Adding support to deploy LPARs with VNIC failover and 4 backing devices

A couple of customers raised the question, how they can deploy LPARs with more than 2 backing devices through the PowerVC guy. From my actual state of information, this is only possible by using the the OpenStack API of PowerVC.

The customers who had asked for this, usually want to deploy LPARs with 4 backing devices to archive the same level of redundancy like they had before with SEA failover technology and LACP with 2 ports through the GUI.

They want to have the VNIC failover deployed the following way:

- 2 VFs must be created on NIC ports attached to network switch A

- 2VFs must be created on NIC ports attached to network switch B

- In a dual VIOS architecture, the VNIC Server adapter of 1 VF from switch A and B must be served by each VIOS

In my opinion, the easiest way to implement this would be a logic like (would apply for API and GUI):

- Label all ports attached to switch A with: SW_A

- Label all ports attached to switch B with: SW_B

- In the PowerVC network definition I define the primary SR-IOV label with SW_A and for the redundancy SW_B

- During deployment (GUI, or API) I request the overall amount of VFs I want to have and PowerVC equally distributes them across the ones I defined in the network as primary and redundant adapter. Then it ensures that 1 VFs per primary and redundant adapter are attached to each VIOS. 

Maybe we can define how PowerVC is distributing VFs by something like we have it for "Fabric Redundancy" in the Storage Connectivity Group, to offer more flexibility in the future?