Create Subnet in Multiple Clusters from PC

No. You need a Multi Cluster Virtual Switch for that. More info here: https://www.jeroentielen.nl/nutanix-multicluster-virtual-switch-management/

So what is the solution? do you have to create the same subnet 10 times even though you are doing it from PC? Does the cluster selection drop down not let you check a box next to each VS/Cluster you want the subnet on?

Don’t have the free NICs available to create a MCVS.

Previously each subnet had to be created on each virtual switch for each cluster (vs0), That is still the case if you want a subnet to only existing on a specific cluster.

If you need a subnet that spans across multiple clusters, that’s where the multi-cluster vswitch comes into play. 

To answer your question “do you have to create the same subnet 10 times even though you are doing it from PC?”, Yes or you can use an automation approach such as script to do this for you. 

If I only have 2 NICs is there anyway to migrate to a MCVS one NIC at a time? Similar to how you can migrate a vSphere standard switch to a DVS one NIC at a time?

That’s something I am looking to test in May as I am in the same boat. Only 2 NICS and I want to move over to the mvcs. 

I am confused as this documentation states “Prism Central deploys the Network Controller that provides Flow Virtual Networking (application-driven network virtualization) as well as centralized VLAN management.”

In my mind Centralized VLAN management means being able to create/update/delete a VLAN across all cluster managed my PC.

This is in this guide - https://portal.nutanix.com/page/documents/details?targetId=Nutanix-Flow-Virtual-Networking-Guide-v7_0_0:ear-flow-nw-overview-pc.html

Just a little bit above that, you will see the 

“Network Controller

The Network Controller is the networking component of Prism Central that manages and controls configuration, monitoring and optimization of network resources for Flow Virtual Networking VPCs and VLAN subnets. It provides programmability, automation, and centralized control for configuring and managing network flows.

Network Controller is necessary to use centralized VLAN management, Flow Virtual Networking and Flow Network Security Next Generation.”

So the network controller needs to be deployed to enable you to use centralized VLAN management. 
The Underlying ‘switch’, either single vs0 or multi-cluster are configured independently of flow networking. 

Network Controller was deployed automatically with version 7.5 and XL size PC so does this mean we can create a single VLAN and have span all clusters?

So there was a similar post on reddit that is similar to this question.

https://www.reddit.com/r/nutanix/comments/1syurwp/prism_central_75_and_centralized_vlan_management/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button

This was posted by someone who works on the solution:

“Disclaimer: I work on this feature area at Nutanix.

Instead of the multi-cluster virtual switch, for this use case I'd look at automation with either aCLI or API to make sure all the VLANs are created the same across all the clusters.

For the NCI 7.5 release, multi-cluster virtual switch does have some limitations with cross-cluster live migration, so I'd read up on it first before going through with a deployment. You do need another set of physical interfaces for the multi-cluster virtual switch, and it's awesome for scenarios where you want to do something like share a VPC External Subnet that holds NAT and NoNAT hosts across clusters. The MCVS is also good to have a unified IPAM across clusters for VLANs.

Once the CCLM limitations are ironed out, then MCVS should be better for more scenarios.

The other comment I'd have about sharing VLANs across clusters is that you should consider both the number of VMs in a single layer 2 broadcast domain and MAC and IP address conflicts if those broadcast domains are shared. You can check out the MAC prefix feature to create a unique MAC prefix per AHV cluster if you don't go down the MCVS path.”

You need 4 nics per node. (if you want to use a bond). Because the internal (AHV/AOS) interfaces cannot be part of a multicluster virtual switch. 

Since you only have two (2) then I would advice not to use the multicluster virtual switch and YES, then you need to make subnets per cluster. That is why I ALWAYS suggest to take at least 4 interfaces in a node.