I have been performing some tests with a new FAS2240 unit and have noticed a discrepency between the size usage on the NetApp LUN vs what VMware is actually using. I believe I understand well enough what I am seeing, but would like clarification.
Netapp setup: 1 volume in NetApp of 500GB with a 20% snapshot reserve. So 400GB for data and 100GB for snapshots. 1 LUN in this 500GB volume. I have made the LUN 350GB in size. Thick provisioned on both the volume and LUN. No snapshots are being taken, fractional reserve is turned off.
VMware setup: iSCSI SAN - Added a datastore using all of the NetApp LUN space. So the datastore is 350GB in size. There are currently 330GB of VM's located on this datastore. All of these VM's are thick provisioned lazy zeroed.
What I end up with is the VMware datastore reporting 330GB in use, but the NetApp LUN reports ~100GB in use.
1. I assume that NetApp is reporting only on blocks that have actually physically changed/been written too. For example even though I create a VM that is 100GB in size, this doesnt mean that 100GB of data is being written to the backend storage at the block level, especially when using lazy zeroed for the VM. Is this correct?
2. As more blocks are written on the LUN, the usage reported by NetApp goes up. Usage however does not go down. For example, I delete 300GB of VM's from the 350GB LUN. Vmware reports the free space, but Netapp stays the same. Are there any issues with the LUN finally hitting 100% usage even though on the VMware side there is plenty of free space? Issues with snapshots or snymirror?
The simple answers to your questions are:
2. No issue with your lun hitting 100% usage (it most likely will eventually). You said you aren't doing snapshots. If you start doing snapshots you will have to pay attention to your backend volume space utilization as your lun could go offline if the volume gets full. Snapshots will use snapshot reserve space, but they will spill over to data space if they run out of snap reserve space. You might also want to look into allowing your volume to autogrow.
It's not the volume that will go offline. Your volume will never go offline unless you explicitly offline it. It's your lun(s) that could go offline. Yes, you can use snapshot autodelete (most people use a combination of both snapshot autodelete and volume autogrow). In our company, we use snapshots as backups and we'd rather not delete them, so we don't use that option.