Disk partition flexibility in virtual systems

[Svend]

I have only just started looking at SANs and virtualization and have a couple of questions mostly about the virtualisation of the storage.

I've played with VMware's free GSX server and seen that if I want I can create a disk that can grow, so I make it 400GB even though I only have 60GB hdd on the workstation I'm playing with. Then I can tell the server that it can have a 100GB C: drive and a 300GB D: drive.

I looked for this sort of feature on VMware's ESX server (2.5) but could not see it.

Isn't this the way you future proof your server storage? Just think of how much space you'll need for the next four years and then quadruple that number and make that the virtual disk size. Then when you outgrow your SAN you just swing the virtual servers onto a new or expanded SAN. No more need to mess around repartitioning the serevr disks at the OS level.

Is this not the way it works? Is it not supported in ESX 2.5? Does version 3 support disks that grow, or do I need to hard allocate the disk space up front?

Svend.

 

[stupman]

Yes it does, but it's not such a great idea.
Each time the disk grows, the esx server needs to place a lock on the vmfs volume, making it inaccessible during the time that the disk is allocated.

Having said that, you can create 'thin' disks using vmkfstools (service console).
It goes something like this:

vmkfstools -c 10G -d thin /vmfs/volumes/vmfsvol/vmfolder/disk.vmdk

Hit enter, and your 'thin' disk will be created. You then will have to attach that disk to the VM manually, using the 'use existing disk' option.