Flash Storage​

Carlos Mazaro posted 09-27-2022 10:00

Both are good choices for resiliency, for rebuild time 15TB - correction copy time is 6 h 40 m for 6D+2P and in the case of RAID6 (14D+2P), the correction copy time of NVMe SSDs might be two times longer than described - with NO HOST I/O.

John Yarborough posted 10-31-2022 19:51

I also think about a concept of "RISK Pool" when making these decisions.  What if the worst possible scenario occurs and I have to restore all this? If the data on the disks are things like executive staff email I go with the smaller 6+2. Losing their mail can be a resume generating event.

When the data on the disks are things like regular users home directories I use the bigger 14+2. Yes they'll be lots of complaining while the restores are running but I'll survive that.

Sumit Keshri posted 11-04-2022 07:19

In terms of resiliency point of view, RAID 6 (6D+2P) would be the best choice when compared with RAID6 (14D+2P), because 6D+2P can sustain 2 drives failures out of 8 drives, but with 14D +2P configuration, it can sustain only 2 drives failure out of 16 drive.

The rebuild time in RAID 6 (14D+2P) will take lesser time because the data parity will be distributed across all the drives & all the healthy drives will be used in the rebuild operation. So, If you have a higher number of drives in the parity group, the rebuild operation would be quicker.

Shubhadip Pal posted 11-04-2022 13:01

Raid comparison from Hitachi documentations. 

Reliability Index:
RAID6(6D+2P) > RAID6(14D+2P) >> RAID1(4D+4D) = RAID1(2D+2D) = RAID5(3D+1P) > RAID5(7D+1P)

Performance index:
RAID6(14D+2P) > RAID5(7D+1P) > RAID1(4D+4D) > RAID6(6D+2P) > RAID1(2D+2D) = RAID5(3D+1P)

Space efficiency index:
RAID5 (7D+1P) = RAID6 (14D+2P) > RAID5 (3D+1P) = RAID6 (6D+2P) > RAID1 (4D+4D) = RAID1 (2D+2D)