You can configure ESXi hosts in your network to measure: pNIC to vNIC, vNIC to pNIC, vNIC to vNIC, and VTEP to VTEP network latency statistics.

Configuration is supported only using the NSX REST APIs. The steps in the following procedure list the management plane APIs that you must run to configure the calculation of various network latency statistics. For a detailed information about the API schema, example request, example response, and error messages of all the APIs, you must read the NSX-T Data Center API Guide.


Both vCenter-managed hosts and standalone ESXi hosts that you want to configure for measuring network latency statistics must be prepared for NSX-T Data Center. That is, NSX-T Data Center components must be installed on all the ESXi hosts in your network.


  1. To compute vNIC to vNIC, pNIC to vNIC, and vNIC to pNIC network latency statistics, do these steps:
    1. Create a latency profile with the following POST API:
      POST https://<NSX-Manager-IP>/api/v1/latency-profiles

      By default, vNIC to vNIC latency is measured for all the vNICs on the host transport node.

      In the request body of this API, configure the following information:
      • Enable or disable pNIC latency on the host. When it is enabled, pNIC to vNIC and vNIC to pNIC latency are calculated for each vNIC on the host transport node.
      • Specify either the sampling rate or the sampling interval.

        If you configure both, sampling interval takes a precedence.

    2. Create a NSGroup with transport nodes as the target type in the NSGroupsSimpleExpression with the following POST API:
      POST https://<NSX-Manager-IP>/api/v1/ns-groups

      If you have Manager mode enabled in your UI, you can use the UI to create the NSGroup, and specify the transport nodes in the membership criteria.

    3. Create a Service Configuration profile with the following POST API:
      POST https://<NSX-Manager-IP>/api/v1/service-configs

      This API combines the latency profile and the NSGroup that you created in the previous steps.

  2. To measure VTEP to VTEP latency statistics, enable latency in the BFD health monitoring profile, which is a resource type in the transport zone profile. Run either the following PUT or the POST API:
    • POST https://<NSX-Manager-IP>/api/v1/transportzone-profiles
    • PUT https://<NSX-Manager-IP>/api/v1/transportzone-profiles/<transportzone-profile-id>

What to do next

Export the statistics to an external collector for a deeper network insight and troubleshooting network-specific latency problems.