That's a very interesting question - and a bit of dark art to answer this!
In general bigger RG means more capacity & more performance, as more spindles are actively serving data (or less parity drives in other words). The downside is, bigger RG will require more time to rebuild if a disk fails (however pre-emptive data copying from a "dodgy" disk via Disk Maintenance can help here).
I normally stick to NetApp Synergy tool suggested RG size, which also tries to make all RGs with the same aggregate as even as possible (which has an impact on performance)
For 60 SAS disks Synergy says RG=20 (perfectly balanced), for 70 SAS disks it suggests RG=18. You can increase this up to RG=28, but preferred range is 12-20.
on netapp support site, there is a storage subsystem configuration faq which goes into details on what aggrs you should create. usualy sweet spot for sas drives is between 16 and 22 disks.
small sequential io expected, rather small rg´s due to smaller stripes to calc parity and write
large sequential io expected, rather big rg´s due to big stripes
THis is what I've created for myself
The maximum RAID group size for SATA is 20 and not 16.
thanks for the info guys.
for my SAS disks:
as of now my current RG size is 21+2 (so that i can get use of all spindles in one shelf. )
i am planning to add 6 additional disks to my pool ( as a growth )
so when i check my aggr i see my RG can accommodate 28 disks in RG0, so if i add 6 additional disks to it, it would look like 26 + 2 but leaving 1 disk aside 9 as a spare )
what would be the complexity if i rearrange my existing RG ( rg0 ) to rg0 -> 20+2 and rg1 -> 20+2
and keep adding all my future disks to RG1 so that i can make use of more spindles and balance the RG's
For SATA of 60 disks:
i will go with 18+2 's