In this part you add VCF Operations HCX, the component that moves workloads into your VCF 9.1 fleet. It is optional, and it is the tool you reach for when you want to migrate live virtual machines from an older vSphere or another site rather than rebuild them. It pairs well with the brownfield import path when workloads need to move rather than stay in place.

Install the HCX Manager from VCF Operations, pair it with the source site, define the network and compute profiles, then deploy the service mesh that carries the traffic. Plan the IP pools for the service mesh appliances before you start, because the mesh will not deploy without them.

Prerequisites

Step 1 create the HCX Manager DNS record

Prepare the name before you install.

1. Choose the HCX Manager FQDN.
2. Create a forward A record and a matching PTR record.
3. Confirm both resolve from the management network.

Step 2 install the HCX Manager

Download the binary in VCF Operations, then start the install from the domain.

1. In VCF Operations, open Build, expand Lifecycle, VCF Instances, and select your instance.
2. On the Binary Management tab, open Install Binaries, select VCF Operations HCX version 9.1.0.0 or later, and click Download.
3. Select the domain, open the Software Versions tab, and click Manage Components.
4. Locate the VCF Operations HCX tile and click Install.
5. On the General step, select the downloaded HCX Manager version 9.1.0.0 or later, then click Next.

Step 3 set deployment settings

Give the manager its identity and credentials.

1. On the Deployment Settings screen, enter the HCX Manager FQDN that matches the DNS A record.
2. Set the admin and root credentials, then click Next. The management domain drives the remaining tasks, so no further input is needed here.
3. Pass the Validation screen, confirm the Summary, then click Finish on the Next Steps screen.
4. Wait for the manager to finish deploying and reach a healthy state.

Step 4 create the site pair

Connect the two HCX Managers over the control plane. Initiate the pair from the source toward VCF 9.1.

1. Open the HCX interface at https://<hcx-manager-fqdn> and go to Sites.
2. Click New Site Pair.
3. Enter the remote HCX URL of the destination and its administrative credentials, connecting over TCP 443.
4. Click Create and confirm the pair shows a connected status.

Step 5 create network and compute profiles

Define where the mesh appliances live and which addresses they use.

1. Open Interconnect, click the Network Profiles tab, and click Create Network Profile for management and replication, uplink, and vMotion.
2. Give each profile an IP pool, prefix, gateway and MTU, check the matching service tags, and click Create.
3. Click the Compute Profiles tab and click Create Compute Profile.
4. Select the services, the service and deployment clusters, attach the network profiles, then click Finish.

Step 6 deploy the service mesh

Build the data plane that carries migrations.

1. Open Interconnect, click the Service Mesh tab, and click Create Service Mesh.
2. Select the paired source and destination sites and the matching source and target compute profiles, then click Continue.
3. Review the uplink, scale out, traffic engineering and encryption settings, name the mesh, then click Create Service Mesh.
4. Wait for the Interconnect and Network Extension appliances to deploy and show green.

Step 7 verify with a test migration

Prove the path before you move production.

1. Open Migration and click Migrate.
2. Select a small test virtual machine and a migration type.
3. Start the migration and confirm it completes and the VM runs on the target.

flowchart TD
A[Create HCX Manager DNS record] --> B[Install HCX Manager from VCF Operations]
B --> C[Set FQDN and credentials]
C --> D[Create the site pair over 443]
D --> E[Create network and compute profiles]
E --> F[Deploy the service mesh]
F --> G[Run a test migration]

Deploy the source HCX Manager

VCF Operations automates the destination install, but the source site sits outside VCF, so its HCX Manager needs a manual OVA deployment. Deploy the same 9.1 appliance on the legacy vCenter, then run its first time setup before you pair.

1. Download the HCX Manager OVA from the Broadcom support portal.
2. On the legacy vCenter, right click the target cluster, choose Deploy OVF Template, then select the OVA and name the VM.
3. Set the admin and root passwords and the network properties, hostname, static IP, prefix, gateway and DNS, then power the VM on.
4. Open https://<source-hcx-manager-fqdn>:9443 and log in with the admin account.
5. Enter a system name, select vSphere as the instance type, and connect the source vCenter. Leave the NSX Manager fields blank if the legacy site has no NSX.
6. Click Restart to apply, and wait for the dashboard to show vCenter and SSO connected.

Size the HCX appliances

Plan capacity for the manager and for the data plane appliances the service mesh creates. The VCF Operations HCX Manager appliance uses 4 vCPU and 12 GB of memory, and its services can take up to five minutes after deployment to start. The service mesh then adds an Interconnect and a Network Extension appliance on each side, placed on the deployment cluster and datastore you pick in the compute profile, so leave headroom there before you build the mesh.

Open the required ports

Confirm the firewall before you pair, because a blocked port stalls the site pair or the mesh deployment. TCP 443 carries the management control plane between the two HCX Managers. The data plane appliances need IPSec UDP 4500 for their tunnels and TCP or UDP 5201 for the built in Perftest diagnostics. Open these both ways, and check the Broadcom ports and protocols portal for the full list before a production wave.

Extend a source network

Once the service mesh shows green, the Network Extension appliance can stretch a source segment into VCF 9.1 so a migrated virtual machine keeps its IP address and gateway. This is what lets you move a workload without re addressing it, and it is the piece most migration plans depend on.

1. In the HCX interface, open Network Extension and click Extend Networks.
2. Select the source segments to stretch and confirm the target gateway details.
3. Submit the extension and wait for each network to show Extension complete.
4. Migrate the workloads while the network stays extended, then move the gateway to the target once the last VM has moved.

Keep an extension in place only while a wave is in flight. A stretched network still hairpins traffic back to the source gateway until you migrate it and cut the gateway over, so unextend each network as soon as its workloads land on the target fleet. Plan the extensions per wave rather than stretching everything at once, and confirm the uplink can carry the added east west traffic each extension puts on the tunnel.

Verify workload mobility

Confirm the site pair shows connected and the service mesh appliances are green. Run a test migration of a small virtual machine and confirm it completes and runs on the target fleet. A clean test migration is the signal that you can plan the real moves.

Plan the service mesh networks

The service mesh will not deploy without addresses, so plan the network profiles first. HCX needs a management network for the appliances, an uplink network that carries traffic between sites, a vMotion network for live moves, and a replication network for bulk transfer. Some designs collapse these onto fewer networks, but keep management and uplink separate at minimum. Size an IP pool for each profile with room for more than one service mesh if you will pair more than one source. Confirm the uplink path has the throughput and MTU your migrations need, because that link sets the pace of every move.

Choose the right migration type

HCX offers more than one way to move a workload, and the right one depends on the workload and the downtime you can accept. Match the type to the workload before you schedule a wave, and always prove the path with a small test migration first.

Notes and best practices

Plan the network profiles before you touch the service mesh, because the mesh simply will not deploy without addresses, and a half planned profile wastes a deployment cycle. Keep the management and uplink networks separate at minimum, and size each IP pool for more than one service mesh if you will pair more than one source site. Prove the whole path with a small test migration before you schedule a wave, so you find a blocked port or a slow uplink on a throwaway VM rather than a production one. Pick the migration type per workload rather than forcing everything through one method, since a live database and a stateless web VM want different approaches. Watch the uplink MTU and throughput, because that single link sets the pace of every move, and a mismatch there turns a quick migration into an overnight one.

Common errors and fixes

Site pair will not connect.
TCP 443 is blocked between the sites or the source endpoint URL is wrong. Open 443 both ways and confirm the source HCX endpoint and credentials.

Service mesh fails to deploy.
A network profile is missing addresses. Assign IP pools to the management, uplink, vMotion and replication profiles, then redeploy the mesh.

HCX Manager install stalls.
The FQDN has no DNS record or the credentials were rejected. Create the A and PTR records, confirm the credentials, and retry the install.

Migration is slow or drops.
The uplink profile does not have enough throughput or MTU. Confirm the uplink network and MTU match the design, and place the Interconnect appliance on the right path.

Common questions

Is HCX mandatory
No. It is an optional workload mobility component, used when you want to migrate running virtual machines into the fleet.

Where do I install the HCX Manager
From the VCF Operations HCX tile, choosing version 9.1.0.0 or later.

What port does the site pair use
TCP 443 for the management control plane between the sites.

What does the service mesh deploy
The data plane appliances, the Interconnect and the Network Extension, that carry migration and stretched network traffic.

References

• VCF Blog: Workload Mobility with VCF Operations HCX 9.1, Part 1 Deployment
• VCF Blog: VCF Operations HCX 9.1, Part 2 Service Mesh
• Broadcom TechDocs: About the HCX Service Mesh

About The Author

Dr. Pranay Jha

Dr. Pranay Jha is a Cloud and AI Consultant with 18+ years of experience in hybrid cloud, virtualization, and enterprise infrastructure transformation. He specializes in VMware technologies, multi-cloud strategy, and Generative AI solutions. He holds a PhD in Computer Applications with research focused on Cloud and AI, has published multiple research papers, and has been a VMware vExpert since 2016 and a VMUG Community Leader.

See author's posts