Already given in the previous replies, the Hyper-V switching is only providing with an intermediate layer. In order to get an easier understanding of it all, enabling a virtual switch on top of a physical NIC (or NIC team) is actually creating a switch.
The switch in question has an uplink which is connected to the physical interface. By enabling for the management OS to be allowed to connect to the switch, this is virtually creating a virtual NIC in the management OS with the same name as the virtual switch.
That NIC is connected to virtual switch and inherits the network configuration of the physical NIC to which the switch uplink is connected.
The effect of this is that if utilising more than one NIC on a server, the best way of simplifying interface aggregation and getting an embryo of a failover solution, is to team all the NICs that are connected to the same network in a "Switch Independent"
teaming configuration. This is, given that the different cables are connected to different switches. This will present a single teamed interface to the management OS and then that interface can be utilized to create a virtual switch.
Apart from creating an external switch, you can also create internal switches if there is a need for keeping IP addressing inside of the Hyper-V host but to also have a single VM to provide for routing if the routing capabilities is what you are
asking for specifically.
This answer is provided as is and based on my personal understanding of your current situation. If something in this mapping doesn't apply to your specific situation, there may or may not be discrepancies in how to apply the potential solution provided above.
Yours
Richard Johansson
Hyper-V Specialist Consultant