For a typical VMware HCX Connector-to-HCX Cloud pairing, start the reverse migration from the HCX Connector in your on-premises data center. In the HCX migration wizard, enable Reverse Migrations, prepare the destination compute, datastore and network, then replicate each workload and schedule its cutover. Bulk Migration copies most disk data before the maintenance window, but the final switchover powers off the cloud VM for an offline synchronization before the destination VM is created. It is therefore a planned-outage process, not a guaranteed zero-downtime move.
The exact procedure depends on the provider’s HCX topology, permissions, software versions and datastore types. Confirm that your hosted service permits reverse migration and that the chosen migration method is supported before committing to a window.
As an Amazon Associate I earn from qualifying purchases.
Can you move VMware VMs from a hosted cloud back to on-premises?
Yes, when the hosted environment exposes an HCX deployment that supports the operation and the provider allows workloads to return to your customer-managed site. In the common Connector-to-Cloud design, the on-premises HCX Connector is the control point for reverse migration. The hosted HCX Cloud side does not necessarily present the same menus or permissions as another provider’s service, so do not assume that a procedure from one hosted VMware platform applies to all others.
Free tools Windows power users keep installed
One-click scans. No signup required.
Before touching a VM, establish four facts with the provider and your internal VMware team:
#1 Best Overall
- Which site owns the HCX Connector and which site owns the HCX Cloud instance in the pairing.
- Whether the provider permits outbound migration to your data center, including the required HCX service and network ports.
- Which HCX, vSphere, ESXi and datastore versions are supported together.
- Whether the workload is an ordinary supported VM rather than a VMware management appliance.
For the documented Connector-to-Cloud workflow, migration initiation belongs on the on-premises Connector. If your service uses a different topology, obtain the provider’s runbook before scheduling a reverse move.
What happens during an HCX reverse migration?
HCX Bulk Migration reduces the work in the outage window by copying VM disk data while the source VM is still running. At cutover, HCX powers off the source, performs an offline synchronization of changes made since replication began, and instantiates the VM at the destination. The source therefore experiences an interruption, even though the full disk copy happened earlier.
The duration is workload- and infrastructure-dependent. Slow links, storage contention, firewall inspection, MTU problems and very large VMs can extend transfer or switchover time. No universal downtime or speed figure is published for all environments; use the HCX estimate and your own trial migration rather than promising application owners a fixed outage.
Recommended Free Tools
Step 1: Map the hosted topology and migration authority
Identify the HCX pairing
Document the hosted vCenter, on-premises vCenter, HCX Manager locations, service mesh, network profiles and the compute and datastore clusters visible at each site. Record which HCX Manager can see the other site and where the Reverse Migrations option is available. Take screenshots or export the current configuration for the change record.
Confirm provider permission
A provider may restrict destination networks, datastore types, migration direction or the use of HCX services. Ask the provider to confirm that the destination vCenter and clusters are approved targets and that the HCX entitlement covers reverse migration. This check is essential because the general Connector-to-Cloud workflow does not prove that every hosted service exposes the same capability.
Classify what you are moving
List ordinary application VMs separately from infrastructure appliances. HCX cannot migrate VMware software appliances such as vCenter Server and NSX Manager. An independently migrated vCenter Server Appliance is explicitly unsupported. Plan a separate, vendor-supported method for management appliances and do not include them in an HCX workload batch.
Step 2: Verify versions, licensing and datastore compatibility
Check the actual software matrix
Validate the HCX and vSphere versions on both sides, ESXi compatibility, licensing and the provider’s supported migration methods. Support is specific to the version pair and storage combination. A Broadcom example for VCF 5.x vVol at the source and VCF 9.0 VMFS at the destination supports Bulk Migration but not vMotion or Replication Assisted vMotion (RAV) to or from vVol datastores. That example must not be generalized to other releases or datastore pairs.
Also check product lifecycle status. Broadcom lists HCX 4.10.x as reaching end of general support on July 27, 2025. That date applies to the 4.10.x line; it is not a statement that every current HCX release is unsupported. Use the current release and upgrade guidance for your installed version before designing a new migration.
Confirm capacity at the destination
Measure the provisioned and consumed space of every VM, plus snapshots, swap requirements and any temporary overhead required by the migration. Broadcom’s Bulk Migration guidance calls for enough free space on the target datastore and notes that additional space may be used temporarily. Leave headroom for the destination VM to be created after offline synchronization; a datastore that only fits the nominal disk size can still fail at cutover.
Prepare compute and service profiles
Choose the destination cluster, resource pool, folder, VM placement policy and DRS behavior. Confirm that the destination service mesh, compute profile, network profile and required HCX services are healthy. Check CPU, memory, encryption, hardware-version and virtual-device compatibility before adding a VM to a batch.
Step 3: Design the destination network before copying data
Map each dependency
For every VM, record its VLAN or segment, subnet, gateway, DNS servers, static routes, firewall rules, load-balancer membership and application dependencies. Decide whether the VM will retain its IP address through a supported Layer 2 extension or move to a new routed network with an IP and DNS change. Document the order in which tiers must be stopped and started.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Validate paths and security controls
Test reachability between source and destination HCX components, the migration network, target ESXi hosts and target datastores. Confirm that required ports are allowed in both directions and that inspection devices do not unexpectedly alter or drop replication traffic. After a test VM is moved, verify guest-to-gateway, DNS, application and external-service communication from the destination network.
Avoid Layer 2 extension loops
Do not extend the same network from two on-premises sites to the cloud at the same time unless the design explicitly supports that arrangement. Broadcom warns that simultaneous extensions can create Layer 2 loops. A documented transition removes the existing site’s extension before extending that network from the replacement site. Coordinate this sequence with the network team and check MAC-learning and spanning-tree behavior before changing ownership of an extension.
Step 4: Prepare a migration and rollback runbook
Group workloads by application, dependency and acceptable outage rather than selecting an arbitrary number of VMs. For each group, record the source VM name, destination cluster and datastore, target network, owner, backup status, test method, maintenance window and rollback decision point.
- Take or verify an application-consistent backup according to the application’s recovery policy.
- Confirm that monitoring, backup agents, security controls and automation will recognize the destination VM.
- Define who can approve cutover and who can stop it if replication, storage or application checks fail.
- Keep the original source available until application owners sign off, unless the provider’s workflow removes it as part of a successful migration.
- Specify whether rollback means restarting the original VM, restoring a backup or reversing the migration, and identify the data-loss boundary for each choice.
Run a small, representative test first. Include a VM with the same datastore type, network behavior and application dependencies as the production batch. A test validates permissions and paths; it does not establish a universal production downtime figure.
Step 5: Start reverse migration in the HCX Connector
Use the following sequence for a Connector-to-Cloud pairing. Labels can vary slightly by HCX release, so follow the provider’s screen names if they differ.
- Sign in to the HCX Manager on the on-premises Connector side.
- Open the migration view and create or edit a migration job.
- Enable Reverse Migrations. This makes workloads in the hosted cloud available as migration sources.
- Select the cloud VM or batch, then choose the on-premises destination site, compute resource, datastore and destination network.
- Review the compatibility and capacity checks. Resolve unsupported devices, insufficient datastore space, disconnected networks or missing profiles before proceeding.
- Choose Bulk Migration when it is the supported method for the source and destination combination. Set the cutover or maintenance schedule and confirm the planned source power-off behavior.
- Submit the job and monitor replication, warnings and the estimated time remaining. Do not treat an estimate as a guaranteed completion time.
If the cloud VM does not appear after Reverse Migrations is enabled, stop rather than improvising. Check the pairing state, provider permissions, source inventory visibility and HCX version support with the service owner.
Step 6: Replicate, schedule and execute cutover
During replication
Bulk Migration transfers the VM’s disk data before the maintenance window. Watch transfer rate, remaining data, HCX events, source datastore latency, destination capacity and network utilization. Slow infrastructure or network conditions can lengthen the copy. Avoid unrelated storage migrations or network changes while a large batch is synchronizing.
Rank #4
At the maintenance window
- Confirm that application owners have stopped writes or placed the application in its approved maintenance state.
- Check that the latest backup and the rollback decision are recorded.
- Start the HCX cutover from the migration job.
- Allow HCX to power off the hosted source and complete the offline synchronization of final changes.
- Wait for the destination VM to be instantiated and registered on the selected on-premises resources.
- Power on and validate the destination VM only after HCX reports the migration stage complete.
The offline synchronization and VM instantiation are the critical outage stages. Very large VMs may take longer to finish this phase even when most data was replicated earlier. Keep the window open for checksum or final-copy work rather than scheduling the next change immediately after the nominal start time.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallStep 7: Validate the destination before releasing the source
Validation should be specific to the guest and its application, not limited to seeing a green HCX status.
- Confirm that the VM boots, VMware Tools or the applicable guest integration is healthy, and expected virtual devices are present.
- Check the IP address, gateway, routes, DNS registration and reverse lookup.
- Test application transactions, service-to-service calls, authentication, storage mounts, queues and external integrations.
- Verify monitoring, alerting, backup scheduling, vulnerability scanning and security policy attachment.
- Check performance, time synchronization and log forwarding under representative load.
- Obtain application-owner approval and record the validation evidence.
Keep the hosted source and its rollback information available until that approval. Do not delete or repurpose the source merely because the destination VM powered on successfully.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Which migration method should you use?
Compare methods using the exact versions, datastore types and network design in your environment. The following distinctions are established by the HCX guidance:
| Decision factor | Bulk Migration | vMotion or RAV |
|---|---|---|
| Data movement | Replicates disk data before cutover, then performs offline synchronization. | Support depends on the source and target versions and datastores. |
| Cutover behavior | Powers off the source during final synchronization and destination instantiation. | Not supported in the cited VCF 5.x vVol to VCF 9.0 VMFS example. |
| vVol compatibility in the cited example | Supported for the stated VCF 5.x vVol to VCF 9.0 VMFS case. | Not supported to or from vVol datastores in that stated case. |
| Planning implication | Reserve a maintenance window and target datastore headroom. | Do not select it without confirming the exact compatibility matrix. |
These are not interchangeable labels for every HCX release. A provider’s supported-migration matrix takes precedence over assumptions based on the method name.
Troubleshoot slow transfers and failed jobs
Transfer rate is unexpectedly low
First verify that replication is using the intended private or Direct Connect path rather than an unintended route. Compare MTU settings end to end, including tunnel and physical interfaces. Review Layer 4–7 firewall inspection, packet drops, retransmissions and HCX logs. MTU mismatch and firewall packet inspection are documented possibilities, not guaranteed explanations; measure before changing production settings.
Best Value
The VM cannot be selected as a source
Confirm that Reverse Migrations is enabled on the on-premises Connector, that the site pairing and service mesh are healthy, and that the provider has exposed the cloud inventory for outbound migration. Check whether the VM uses an unsupported device or belongs to a protected management appliance category.
Cutover stops for capacity or compatibility
Recheck free space on the destination datastore, temporary overhead, snapshots and swap. Compare the VM’s virtual hardware, disks, encryption and source datastore type with the supported matrix. Do not force a job through a failed validation; resize or select an approved destination only after the change is documented.
Destination starts but applications fail
Test DNS, routes, firewall policy, identity services and dependent databases separately. If the network was extended, verify that only the intended site owns the extension and that no Layer 2 loop or duplicate gateway exists. Use the rollback plan if the application cannot be restored within the agreed window.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Special planning for vCenter, NSX and other appliances
HCX is for supported workload VMs, not a universal relocation mechanism for the VMware control plane. vCenter Server and NSX Manager are VMware software appliances that HCX cannot migrate. Treat their relocation as a separate architecture project with its own supported upgrade, backup, replacement or federation procedure. Keep management services available while moving application workloads and verify that the destination HCX and vSphere control paths do not depend on an appliance being moved in the same batch.
How to decide whether the move is ready
Proceed only when each answer below is documented:
- The provider confirms reverse migration and identifies the initiating HCX Manager.
- HCX, vSphere, ESXi, licensing and datastore compatibility are confirmed for the exact pair.
- The destination cluster and datastore have tested capacity, including temporary migration overhead.
- Routes, firewall rules, MTU, DNS and any Layer 2 extension sequence have been validated.
- A representative test VM completed replication, cutover and application checks.
- Owners accept the outage window and understand that Bulk Migration still includes power-off and offline synchronization.
- Backups, monitoring, security, rollback and sign-off responsibilities are assigned.
- Management appliances are excluded from the workload batch and have a separate supported plan.
HCX documentation cautions that a migration cannot be guaranteed under every infrastructure and network condition. The practical response is controlled scope, a tested runbook, observable checkpoints and a rollback decision—not a promise of zero downtime.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
