Before deploying HCX, consider the required resources for both the source and the destination environment site managers.

Virtual Hardware Requirements for HCX Manager Appliances

Appliance

vCPU

Memory

Disk Space/IOPS

HCX Manager

4

12 GB

60 GB

HCX Interconnect (HCX-IX)

8

6 GB

2 GB

HCX Network Extension (HCX-NE)

8

3 GB

2 GB

HCX WAN Optimization (HCX-WAN-OPT)

8

14 GB

100 GB / 5000 IOPS

HCX Sentinel Gateway (HCX-SGW [source only])

8

8 GB

  • HCX:

    21 GB (disks: 2 GB, 6 GB, 4 GB, and 9 GB)

HCX Sentinel Data Receiver (HCX-SDR [destination only])

8

8 GB

  • HCX:

    21 GB (disks: 2 GB, 6 GB, 4 GB, and 9 GB)

Note:

The storage requirement per appliance is doubled during the upgrade and redeploy operations, as a second appliance is created for operation.

Migration Host and Datastore Requirements

During a virtual machine migration operation, the destination compute and storage resources are selected. These resources must meet the following general requirements:

Requirement ID

Requirement

Description

Host-Req-1

Host CPU and Memory Capacity

The destination host must support the original vCPU and Memory configurations of the virtual machine.

Host-Req-2

Host CPU and Memory Usage

The destination host capacity must satisfy admission control requirements.

Host-Req-3

Host Compatibility with VMware Tools

The destination hosts must support the VMware Tools version of the original virtual machine. For more information, see the VMware Certified Compatibility Guides.

Host-Req-4

Host Compatibility with VM Hardware

The destination hosts must support the VM Hardware version of the original virtual machine. For more information, see VMware KB 2007240.

Datastore-Req-1

Datastore Type

The destination datastore type must be VMFS5, VMFS6, NFS3, NFS4.1, vSAN, or vVOL.
Note: vVOL is not supported for RAV.

Datastore-Req-2

Datastore Capacity

The destination datastore capacity must be able to store the following:

  • Complete original virtual machine data (everything that makes up the virtual machine)

  • Additional replica instance vmdk file on target datastore (applicable for Bulk or RAV Migration during the delta sync)

  • Redo logs (virtual machine data changes over a one hour interval)

  • Additional data for multiple point in time (MPIT) recovery points

  • Full copy of the virtual machine when using test recovery

  • With VSAN, additional data required by the Primary level of failures to tolerate (PFTT) configuration

System Requirements for Scaling Out Deployments

Consider the following requirements when scaling out HCX deployments:
Note: For information regarding the scaling of HCX resources, see VMware KB article 93605.
  • When scaling HCX deployments, the hardware requirements apply for each additional appliance. For appliance limit information, refer to the VMware Configurations Maximums.

  • In environments with multiple vCenter Servers, a maximum of one site manager can be deployed per vCenter Server.

  • By default, the migration service encompasses all source and destination clusters in a service mesh with a single IX appliance. The migration service can be scaled out when there are additional clusters in either the source or destination environments (you can create a service mesh for each unique source/destination cluster pair).

    For example, if there is source site with clusters A and B paired to a destination cloud with clusters X, Y, and Z, you can scale out by creating service meshes AX, AY, AZ, BX, BY, BZ.

    Note: In this scale out example, cluster A is hosting three IX appliances, and you can configure host anti-affinity to enable parallel switchovers with HCX vMotion and RAV migrations.
  • Multiple Network Extension appliances can be deployed per distributed virtual switch (DVS) or NSX Transport Zone.

  • When using Network Extension High Availability (HA), a maximum of two Network Extension appliance can be deployed for each VLAN or NSX segment. In the HA group, one appliance is Active and one is Standby.

    Note:

    HA Standby appliances must be considered when calculating the overall hardware resource requirements.

  • When not using Network Extension HA, a maximum of one Network Extension appliance can be deployed for each VLAN or NSX segment.

  • The Network Extension appliance count in the Service Mesh configuration must equal the sum of all planned standalone, HA Active, and HA Standby appliances.