Migration hubs

Strategy, architecture and tooling for migrating off legacy virtualization

Migration guides for moving onto the Ænix platforms from an existing virtualization or cloud stack. Each hub covers the same three questions: what maps mechanically, what has to be redesigned, and when staying put is the right answer. Cohort-based, parallel-run validated, with Konveyor Forklift shipping in the platform as the VM transfer engine.


Source platforms
VMwareOpenStackCloudStackProxmoxNutanixVirtuozzoIBM Power
assess
Migration path
Readiness AssessmentPilot cohortRolling migration
cutover onto
Cozystack
KubeVirtLINSTORCilium
delivers
Outcome
One control planeSmaller ops team

Browse migration hubs

VMware

The largest single migration trigger of 2026 — Broadcom subscription pressure and licence-model uncertainty. Forklift performs cold and warm transfer from vSphere; the page covers what that actually requires, including the VDDK licence problem you inherit and the guests that do not convert.

Typical pattern: assessment (14-28 days) → destination foundation → cohort migration sequenced against subscription expirations → decommission. A 100-VM estate typically completes in 8-12 months.

OpenStack

An operational case, not a technical-superiority one. Nova and Glance move mechanically because both sides are KVM; Neutron is a genuine redesign; Heat and Horizon are rebuilt; Ironic has no equivalent.

Typical pattern: architecture review → parallel deployment → cohort migration by tenant → decommission with hardware recycled forward. 4-12 months mid-size.

CloudStack

CloudStack is actively maintained, so the case for moving is catalogue breadth, not decay. The work is in the virtual router redesign, the tenancy re-model and the billing integration.

Typical pattern: architecture review → parallel deployment → billing re-pointed first → cohort migration by customer → decommission and catalogue expansion.

Virtuozzo

Three products renamed in 2026, exiting three different ways. Virtuozzo Infrastructure (ex-Hybrid Infrastructure) is a real OpenStack deployment and the one place standard tooling applies. Virtuozzo Application Management (ex-Application Platform, ex-Jelastic) is a PaaS whose export only imports into another Virtuozzo installation — a re-platform, not a lift-and-shift. Virtuozzo Server (ex-Hybrid Server) is past end of maintenance and its system-container line has no announced successor.

Proxmox

For Proxmox deployments that have outgrown the single-organization model — multi-tenancy, a service catalogue beyond VMs, billing.

Typical pattern: productized installer → workload migration via standard VM export/import → tenant model design → cutover. Under 50 hosts and single-tenant, staying on Proxmox is the honest recommendation.

Nutanix

HCI licence and renewal pressure, plus the wish to run VMs and containers on one platform rather than two.

IBM AIX / Power

The longest-horizon migration on this list: an architecture change from POWER to commodity x86, not just a hypervisor change.

Cloud migration strategy

If the trigger is not one specific vendor, start with placement rather than destination.


Migration tooling in the platform

The Ænix platforms ship Konveyor Forklift, the Kubernetes migration toolkit for virtualization, as the VM transfer engine. It is configured through Kubernetes objects — a Provider for the source connection, a NetworkMap and a StorageMap for the mappings, and a Plan executed by a Migration — and virt-v2v handles guest conversion: VirtIO driver injection, VMware Tools removal, static IP and drive-letter preservation.

SourceForklift supportNotes
VMware vSphere (6.5+)Cold and warmWarm uses changed block tracking. VDDK image required for vSAN, strongly recommended otherwise, and supplied by you — it cannot be redistributed.
oVirt / RHVCold and warmWarm uses the same precopy and cutover model.
OpenStackCold onlyEvery instance needs a real power-off window.
OVA filesCold onlyUseful for anything that can export an OVA. Vendor appliances may fall outside vendor support once converted.
Remote KubeVirt / OpenShiftColdCluster-to-cluster.

Warm migration is not live migration: RAM state is not carried across, so a reboot still occurs. What warm buys is an outage measured by the final delta rather than by a full disk copy.

Forklift covers the disk and guest layer only. Tenancy, addressing, billing integration and cutover order are engineering work, and that is what the assessment and the cohort plan exist for. For sources Forklift does not cover — Virtuozzo among them — Ænix engineering builds the export and conversion path as part of the engagement.

Upstream status: Forklift ships in the Ænix platform today. The work to expose it as tenant self-service VM import in upstream open-source Cozystack is in review and is not yet in a released Cozystack version.


Migration economics

Use the TCO calculator to model five-year cost against your current platform, or the Cloud Repatriation TCO Worksheet (free Excel and Google Sheets) for migration cost against ongoing savings and cumulative net cash flow. Most repatriation cases break even within 18-36 months on sustained workloads.


How we engage on migrations

  1. Discovery call — confirm fit, scope, timeline
  2. Platform Readiness Assessment (5-10 days, fixed price agreed up front) — current-state audit, target architecture, cohort plan, risk register. This is also where a no-go decision is still cheap.
  3. Pilot cohort (3-6 months) — first cohort migrated with Ænix engineers inside your team, producing the runbooks for the rest
  4. Rolling migration — cohort by cohort with parallel-run validation, source hardware recycled forward as it frees up
  5. Decommission — source platform shutdown after final validation

See also: Alternatives → for vendor-by-vendor comparison; Solutions → for repatriation strategy; Resources → for free assessment tools.

CloudStack migration — moving a service-provider cloud to Kubernetes

Apache CloudStack to Cozystack migration: zones and offerings, the virtual router redesign, accounts to tenants, billing, and when CloudStack should stay.

OpenStack migration — service-by-service move to a Kubernetes control plane

OpenStack to Cozystack migration: what maps, what is redesigned (Neutron), what does not move (Ironic, Heat), and when staying on OpenStack is right.

Virtuozzo migration — three products, three different exits

Migrating off Virtuozzo: Application Management (ex-Jelastic) is a re-platform, Infrastructure is real OpenStack, and the container line has no successor.

Cloud migration services — strategy for private and hybrid infrastructure

Cloud migration in 2026 is a workload-placement decision, not a race to public cloud. Aenix runs structured migrations: repatriation, VMware exit, greenfield.

Nutanix Migration: Exit HCI Lock-In to a K8s Platform

Nutanix migration to a Kubernetes-native platform: leave HCI lock-in and per-node licensing for KubeVirt VMs, containers, and LINSTOR storage you own.

IBM AIX / Power migration — exit Power to an open cloud

Migrate off IBM AIX/Power and Cloud Pak/OpenShift to an open, Kubernetes-native platform on commodity x86. Honest TCO, Oracle-safe design, EU engineers.

Proxmox to Cozystack migration — when SMB virtualization stops fitting

Proxmox VE is excellent at SMB scale. When deployments grow into multi-tenant cloud builders or service-provider models, the operational model strains....

VMware migration — exit VCF without breaking the application

VMware migration end to end: Forklift cold and warm transfer, VDDK and virt-v2v realities, cohort cutover with parallel-run validation, VCF decommission.

Planning your migration?

Start with a fixed-price Platform Readiness Assessment — architecture review, gap analysis, and a roadmap you can take to the board.