Hello Hyper-V experts,
This definitely falls under a "general" question, but I am trying to avoid some common pitfalls and I'm hoping that some of you have some excellent advice for me to save time/headaches.
I am tasked with turning a poorly configured 2-node Hyper-V "instance" into a true failover cluster. Basically, "Company A" has 2 Hyper-V servers (Hyper-V Server 2008 R2...Core) running completely independent of each other. There is a fibre switch and a SAN behind them, but no zoning is performed in the fibre switch and also (naturally) the LUNs are not shared between the two servers, but rather statically mapped individually to their respective server.
My task is to correct the LUN mappings, create zoning in the fibre switch, and enable failover clustering. There are about 15 VMs running between the two Hyper-V nodes, both Windows and Linux.
The Hyper-V servers themselves have 4 NICs, but they are configured in the worst possible fashion. Each NIC is configured as its own virtual switch and virtual machines are just spread out all over the place between the various virtual switches. Obviously, I will need to create a Live Migration network, a host management network, a *new* virtual switch for the VMs, and a cluster communication network (i.e. crossover cable).
Based on past experience, when I make these changes, I am going to not only lose network connectivity to/from the virtual machines, but I am expecting that I'll also have to readdress each of them.
Am I correct in this assumption that each VM will have to be readdressed individually? Are there other pitfalls/problems that I need to look out for from the Hyper-V side of things or in general?
I am not a big fan of "changing the tire while I'm driving down the road", but that is what I've been tasked with for this customer.
Thanks in advance for any advice that you may have for me.