I am currently setting up a ESXi 4.1 integration to a Netapp 2040 via iscsi. The netapp has two shelves/arrays. First is 24 - 300GB 15k SAS disks. Second is 12 - 1TB SATA disks. I am using RAID DP which puts me at roughly (very roughly) 5TB of useable space on each array. I have roughly 50 virtual machines which take up about 4 TB of space. 2 TB of that are VMs with our our main file server on it. The other two TB is a VM with our archive file server on it. Question.
How big should I make the aggregates/volumes on the Netapp size?. I was thinking about just creating One 5TB Volume on the first array. Then creating another 5TB volume on the other array, and then just dump all my VM's in there. Simple but I have read that it may not be the best performance wise. Anyone have any best practices on how I should carve this data up?
Are you running ONTAP 8.0.1? If not, then be wary that for 7.3.x the max deduplicated volume size is 3TB.
Also you should leave some 'breathing space' in the aggregate, so a sum of all volumes capacities should be smaller than the aggregate size - unless you make your volumes thin (space guarantee set to none) when it makes perfect sense to make them as big, as possible.
I think I figured this out a little further as far as LUNs vs. Volumes... so to ask this question a little further. I could create a volume the same size as the aggregate or should I created multiple volumes roughly the same size as the LUNs. See the below example.