We have Rack server with two VIC 1225, managed by UCSM(single wire management), Single Wire Management of rack serveer only supports single NIC. Is there anyway to use second NIC as redundancy , in case one VIC 1225 fail?
Thanks in advance for your help.
It should be possible to use multiple VIC 1225 with C-Series integration. However, with the VIC it must be connected to the Fabric in the UCS through the FEX.
There is no external tool (not managed by UCS), to configure the VIC, once the server is integrated into the UCSM.
Therefore, any VIC would have to be connected to the system or it would be imporoperly connected to manage the links.
When VIC card fails, vnics configured would not be moved to other VIC. You can achieve redundancy by combination of Fabric Failover and NIC teaming at the OS level.
For example, we create one vNIC on each VIC card and configure fabric failover to provide redundancy for upstream FEX failure and NIC teaming at OS level for VIC card failure.
VIC 1 - A1 and A1"
VIC 2 - B1 and B1"
A1 and B1 are teamed at OS level and A1" and B1" are fabric failover vNICs that are not visible to OS.
Please refer to server datasheets on server models that supports two VIC cards.
Thanks for quick response. Please refer page 6 to 13 of below link. All scenario is just using single NIC card. None of scenario refers to using multiple physical NIC in singe cable mgmt.
Thanks for your help.
I was going through the above example, I couldn't understand why you are suggesting to use Fabric Failover and Teaming. How will this work can you please explain.
I am having similar scenerio I have VIC 1225T and VIC1227T installed in C220 M4 servers, to avoid VIC card SPOF I will make a bond using 1 VNIC from each VIC. I hope this thing will work if yes than what is the purpose of enabling Fabric Failover(What is the use of those two VNIC's A" & B")
Thanks are regards
I think fabric failover is a rather old artifact, from the time (Windows Server 2008) when it was not possible for this OS to have 2 vnic's.
Today, with W2012 and newer this is no more the case.
Therefore your reasoning is 100% correct ! Use the OS teaming, and avoid to enable in addition failover; 2 control loops are one too much.
Is rack server with two VIC cards managed by UCSM ?
To achieve both VIC and path redundancy, you can team the NIC in OS without enabling fabric failover for the vNICs ( team vNIC created from each VIC card )
# applies to rack connected to FI / UCSM
If you have single VIC card, then Fabric failover is preferred as it eliminates config overhead in the OS
in my opinion refers to standalone (not UCSM managed) C-series servers.
failover is a feature which only exists for UCSM managed servers (during the definition of a service profile).
Even with a single I/O adaptor, one would not anymore use failover today (don't forget this means load on your FI's).
yes, fabric failover feature is available in UCSM managed mode.
I shared the doc as I am not clear about the deployment.
Fabric failover is recommended in single VIC card scenario to achieve the redundancy