RAID5 with 8TB drives is asking for trouble. Personally, with drives over ~4TB, full mirrors is the way to go. Whether using actual RAID5 or the ZFS equivalent, you have an extremely high probability of losing your volume when using calculated parity on large drives. The I/O-intensive rebuild is likely to cause a second failure.
I am also building a server pc with 8 x 4tb seagate ironwolf nas in it … would you say raid 6 is a good option ? I do want the option to add another 3 hdds in the future as my case (fractal design r6) can hold 11 in total
RAID6 (or RAIDZ2) May be a good option with 8x4TB drives. You’ll get 24TB of usable space (vs 16TB if you do mirrors). Personally, I sacrifice the space and go with the mirrors over calculated parity, but, with 4TB drives, I don’t think RAID6 would be irresponsible. Are you limited to what size drives you can use? 10TB white label WD’s can be had for ~$150. I’d rather have four of those in mirrors, giving me 20TB of storage with room to double it going forward.
I already have 4 x 4tb seagate ironwolf and I am on a slight budget so buying another 4 to give me 26tb on raid 6 just seems the best option for me , also I can easily add another 3 at another date giving me another 12tb which should last me years ?
The best way to add capacity to calculated parity schemes (RAID5/6,RAIDZ/Z2, etc) would be to replace single drives (one at a time) with larger drives (that said, I wouldn’t use >4TB drives with RAID5/6). Adding to the number of drives in a RAID6, while possible under many configurations, requires intensive disk I/O (encouraging an un-recoverable failure due to multiple drives dying simultaneously under heavy load) and should generally be avoided when possible. If it were my data, I would prefer a RAID configuration that did not require computation based parity. That said, I imagine we’re talking about media (vs “important” data) so a failure during the expansion probably wouldn’t be earth shattering. Disks are so cheap these days that I made a decision a few years ago to discontinue my use of RAID5/6/RAIDZ entirely and transition to RAID1/10 (or ZFS equivalents) exclusively. Gives me more peace of mind.
On another note, 8x4TB under RAID6 would be ~24TB not ~26TB (2 parity drives, leaves 6x4=24). RAID5 would be 28TB. I REALLY would advise against single parity with that many drives.
While I am not suggesting you are incorrect, different priorities can lead to alternate solutions.
I am happy with BTRFS on 8TB x 6 w raid5. This volume has growth through 4TB/6TB to 8TB disks and from 4 disks to 6 (soon to be 7)
I would rather use my extra disks to facilitate extra backups. In fact , that’s where my 4/6TB disks have been re-homed to. For media volumes sequential performance is what I am looking for (10gig E) and RAID 5 (with enough spindles) is good at that. BTRFS is helpful for the integrity piece (sure maybe not ZFS level but …) What remains is availability. Sure 2 disk redundancy would be better, but for my money, I saw better value in using those $ for better backup options. the downtime if I had 2 simultaneous failures is survivable.
If your data is backed up elsewhere, I say have at it. If your data is only protected by RAID5, I personally believe things will eventually go south. Two parity drives would only be marginally better.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.
