I’m so ready for the first Windows 2012 R2 preview bits. Yes that’s what our current setup looks like. Two RDMA capable NICs at the ready 
… let the bits come. I’m pretty excited to test Live Migration over Multichannel & RDMA.
.
We choose to get RDMA NICs for new servers and replace non RDMA card in host where there’s a clear benefit. By the way have you seen the news on the Mellanox ConnectX®-3 Pro Single/Dual-Port Adapters => NIC support for NVGRE is here people!
Do you remember my blog from 2011 on optimizing some system settings to get way better Live Migration performance with 10Gbps NICs? It’s over here Optimizing Live Migrations with a 10Gbps Network in a Hyper-V Cluster. This advice still holds true, but the power optimization settings & interaction between DELL Generation 12 Server and with Windows Server 2012 has improved significantly. Where with Windows Server 2008 R2 we could hardly get above 16% bandwidth consumption out of the box with Live Migration over a 10Gbps NIC today this just works fine.
Don’t believe me?
You do now? A cool 
For overall peak system performance you might want to adjust your Windows configuration settings to run the High Performance preferred power plan, if that’s needed.
You might do no longer need to dive into the BIOS. Of cause if you have issues because your hardware isn’t that intelligent and/or are still running Windows 2008 R2 you do want to there. As when it comes to speed we want it all and we want it now 
and than you still want to dive into the BIOS and tweak it even on the DELL 12th Generation hardware. Test & confirm I’d say but you should notice a difference, all be it smaller than with Windows Server 2008 R2.
Well let’s revisit this again as we are now no longer working with Generation 10 or 11 servers with an “aged” BIOS. Now we have decommissioned the Generation 10 server, upgrade the BIOS of our Generation 11 Servers and acquired Generation 12 servers. We also no use UEFI for our Hyper-V host installations. The time has come to become familiar with those and the benefits they bring. It also future proofs our host installations.
So where and how do I change the power configuration settings now? Let’s walk through one together. Reboot your server and during the boot cycle hit F2 to enter System Setup.
Select System BIOS
Click on System Profile Settings
The settings you want to adapt are:
That’s it. The below configuration has optimized your power settings on a DELL Generation 12 server like the R720.
When don, click “Back” and than Finish. A warning will pop up and you need to confirm you want to safe your changes. Click “Yes” if you indeed want to do this.
You’ll get a nice confirmation that your settings have been saved. Click “OK” and then click Finish.
Confirm that you want to exit and reboot by clicking yes and voila, when the server comes back on it will be running a full speed at the cost of more power consumption, extra generated heat and cooling.
Remember, if you don’t need to run at full power, don’t. And if you consider using Dynamic Optimization and Power Optimization in System Center Virtual Machine Manager 2012. Save a penguin!
See update in yellow in line (April 11th 2013)
I recently saw KB2834898 (pulled) appear and it’s an important one. This fast publish statement is important as until recently it was accepted that Live Migration with pass through disks was supported with Windows Server 2012 Hyper-V Live Migration (just like with Windows Server 2008 R2 Hyper-V) as long as the live migration is managed by the Hyper-V cluster, i.e. the pass through disk is a clustered resource => see http://social.technet.microsoft.com/wiki/contents/articles/440.hyper-v-how-to-add-a-pass-through-disk-on-a-failover-cluster.aspx
UPDATE April 11th 2013: Now after consulting some very knowledgeable people at Microsoft (like Jeff Woolsey and Ben Armstrong) this KB article is not factual correct and leaves much to be desired. It’s wrong, as pass-through disks are still supported with Live Migration in Windows Server 2012 Hyper-V, when managed by the cluster, just like before in Windows 2008 R2. The KB article has been pulled meanwhile.
Mind you that Shared Nothing Live Migration with pass through disks have never been supported as there is no way to move the pass through disk between hosts. Storage Live Migration is not really relevant in this scenario either, there are no VHDX file to copy apart fro the OS VHDX. Live migrations between stand alone host are equally irrelevant. Hence it’s a Hyper-V Cluster game only for pass through disks.
I have never been a fan of pass through disks and we have never used them in production. Not in the Windows Server 2008 R2 era let alone in the Windows Server 2012 time frame. No really we never used them, not even in our SQL Server virtualization efforts as we just don’t like the loss of flexibility of VHDX files and due to the fact that they tend to complicate things (i.e. things fail like live migration).
I advise people to strongly reconsider if they think they need them and only to use them if they are really sure they actually do have a valid use case. I know some people had various reasons to use them in the past but I have always found them to be a bit of over engineering. One of the better reasons might have been that you needed disks larger then 2TB but than I would advise iSCSI and now with Windows Server 2012 also virtual Fibre Channel (vFC), which is however not needed due to VHDX now supporting up to 64TB in size. Both these options support Live Migration and are useful for in guest clustering, but not as much for size or performance issues in Windows Server 2012 Hyper-V. On the performance side of things we might have eaten a small IO hit before in lieu of the nice benefits of using VHDs. But even a MSFT health check of our Virtualized SQL Server environment didn’t show any performance issues, Sure your needs may be different from ours but the performance argument with Windows Server 2012 and VHDX can be laid to rest. I refer you to my blog Hyper-V Guest Storage Performance: Above & Beyond 1 Million IOPS for more information of VHDX performance improvements and to Windows Server 2012 with Hyper-V & The New VHDX Format Leads The Way for VHDX capabilities in general (size, unmap, …).
Is see only one valid reason why you might have to use them today. You have > 2TB disks in the VM and your backup vendor doesn’t support the VHDX format. Still a reality today unfortunately 
But that can be fixed by changing to another one 
One of the nice features in Windows 8 Hyper-V clustering are the “drain node” capability and the virtual machine priorities. You can see this in action in a video by Aidan Finn here.
For more details on draining a node see Draining Nodes for Planned Maintenance with Windows Server "8" for a detailed explanation.
Now another very important feature is the fact that the cluster is intelligent enough to determine what node is the best suited as a target during failover or live migration. Not only CPU and memory load of the hosts are taking into consideration but also the resource needs of the VM and the priority you have given those. This entire process is NUMA aware and as such with windows 8 can be evaluated on a per virtual machine basis. This means that you that the cluster will always try to get the best possible placement and thus performance for your virtual machines.
Now we also have affinity and anti affinity rules. Anti Affinity ensures that the nodes of a virtualized NLB farm will be placed on separate hosts to minimize risk. You don’t want one host to house all the nodes of you NLB farm!
On the other hand sometime you want virtual machines to stick together lets say you have an NLB farm but the virtual machines with the front end and middle tier need to stay together. In that case you use affinity rules to achieve this. On top of this the anti affinity rules will ensure that the NLB farm virtual machines are on different nodes.
Do note that when the cluster has to choose to break these rules versus bot being able to run the virtual machines it will choose to keep them running. It knows its priorities! Now if in such a situation there are not enough resource the priority will also come into play and the low priority machine may be shut down to ensure the higher priority ones can be up and running.
As you can imagine there are potentially a lot of factors/permutations at play here and I’m looking into doing some more test of these features and the intelligence in the process to see if we make the same decisions and how to best configure this for maximal performance & availability.