During the initial VxRail cluster build, IP addresses that you enter are assigned to the VxRail components in the External Management network. Optionally, they are assigned to VxRail components in the vCenter Server management network. The IP addresses must follow certain rules:

• The IP address scheme must be a public IP address range.
• The IP address must be fixed (no DHCP).
• The IP addresses cannot be in use.
• The IP address range must all be in the same subnet.

You have some flexibility in how the IP addresses are assigned to the VxRail management components. For the ESXi hosts:

• You can manually assign the IP addresses to the ESXi hosts.
• You can have the IP addresses automatically assigned during VxRail initial build.

The decisions that you make on the final VxRail configuration that is planned for your data center impacts the number of IP addresses you must reserve:

• Decide if you want to reserve additional IP addresses for VxRail management to assign to VxRail nodes in the future for expansion purposes in a single rack. When a new node is added to an existing VxRail cluster:

• It assigns an IP address from the unused reserve pool, or:
• If no addresses are in reserve and unused, it prompts you to enter an IP address manually.

• Decide whether you want to use the vCenter instance that is deployed in the VxRail cluster or use an external vCenter already operational in your data center. 

• For VxRail versions 7.0 or later, you must reserve an IP address for vCenter if you use the vCenter instance that is deployed on the VxRail cluster. The Platform Service Controller is bundled into the vCenter instance. 
• For VxRail versions earlier than version 7.0, reserve an IP address for the vCenter instance) if VxRail deploys vCenter. Also reserve an IP address for the Platform Service Controller (PSC) if VxRail deploys vCenter.

• Decide if you plan to use vSphere Log Insight that can be deployed in the VxRail cluster:

• For VxRail version 7.0 and earlier, if you use the vCenter instance that is deployed in the VxRail cluster, you can deploy vSphere Log Insight on the cluster. You can also choose to connect to an existing syslog server in your data center, or no logging at all. If you choose to deploy vSphere Log Insight in the VxRail cluster, you must reserve one IP address.
• vRealize Log Insight is not an option for deployment during the initial VxRail configuration process starting in version 7.0.010.  
• If you use an external vCenter already operational in your data center for VxRail, vSphere Log Insight cannot be deployed.

• If you plan to deploy a VxRail cluster that requires a Witness at a remote third site, two IP addresses are required to deploy the witness virtual appliance. For example, a VxRail stretched cluster or 2-Node cluster:

• One IP address is assigned to the witness management network.
• One IP address is assigned to the witness vSAN network. 
• Both networks must be able to route to the VxRail cluster requiring the remote site witness.

If the VxRail clusters are stretched clusters, an existing vSAN witness can be shared in your remote site. The vSAN witness can support vSAN datastores at version 7 Update 1 or later. 

• For a 2-Node Cluster, the VxRail nodes must connect to the Witness over a separate Witness traffic separation network. For this network, an additional IP address is required for each of the two VxRail nodes.

• The VxRail nodes must be able to route to the remote site Witness.
• The traffic must be able to pass through the Witness traffic separation VLAN.

Use the following table to determine the number of public IP addresses required for external connectivity:

Table 5.      Public IP addresses required for external connectivity

Request your networking team to reserve a subnet range that has sufficient open IP addresses to cover VxRail initial build and any planned future expansion.