Throughout the development of Windows Server 2012, we have regularly pushed the scale boundaries as test resources and engineering schedule allowed  For example, in pushing Hyper-v scale to support up to 64 virtual processors per VM, it isn't just replacing some static variables and retesting.

There is serious engineering prowees that is brought to bear to ensure uniform and consistent scaling across the board such as:

  1. Virtual NUMA
  2. SR-IOV & Networking
  3. Live Migration
  4. Storage Optimisations (1 million IOPs from a single VM :) )

Scalability means systematically breaking through one scalability barrier only to find the next one, addressing that to ultimately achieve system balance.  To that end, Windows Server 2012 is redefining performance and scale and the reason for its success has been the tremendous work cross-team to identifiy the mitigate theses issues across the platform.

Because the Windows Server 2012 scale limits have continuously grown throughout development, I wanted to let you know that Windows Server 2012 RTM will now support up to

8,000 Virtual Machines in a Cluster!

Yes that's right, 8,000 VM's in a cluster. To put that in perspective in terms of numbers for VMs per server in a cluster that translates to:

  • 8 nodes in a cluster - 1,000 VMs per server
  • 16 nodes in a cluster - 500 VMs per server
  • 32 nodes in a cluster - 250 VMs per server
  • 64 nodes in a cluster - 125 VMs per server

As you can see, you could do an 8 node cluster with 1,000 VMs per server, but failing over 1,000 VMs per server will logically take some time and be dependant on a multitude of factors most of them storage related.  Thus, the ability to support up to 64 nodes (twice as many as VMware) to disperse the load is significant.