As you saw from my earlier post, I've been loaned some new hardware and I have a plan for it as follows:
I can now demonstrate a cluster larger than three servers. In fact, I’ve already created a five node cluster and it’s pretty cool! The only downside to my five node cluster is that three of the nodes are AMD CPUs and the other two are Intel CPUs. Yes, your cluster can have a mix of CPUs, but you can only Live Migrate from an Intel CPU to another Intel CPU, or an AMD CPU to another AMD CPU. Remember, I talked about that here. The only thing the mixed CPU cluster gives me, from a Hyper-V perspective, is failover. If my workloads are running on an Intel CPU and the server fails, an AMD based server can restart the workloads since the workloads have to be rebooted. I get HA, but not Live Migration. In my environment, I actually shut down a number of my computers over Thanksgiving. One of my DCs is an Intel machine, the other is an AMD. I left the two of them running in a cluster. While I couldn’t Live Migrate from one to the other, I didn’t need to. My goal was just to make sure that if I lost a server, my workloads would be restarted. I don’t see this as a very viable production scenario, but scenarios like this have made my lab a lot more flexible.
I’ve also been able to build two separate clusters and plan to show how SCVMM can manage more than one cluster at a time. I’ve taken my five node cluster and broken it into two clusters. One three node AMD cluster, and one two node Intel cluster. This is pretty cool and it gives me a bit more flexibility from a test perspective. I’ve already moved a workload from my AMD cluster to my Intel cluster via SCVMM. I had to take an outage to move the workload since I was moving CPU architecture, but it gave me some additional insight. I’ll show you SCVMM and how we can manage both clusters while of course still utilizing the one library they both share.
With the loan of the Intel CPU and motherboard, I’m now able to cluster together two Intel machines running Hyper-V. My original Intel machine is a Dell Precision 490 with a 3 year old Xeon 5050 CPU in it. The newest Intel CPU is the XEON 3440. My goal is to show you how you can Live Migrate between two different generations of CPUs. Why? Well let’s look at our customers that have already virtualized, or even those that are about to. As quickly as the hardware is progressing, if our customers build a three node cluster now; what do they do when that cluster needs to be expanded in six to twelve months? We need to give them the confidence that they can scale their cluster out as their demand increases. We can define a clear starting point and growth path as their business dictates. This is another area where Virtualization will allow our customers to scale at their pace. I will show you that as their clusters need to grow they can continue to economically scale their cluster and continue to leverage new innovations without having to start over on new platforms.
For now, I’ll go to work on this list of ideas and get them delivered to you as soon as I can. As always, I’ll update my Hyper-V summary page here with these new updates as I get them posted.
Until next time!