New: leaving Nutanix, end to end, with diagrams and a measured run
Transiva · workload mobility
Escape the
renewal trap.
A Go control plane that discovers, inventories and orchestrates workload exports from VMware vSphere and Nutanix AHV, hands the artifacts to h2kvm for conversion, and runs fleet jobs over REST. Apache-2.0, Community Edition.
Abridged and illustrative; the real run prints live logs and a JSON result.
Why Transiva
Getting out should be a job, not a project.
Every hypervisor renewal buries your VMs deeper in someone else's proprietary API. Getting out usually means a spreadsheet, a maintenance window, and a pile of one-off ovftool invocations that fail mid-transfer.
What changes
- One CLI and job model for both vSphere and Nutanix AHV.
- Resumable exports through a REST daemon.
- Offline artifacts: the source VM stays untouched until cutover.
The stack around it
Export with Transiva, convert with h2kvm, assure with GuestKit, operate on Zeus OS.
Capabilities
Two sources. One workflow.
How it flows
Four steps. No agent in the guest.
1 · Discover
Inventory VMs, disks and storage containers from vSphere or Prism.
2 · Export
Pick the disk up (vSphere API, or NFS for Nutanix) and write a qcow2 and a manifest.
3 · Hand off
h2kvm repairs and converts the artifact offline.
4 · Land
Boot it on KVM or KubeVirt, or hand it to Kairon or Machina.
New · Exit Nutanix
Out of Nutanix in minutes.
One script exports a VM through Prism v4, converts it with h2kvm and boots it on plain KVM and KubeVirt.
Open by default
Apache-2.0. No phone-home.
No agent in the guest. Source, Docker and RPM packaging. Community Edition is free forever for labs and single-host PoCs: two source hypervisors, CLI and REST, support via GitHub Issues.
Moving an estate, not a lab?
CE proves export. Platform moves the estate.
If you're moving 50 to 10,000+ VMs, HyperSDK Platform adds CBT incremental sync, wave planning, SSO/RBAC and a direct line to Zyvor on cutover night: 10 to 11 source providers instead of 2, and KubeVirt as a deploy target.
Start with one VM.
Prove the export in your own lab, then plan the wave.