In this model, multiple vSphere clusters host the NSX Edge devices and tenant workloads. When you plan to expand the physical infrastructure that is available for workloads, you distribute edge devices across each of the compute clusters involved.

As you deploy building blocks, you deploy additional NSX Edge devices in each building block to handle the north-south traffic into and out of the SDN for the workloads deployed on the new block.

Figure 1. Multiple Shared Edge and Compute Clusters


  • This model limits the impact on network traffic when an outage of a cluster occurs because only certain NSX Edge devices become unavailable. If you use a dedicated edge cluster model, then an outage of the edge cluster impacts the routing of all workloads.

  • This model supports network separation and limits broadcast domains, and can reduce the scope of transport zones.
  • Distributing edge devices across compute clusters can be with multiple clusters inside a single workload domain or when creating multiple workload domains. When multiple workload domains exist in an environment that uses NSX for vSphere, then this model is more cost effective than a dedicated edge cluster model.

  • To optimize traffic, each cluster can have its own set of networks, with the edge gateway for each of those networks being located on the same cluster.

  • Consider this topology when horizontal movement of workloads between clusters is not required. Horizontal movement would create an inefficient and unequal traffic flow between workloads and their gateways according to the cluster the workload was on.

  • Evaluate the benefits of the extra separation and distribution against the cost of routing any traffic between clusters in a north-south manner by using edge gateways compared to being able to route the same traffic in an east-west manner .

  • You must provide more network interfaces to the upstream network than with a dedicated edge or single shared edge and compute cluster design.