Ferdinand,
Let me add to this discussion to help clarify a few things.
Server platforms with a single CIMC interface will have the CIMC communication shared with a single LOM interface - usually LOM1. Platforms with a single CIMC interface (Ex C200) can only use "None" for the CIMC redundancy mode. Servers with dual CIMC interfaces (Ex C260) can share both CIMC interfaces with both LOM ports respectfully. Dual CIMC systems can take advantage of Active/Active, Active/Passive and None redundancy modes when configured for Shared LOM.
When dual CIMC interfaces are in A/A redundancy mode, they operate with TLB (Transmit Load Balancing). This allows both interfaces to be used for transmit, but only one for receive. This requires that upstream the interfaces are NOT port channeled. In A/P mode only a single interfaces is used for send & receive until there is a failure and it fails over to the standby interface.
When using shared LOM with ESX, you should be using the default vSwitch hashing mode (Virtual Port ID). This does not require any upstream port channeling. This allows multiple vSwitch uplinks to be used simultaneously and allow CIMC traffic to operate as it must. Remember, the CIMC interfaces can't be port channeled, so the physical interfaces must operate individually upstream as well.
The CIMC IP and ESX Management or VM Traffic do NOT need to be in the same subnet. The CIMC can be configured for VLAN tagging so you can logically separate your CIMC managment traffic from your ESX & VM traffic. The one requirement is that both INTERFACES must allow the CIMC VLAN. That's what the doc means by "must be in same subnet". From a vSwitch level the interfaces don't even need to be on the same vSwitch when the CIMC is sharing the LOM, as long as upstream you're allowing the CIMC VLAN on each physical uplink.
Regards,
Robert
