8 disk in RAIDZ2 (or 3) with or without vdevs?

[ChapterSevenSeeds]

First off, I want to thank this community for being so responsive and helpful as it has in my journey in becoming acquainted with TrueNAS. And, once again, I come asking for advice.

I currently have a RAIDZ2 volume with 4x 6TB drives. My setup serves principally as our Plex server with movies, pictures, home videos, and music. We do also use it as a backup machine for some of the other computers in the home. However, most of its workload is devoted to streaming Blu-Rays (which only demands at most 4 MBs per second). As such, I am not the most hungry for read/write performance but rather for reducing downtime in the event of a pool failure. All of our data is backed up to the Backblaze cloud so, a pool failure isn't going to ruin me. However, restoring all that data would definitely not be a very fun process.

I am soon going to rebuild the pool with 4 more drives for a total of 8. I am curious as to what the best options are when it comes to reducing the likelihood of a pool failure. From the reading I have been doing, I assume my best options are RAIDZ2 or RAIDZ3 or RAIDZ2 with multiple vdevs. When it comes to vdevs, would that just be two RAIDZ2 pools striped together? If I understand correctly, RAIDZ2 would get me the most storage but with the least amount of redundancy with RAIDZ3 coming next. But, if the probability of 3 or more drives simultaneously dying in a RAIDZ2 pool is mathematically low enough for most people to not have anxiety, I may just go with that setup. I'm just looking for some suggestions and pointers. Thanks!

 

[NugentS]

As you are not fussed about IO I would RAIDZ3 or RAIDZ2 the entire lot. Its not worth splitting 8 drives into multiple vdevs.
Once you get to 10-12 drives, consider more vdevs rather than wider vdevs, but until then a single vdev is fine.

 

[ChapterSevenSeeds]

I know that the difference between RAIDZ3 and RAIDZ2 is that the first has one extra redundant drive. But, in my case with 8x 6TB drives with quick SMART checks 3 times per month, monthly extended SMART checks, and biweekly scrubs, would RAIDZ3 actually offer that much of a reduction in the probability of pool failure over RAIDZ2?

 

[ChrisRJ]

I have been running RAIDZ2 on two systems for the last ten+ years and lost 6 drives (3 WD Red 4 TB CMR in 8 years, 3 Seagate Exos 16 TB in 1 year). But never more than one drive at a time. As long as you have a proper backup it basically comes down to your willingness to pay for an extra drive, in my personal opinion.

 

[ChapterSevenSeeds]

Thanks, ChrisRJ. Your post is what put my paranoia to bed

 

[ChrisRJ]

When it comes to storage, you can't be too paranoid! Seriously. Since I lost a 42 MB (yes, megabytes) drive in 1990, I have spent an absurd amount of money on backup. I see it as an insurance. In hindsight, when everything went according to plan, it may seem like a superfluous expense. But you only know that later.

 

[ChapterSevenSeeds]

Oh I definitely agree. Having my stuff in the Backblaze cloud definitely keeps me from worrying about it too much. It’s just the restoration process that I don’t want to undergo. The thought of it is just terrible. At the same time, I don’t want to have more redundancy if the chances of that extra redundancy failing to protect me are less than once-in-a-lifetime low.

 

[pschatz100]

I would agree with @ChrisRJ that RaidZ2 should be sufficient - along with good backups. Don't forget that RaidZ is not a substitute for a good back up in any case.

I do not wait for disks to actually fail. If an older disk starts showing errors, I replace it. I'm definitely of the "better safe than sorry" mindset.