You can configure and use the NSX Edge in a vSAN stretched cluster in active/active mode.

Keep in mind the following considerations, advantages, and limitations when planning to use NSX networking on a vSAN stretched cluster. You can use two topology options to deploy the NSX Edge Node VMs.
Consolidated NSX Edge Node and vSAN Stretched Cluster topology
The consolidated topology places the NSX Edge VMs in the same compute cluster as the supervisor and workloads. This topology is cost-effective as lesser hosts are required compared to the shared topology. The disadvantage of this topology is the complexity of implementation and it has overlay performance implications if third-party VXLAN is used on the stretched physical network. Its also has a single point of failure.
Shared NSX Edge topology
In this topology, all workloads across all Supervisors use a single shared NSX Edge VM in a separate compute cluster. This topology ensures pNIC availability as it has dedicated hosts. The cost is higher compared to the consolidated topology as more hosts are required. The disadvantage of this topology is that underlay will need to be stretched across all top of the rack (TOR) switches.

Requirements of the Consolidated NSX Edge Node and vSAN Stretched Cluster topology

To use the consolidated topology, verify that you meet the following requirements:
  • Verify that the host compute resources are available for the NSX Edge nodes.
  • Enable BGP across multiple TOR switches in the two sites.
  • Verify that all relevant NSX Distributed Virtual Port Group (NSX DVPG) required for the NSX Edge (Edge TEP, uplink TEP, Edge management) are stretched in the underlay physical network using third-party VXLAN.
  • Create host/NSX Edge VM groups and affinity rules to pin the NSX Edge nodes to site 1 or site 2 to prevent migration of NSX Edge node VMs across the two sites.

Requirements of the Shared NSX Edge topology

The shared topology is the preferred topology. Configure an Edge cluster at each site. To use the shared topology, verify that the following requirements are met:
  • Enable BGP across multiple TOR switches in the two sites.
  • Create host/NSX Edge VM groups and affinity rules to pin the NSX Edge nodes to site 1 or site 2 to prevent migration of NSX Edge node VMs across the two sites.

Placement of NSX Components in an Active/Active Deployment

NSX Manager VMs
NSX Manager VMs are generally deployed in a separate management cluster which is outside the Supervisor or workload cluster.

Deploy the NSX Manager VMs in a separate management cluster that has a vSAN stretched cluster topology. Deploy the NSX Manager VMs evenly across site 1 and site 2 of that cluster.

NSX Edge VMs
Create NSX Edge VM groups and deploy the Edge VMs evenly across site 1 and site 2 for both shared and consolidated topologies.

Host Affinity Rules for NSX Components in an Active/Active Deployment

NSX Manager VMs
Perform the following steps:
  1. Create a VM group for pinning the NSX Manager VMs to site 1 and another VM group for pinning VMs to site 2. For example, NsxMgrVmGroup-A and NsxMgrVmGroup-B.
  2. Create a DRS host group with all the ESXi hosts of site 1 and another DRS host group with the ESXI hosts of site 2. For example, HostGroup-A and HostGroup-B.
  3. Create a should VM-Host affinity rule for NsxMgrVmGroup-A and HostGroup-A.
  4. Create a should VM-Host affinity rule for NsxMgrVmGroup-B and HostGroup-B.
NSX Edge VMs
Perform the following steps:
  1. Create a VM group for half the number of NSX Edge VMs. For example, NsxEdgeVmGroup-A.
  2. Create a VM group for the remaining NSX Edge VMs. For example, NsxEdgeVmGroup-B.
  3. Create a should VM-Host affinity rule for NsxEdgeVmGroup-A and HostGroup-A.
  4. Create a should VM-Host affinity rule for NsxEdgeVmGroup-B and HostGroup-B.