FAS8060 with 6 * 24 800GB ADP sizing question


I have a cluster with 2 FAS8060, they are NOT AFF model, with 6 2246 fully populated with 800GB SSD.  I want to ADP them and need your opinion.  Would it be better to go with RAID group of 28 + 28 + 16 or 25 + 25 + 22?  Is there any limitation I'm hitting?  I think I saw an article somewhere that AFF model supports only upto 36 drives and I was wondering if FAS8060 is the same case.





from best performance and data safety perspective, use the default raid group size value.

for best performance, use max raid group size.


You should follow these guidelines when sizing your RAID groups composed of SSDs:

  • All RAID groups in an aggregate should have a similar number of drives. 

    The RAID groups do not have to be exactly the same size, but you should avoid having any RAID group that is less than one half the size of other RAID groups in the same aggregate when possible.

  • For RAID-DP, the recommended range of RAID group size is between 20 and 28.


You will be hitting a few limitations with your plans - sorry


  • ADP is only supported on "real" AFF systems, even though the FAS8060 with only SSD is essentially the same as an AFF8060, there are system build flags that are different and will prevent the system from initiating ADP configuration on reset. This article outlines how to disruptively convert from FAS to AFF -
  • Even on AFF, ADP is limited to 2 shelves, with additional shelves being used as standard unpartitioned drives. It is ok to mix these two sizes of SSDs inside an aggregate (ie, 800GB for a standard drive and ~760GB on an ADP drive, or 380GB on an ADPv2 drive)
  • Our design rule is to put no more than 4 x 24 drive SSD shelves on a single SAS stack. For this 8060, that would mean running 2 x 3 shelf stacks from the 4 onboard SAS ports, or adding an additional SAS card to split workload. The FAS8060 supports up to 240 SSDs - this would require an additional SAS card per controller (X2065 or X2069), and running 2 x 4 shelf stacks + 1 x 2 shelf stack
  • Our design rules for RAID groups under ONTAP 9 state that if any of the RAID groups are different sizes, the smallest group should be at least half the size of the largest group, so 28+28+16 is acceptable, however it doesn't stasify our 1+1% per controller per disk type spare policy, so it would become 28+28+14 - marginal, but still ok, but the neat-freak in me would prefer 24+23+23

Enjoy your new system!