Cisco Nexus 9000 switch is powerful Layer2 and if needed can be Layer3 switch, supporting standard routing protocol including BGP. In general, its position in networking architecture is in Data Centers access leaf switch or as a border switch.
Depends on customer needs it can be positioned in other network environment as well, like access or aggregation switch in service provider or enterprise environment.
Of course for service provider environment more proper switch would be switches with MetroEthernet feature set or Catalyst switch in traditional Enterprise environment. Still nothing stops you to use Nexus 9000 as a pure Layer2 access/aggregation switch for in SP environments.
And this switch works perfectly in such environment before you send multi-tagged frames to pass frames through it. Multi-tagged frames are expected in SP world as a result of e.g. QinQ technologies, where we want to one or more CE VLANs transport over SP environment, being preserved and not overlapped with other customers VLANs, adding SP VLAN on the TOP of CE VLAN.
In this way we get Layer2 frame with two tags. The most TOP tag is service provider tag, and inner tag is customer CE VLAN tag. Very useful feature. SP needs then to manipulate with SP tag only, not any more interested in CE tag before it reaches egress SP device.
What in case such multi-tagged frames need to pass through Nexus 93180YC-FX switch from left to the right side, transparently without any TAG manipulation. Well by default, Nexus 93180YC-FX doesn' t not know how to process multi-tagged frame and it will drop it. In Lab, I had 93180YC-FX for a test, and here are some solutions for processing multi-tagged frames.
To avoid dropping multi-tagged frames, there are two possible solutions:
1. To configure allow multi-tag on access interface
2. To configure globally system dot1q-tunnel transit
First option is not supported by Cisco in pure VLAN environment, even it works perfectly and no issues were seen in a Lab. Supported environment for this solution would be with VxLAN configuration.
Second option is supported by Cisco, tested and works perfectly. Once applied, frames were propagated correctly without drops. As a sub-option of this command you can define what outer tag to be tunneled and processed by switch.