I have 46disks left on my NetApp and I need to create either multiple small aggregates or 1 large one to cover 13TB of data. I know the recommended and default size is 16disks for a RAID-DP. When the NetApp tech setup the inital aggr ... it was with a pool of 40 disks and RAID-DP size of 20. I just wonder if there's any issue with performance or reliability if put it up to 20disks.?
And a question on spares. The balanced recommendation is 4 per controller. But that doesnt take into play how many aggregates? or should that matter? becuase if you use DP you can still lose 2 disks and be ok.
16 disk per Raid Group is NetApp's best practice and default. That being said, you would generally see some performance boost due to more disks available for striping with more disks in the Raid. The downside is that in the event the Raid has to rebuild, the more disks in the raid group the longer the rebuild takes.
As far as spares, I may be mistaken but I think best practice is 2 spares per disk type per controller. So if you have a clustered system with and each filer has a DS4243 shelf with 2TB drives, you would want 2 spares for Controller A and 2 for Controller B adding up to your 4. We have had our filers in place for about 6 years, with controller upgrades, and added many different shelves and disk types through the years. So in my case, for each disk type I need to spares. So I have some 144 FC GB disks, some 450 GB SAS disks, 2 TB SATA disks, and some 500 GB SATA disks. For each of those I need to have 2 spare.
One thing I would add is for performance try and make sure that all the raid groups within an aggregate are of equal or near equal size; as having 40 drives added to a default aggregate would give rg0 with 16 drives, rg1 with 16drives and rg2 of 8 drives. this rg2 would be a bit of a bottle neck due to it having less drives. bear in mind that each new rg will have 2 parity drives
On the spares topic. 2 is the min recommended per drive type (size and speed) per controller 4 is balanced and 6 is max recommended. There are then rules ie if its an entry level you can get away with 1. if spares lead time is >48 hrs then double, the number of disks in the system etc, etc. The no of aggregates the system has is not important in deciding the no of spares required
Its is good practice to have at least 2 drives as this will allow ONTAP to proceed to Maintence Center which req a min of 2 drives. And if for some freak reason you loose 2 drives before you get any replacements you can unassign a drive from one controller and reassign it to the second.