Cisco UCS Mini – Connecting An External Server

Today I had a rather interesting task of attempting to connect an existing Hyper-V Server to a new Cisco UCS that I have been working on, so that I could basically get the standalone host to communicate with the Cisco UCS blades using Live Migration.  The basic requirement for doing this is so that I can join the Cisco UCS blades into the existing 2012 R2 Cluster, and then start to Live Migrate the Virtual Machines from old to new hardware.  So the scenario for this is:

  • I have a pair of Dell Servers that are in a 2012 R2 Hyper-V Cluster, that have a single 10Gb cable between them for Live Migration & CSV
  • I have a Cisco UCS mini chassis along with 3 blades built to Windows Server 2016
  • I want to be able to join the new blades into the existing 2012 R2 Hyper-V Cluster, and have the ability to Live Migrate any existing virtual machines from old Dell Server to the new Cisco UCS blades

After a lot of whiteboarding with a fellow colleague, we managed to find a solution.  Thankfully our existing Dell Servers had a spare 10Gb port available, so what we did was to cable one of the Dell Servers into one of the spare ports into Fabric A onto the UCS.  Initially we configured the port as a Server port – seemed to make sense at the time as we were connecting an external Server to the Cisco UCS chassis.  What we also wanted to do was to pin the Live Migrate vNIC template to the uplink to the Server, so that we could correctly segregate out the traffic.  However, after configuring the port as a Server port we found that we couldn’t pin the uplink as I guess that is not how Server ports within the Cisco UCS actually work.  With my last Cisco UCS deployment being well over a year, I suddenly thought that the UCS port connected to the Server should be an Appliance port.  After configuring the port to be an appliance port, and setting the correct IP on the new NIC within the Blades I could then see the Live Migration network on the Dell Server!  Rejoice \o/ !

The last piece for me is to sort access to the Fibre Channel storage, and assign the Cisco Blades the LUNs which the Dell Servers currently have access to.  Hopefully then, I can join the new Blades into the existing 2012 R2 Cluster, Live Migrate the Virtual Machines from old to new and decommission the older Servers.  Then all being well I can upgrade the Cluster level to 2016, and start enjoying all the fantastic benefits that Server 2016 has to offer.  Also thankfully I have a great backup product in place (Veeam!), that already supports Windows Server 2016 and I therefore have no worries!

I will post a quick how to for the above shortly, but to be honest I was quite surprised as to how easy it was.  If anyone has any questions or comments, please feel free to ask.