We are using VMware Esxi t.1 for our virtual servers. I am planning to install a new Exchange 2010 SP1 mailbox server for a set of users. I used the latest Exchange mailbox server calculator to determine the system requirements for the new server. There are 2 methods for creating the Exchange VM:
- Create it as just a straight up VM. It will be a 200gb C: (system, app), 200 gb E: (DB), 40gb F: (log). Each disk will be a separate VMDK file that will be configured as a direct attached SCSI disks as far as the OS and app are concerned.
- Create the VM with a 200gb C:, but then present via the iSCSI initiator on the VM the 200gb E: and 40gb F:.
The reason option #1 is most appealing is that it does not require the overhead of MPIO and the iSCSI initiator on the VM. It doesn't require any additional provisioning steps from our SAN type devices. Everything about that method is simple to build, support, expand, etc.
The reason option #2 may be required. Our new blade vSphere ESX host datastores are on NetApp presented via NFS. Even though the VMs see their storage as local SCSI disks, I don't know that it qualifies as non-NAS by Microsoft. If they are STILL taking a hard line against NAS then we might be stuck with the far more complicated iSCSI route.
Are there any red flags for using method 1 in this case?
Thanks!