machine classes, cluster creation, Talos and Kubernetes version selection, scaling, workload deployment, and cluster deletion.
Joint support
Supported by both Oxide and Sidero Labs.
Why Oxide is better than bare metal
Running Kubernetes on bare metal removes virtualization overhead, but it creates a new problem: who manages the hardware?
Oxide takes care of all the additional effort of managing hardware (like firmware updates and hardware fault monitoring) as part of a unified product experience.
When you run Omni clusters on Oxide, Oxide’s VMs are a helpful interaction surface between your workloads and the infrastructure underneath.
Oxide gives you elastic VM, disk, VPC, and IP infrastructure, and Omni uses an Oxide-maintained external node driver to create, scale, and remove nodes.
Meanwhile, the Oxide cloud controller manager adds infrastructure-aware node data and floating-IP service exposure so you can make design decisions based on support, load-balancing, and storage limits.
Predictable economics
Hardware purchased outright with open source software. No surprise consumption bills or software subscription renewals.
One hand to shake
One Oxide team supports the whole stack, from firmware to control plane. No finger-pointing between vendors.
Verifiable infrastructure
Hardware root of trust with open source firmware. Every layer is verified from silicon up.
Your cloud tools, your infrastructure
Terraform, Kubernetes, and your existing automation: first-class, supported integrations, no lock-in.
Plus broad compatibility
Standard VMs mean your existing software runs without modification