VXLAN Mapping Multiple VLANs to a Single VNI

On most network devices, it is not possible to assign multiple VLAN IDs to the same VNI on the same VTEP. Each VLAN is mapped to a unique VNI within a given VTEP, maintaining a strict one-to-one VLAN-to-VNI mapping. For Cisco Nexus devices, this is indeed the case, but some other vendors may implement it slightly differently, allowing for certain exceptions.

It's important to note that there are ways around this limitation, such as using QinQ, where you use a VLAN ID that is mapped to the VNI on the outer tag, and then you can use whatever you want on the inner tag.

It is possible to assign multiple VLAN IDs to the same VNI on different VTEPS, but with some important restrictions. This setup is known as “VLAN normalization,” where different VLAN IDs on separate VTEPs are mapped to a common VNI.

For example:

  • Traffic from VLAN 100 on VTEP1 enters VXLAN and is encapsulated in VNI 5000.
  • When it reaches VTEP2, it is decapsulated and forwarded into VLAN 200.

This works because VXLAN is “agnostic” to the local VLANs once the traffic is encapsulated. This means that the VNI becomes the “real” identifier of the Layer 2 domain in the overlay. In such a scenario, VLAN 100 on VTEP1 and VLAN 200 on VTEP2 are actually part of the same Layer 2 domain.

For more information, take a look at VLAN to VNI Mapping.

https://networklessons.com/vxlan/introduction-to-virtual-extensible-lan-vxlan

https://networklessons.com/vxlan/vxlan-static-ingress-replication

https://networklessons.com/switching/802-1q-tunneling-q-q-configuration-example

Links to this page: