[b:3f3knktv][color=#c00000:3f3knktv]Q:[/color:3f3knktv][/b:3f3knktv]Why can’t multiple nodes access the same logical unit number (LUN) running NTFS in a Windows cluster, as they can with VMware ESX Server and VMFS?

[color=#00b050:3f3knktv][b:3f3knktv]A.[/b:3f3knktv] [/color:3f3knktv]Users of VMware ESX Server are used to formatting LUNs, which are areas of space on shared storage such as a storage area network (SAN), with the VMFS file system. VMFS lets multiple nodes access the same LUN simultaneously. Thus, multiple virtual machines (VMs) can be running on one LUN, and each VM could be running on different nodes because the file system is accessible to all nodes at the same time, as the following diagram shows.

[attachment=1:3f3knktv]ntfsvmfs1.jpg[/attachment:3f3knktv]

VMFS performs file locking on the disk by locking virtual hard disks that are in use, to ensure that multiple nodes can’t start the same VM.

With Windows, the clustering process prevents an NTFS file system from being accessed by multiple nodes at the same time, by restricting the disk resource to being online on a single node only. This is known as the shared-nothing model because any resource can be online on only one node at any given time. This single-access model avoids possible corruption to the file system.

Windows doesn’t have a separate cluster file system, which means it isn’t possible to have a LUN active on multiple nodes in a Windows cluster. A separate LUN that must be separately movable between nodes is needed for each unit of a file system, as the following diagram shows.

[attachment=0:3f3knktv]ntfsvmfs2.jpg[/attachment:3f3knktv]

With a separate LUN for each VM, the flexibility to move each VM individually is achieved by simply moving the LUN that houses the VM to another node.