Your NSX Manager cluster configuration will vary depending on whether your deployment is for single, dual, or multiple sites.

You can use vSphere HA with NSX-T Data Center to enable quick recovery if the host running the NSX Manager node fails.

Note: See Creating and Using vSphere HA Clusters in the vSphere product documentation.

Cluster Requirements

  • In a production environment, the NSX Manager cluster must have three members to avoid an outage to the management and control planes.

    Each cluster member should be placed on a unique hypervisor host with three physical hypervisor hosts in total. This is required to avoid a single physical hypervisor host failure impacting the NSX control plane. It is recommended you apply anti-affinity rules to ensure that all three cluster members are running on different hosts.

    The normal production operating state is a three-node NSX Manager cluster. However, you can add additional, temporary NSX Manager nodes to allow for IP address changes.

    Important: As of NSX-T Data Center 2.4, the NSX Manager contains the NSX Central Control Plane process. This service is critical for the operation of NSX. If there is a complete loss of NSX Managers, or if the cluster is reduced from three NSX Managers to one NSX Manager, you will not be able to make topology changes to your environment, and vMotion of machines depending on NSX will fail.

  • For lab and proof-of-concept deployments where there are no production workloads, you can run a single NSX Manager to save resources. NSX Manager nodes can be deployed on either ESXi or KVM. However, mixed deployments of managers on both ESXi and KVM are not supported.

Single Site Requirements and Recommendations

The following recommendations apply to single site NSX-T Data Center deployments.

  • It is recommended that you place your NSX Managers on different hosts to avoid a single host failure impacting multiple managers.
  • Maximum latency between NSX Managers is 10ms.
  • You can place NSX Managers in different vSphere clusters or in a common vSphere cluster.
  • It is recommended that you place NSX Managers in different management subnets or a shared management subnet. When using vSphere HA it is recommended to use a shared management subnet soNSX Managers that are recovered by vSphere can preserve their IP address.
  • It is recommended that you place NSX Managers on shared storage also. For vSphere HA, please review the requirements for that solution.

You can also use vSphere HA with NSX-T to provide recovery of a lost NSX Manager when the host where the NSX Manager is running fails.

Scenario example:
  • A vSphere cluster in which all three NSX Managers are deployed.
  • The vSphere cluster consists of four or more hosts:
    • Host-01 with nsxmgr-01 deployed
    • Host-02 with nsxmgr-02 deployed
    • Host-03 with nsxmgr-03 deployed
    • Host-04 with no NSX Manager deployed
  • vSphere HA is configured to recover any lost NSX Manager (e.g., nsxmgr-01) from any host (e.g., Host-01) to Host-04.
Thus, upon the loss of any hosts where a NSX Manager is running, vSphere recovers the lost NSX Manager on Host-04.

Dual Site Requirements and Recommendations

The following recommendations apply to dual site (Site A/Site B) NSX-T Data Center deployments.

  • It is not recommended to deploy NSX Managers in a dual-site scenario without vSphere HA. In this scenario, one site requires the deployment of twoNSX Managers and the loss of that site will impact the operation of NSX-T Data Center.
  • Deployment of NSX Managers in a dual site scenario with vSphere HA can be done with the following considerations:
    • A single stretched vSphere cluster contains all the hosts for NSX Managers.
    • All three NSX Managers are deployed to a common management subnet/VLAN to allow IP address preservation upon recovery of a lost NSX Managers.
    • For latency between sites, see the storage product requirements.
Scenario example:
  • A vSphere cluster in which all three NSX Managers are deployed.
  • The vSphere cluster consists of six or more hosts, with three hosts in Site A and three hosts in Site B.
  • The three NSX Managers are deployed to distinct hosts with additional hosts for placement of recovered NSX Managers:
    Site A:
    • Host-01 with nsxmgr-01 deployed
    • Host-02 with nsxmgr-02 deployed
    • Host-03 with nsxmgr-03 deployed
    Site B:
    • Host-04 with no NSX Manager deployed
    • Host-05 with no NSX Manager deployed
    • Host-06 with no NSX Manager deployed
  • vSphere HA is configured to recover any lost NSX Manager (e.g., nsxmgr-01) from any host (e.g., Host-01) in Site A to one of the hosts in Site B.
Thus, upon failure of Site A, vSphere HA will recover all NSX Managers to hosts in site B.
Important: You must you properly configure anti-affinity rules to prevent NSX Managers from being recovered to the same common host.

Multiple (Three or More) Site Requirements and Recommendations

The following recommendations apply to multiple-site (Site A/Site B/Site C) NSX-T Data Center deployments.

In a scenario with three or more sites, you can deploy NSX Managers with or without vSphere HA.

If you deploy without vSphere HA:
  • It is recommended that you use separate management subnets or VLANs per site.
  • Maximum latency between NSX Managers is 10ms.
Scenario example (three sites):
  • Three separate vSphere clusters, one per site.
  • At least one host per site running NSX Manager:
    • Host-01 with nsxmgr-01 deployed
    • Host-02 with nsxmgr-02 deployed
    • Host-03 with nsxmgr-03 deployed
Failure scenarios:
  • Single site failure: Two remaining NSX Managers in other sites continue to operate. NSX-T Data Center is in a degraded state but still operational. It is recommended you manually deploy a third NSX Manager to replace the lost cluster member.
  • Two site failure: Loss of quorum and therefore impact to NSX-T Data Center operations.

Recovery of NSX Managers may take as long as 20 minutes depending on environmental conditions such as CPU speed, disk performance, and other deployment factors.