As Chris said you can use RCU 1.0 with VMFS datastore but some steps are manual or you need to script them (not hard). Same as a NFS datastore you create a "gold" VMFS data LUN. In this LUN you place as many VMs as you want per LUN. Be sure that before you do this that you have the LSI SCSI driver loaded in to the image BEFORE you clone (learned that the hard way!). You can then clone out the VMDK file by using Virtual Center (slow) or by using the RFWM tool included in the RCU 1.0 package (fast!). In this LUN you then want to organize the VMDK files to all be in a single folder. Be sure not to have anything BUT vmdk files in that folder. Once this completes, you need to deduplicate the volume and unmap it from the ESX environment.
Next just like with NFS you make a single VM template that has no disk file as part of the template.
You are now set up. To deploy a new batch of VMs here are the steps to follow:
1. Create a FlexClone of the "gold" volume and map it to the ESX environment. Insure that your ESX environment is set to resignature the volume and mount it. You will want to rename it from "snap-xxxyyy" to something useful
2. Use the RCU 1 Command line to specify the new datastore, the folder in which the VMDKs are stored, the template to use and the customization file to use. Optionally you can choose to create fewer VMs than are in the folder but usually you would want them all.
Instructions for the command line are included in the RCU 1.0 documentation.
As you can see the creation of the FlexClone, mounting and renaming could quite easily be scripted or it is a small manual effort.
Admitedly it is much more effort than the right click you have with RCU 2.0 but still a very fast way to create a large number of VMs in a small amount of time.
The vStorage API promised in VI4 aka vSphere could open up similar functionality to VMFS datastores with a future version of RCU but it is too soon to promise that as VI4 is not yet out.
Keith
