Raid 5 bad idea?

i found this article on google


and they suggest me to buy wd.
how about that?

RAID is kind of a pain in the ass. RAID 5 and 6 puts a lot of faith into the controller. If that goes bad, you have a bad day.

I only use RAID 1 (mirroring). And I backup (not quite as frequently as I should).

You start creating disk arrays bigger than any single disk, and backups get complicated too.

I don't recommend anything but RAID 1.

Raid 1 wasn't practical unless using it in tandem with 0 for raid 10 because I needed expandability and raid 1 doesn't provide that.

1 Like

I don't think you can expand a RAID1 to become RAID10. Regardless, I don't recommend it. Expandability will bite you in the ass.

Im a raid 1 fan. Im become a data whore. Made a digitalocean VPS and put my junk on it then raid 1 on my machine that loaded with youtube-dl {wendell}. I might need a support group.....10+TB drives /drool

My 2 cents is large HD's use raid 1.... 3 or 4 drives in a PC thats huge data...2+ TB drives. Its all down to safety cause drivers are cheap. Raid 5 and 3 drives, best hope a resilver that takes many hours does not fail.

The primary purpose of RAID is high availability and increase in R/W, not necessarily reliability. Reliability, or lack there of, it why you have backups.

That said, ZFS and RAID 6/15 are very reliable. RAID 1 is a wash for any business application. I use it myself on a personal NAS but that's because I don't want to have to make backups of my NAS. If a drive fails I'll just pop a new one in and be groovy and I won't be as spooked as with other arrays.

1 Like

You're right but I didn't mean turn RAID 1 into 10. I meant the ability to add a new drive and expand the array. You can do that in RAID 0,5,6,10 but not 1. I mean you can but you won't actually gain any more capacity.

I think you can only do that in RAID 3,4,5, and 6
Nobody uses RAID 3 and 4
And it's controller dependent.

And you can't do it in RaidZ

Yeah I know raid isn't a backup but I have no place to store a back up :/.

Then why don't you at least use one of your drives for local backup instead of raid?

Yes I know and that was why I used RAID 5.

I then wouldn't have enough drives to do anything outside of RAID 0 or 1 and neither of those suit my needs. Another of the primary reasons I use RAID is so I have one giant volume instead of multiples.

You'd be much better off using LVM instead of RAID 0. You can map together drives as one partition, change size on the fly, hot swap devices, take snapshots and create Hybrid volumes (with a SSD as cache for example).

I'm using raid 5 for the parity.

Great in concept. But in practice can be a nightmare.

Remember, the thing about storage systems like RAID5/6, LVM, RAIDZ, or other funky file system is that they are great WHEN THEY WORK

You shouldn't plan your storage infrastructure around the best possible scenario, but around the worst case scenario. And that's why RAID1, or even just multiple backups on multiple drives is superior. The data is immediately available - no sourcing a replacement RAID card, no hoping that the LVM or RAIDZ is re-recognized or recovers, no funky file systems to remount when you might be up against a number of constraints.

I specifically recommended LVM instead of RAID 0. But if you're really going for flexibiloty, nothing beats LVM on top of RAID:

I think most Yes/No answers are wrong anyway. Just realize what you have and what you don't have and decide for yourself.

Raid 5 will allow for one disk to fail and that's it. If you have a single block error on another disk while recovering, it won't complete. It really is that simple.

Where does that leave you?

Well, 99% of consumers probably get a Synology like NAS with 2 bays, and be in the same situation as any mirror configuration has the exact same issue. So let's not get carried away, you are not a ticking time bomb.

On the other hand disks are getting more dense. Some argue this makes them more prone to block errors, but for sure recovery windows are increasing. Your solution does not allow for active scrubbing, some sectors of the disks might not be read for a very long time. That all comes into play.

My personal OPINION? I understand you can recover the data, but it would be a lot of work. Well, I would get a backup in place first. Time is money as well.

Is raid 1 subject to the same problem though? After all both drives are identical mirrors and it is possible to just mount a raid 1 drive normally.

Anyone really caring about data retention would at least go with raid2z.. ZFS is so fucking strong for datarention and integrity. Sure you need a fuckton of ram for it, but you can get hundreds of gigs of ram pretty cheap now a days.

If what you're backing up is worth money to you, set up a raid2z server at least. Then throw an encrypted backup onto something like amazon glacier if you're really concerned about it.

Well, yes it is. If you lose a drive and the other one has a bad block, that data is gone. Nobody seems to care about that. That was my point. Of course in a 3 member Raid5 you have 2 remaining drives so relatively speaking two times the risk, but the absolute risk is not that much of a difference.

Switching to ZFS or BTRFS is not practical for me and if I did I'd probably go BTRFS RAID 10. Either way it's not really practical.