Which raid with 8 disks?

[aspomwell]

I like pointing people to this site for backup basics. It's written as marketing material for an old solution (Veracity) but the principles are sound: http://www.taobackup.com/

I've recently bought 2 8TB Seagate USB drives in the US. In the week between buying the first and the second it went from $169 to $189 but those two drives will backup all my stuff with plenty of room to grow.

[P3R]

For my own personal backup I then want to use one of the disks in the 8-slot NAS running no raid which is for my own backup files...

Why, why, why?

If instead you do an 8 disk RAID 6 you will get the equivalent storage but it will also be protected by the RAID 6 "for free". In addition to that, all of your storage will have better performance and have less overhead.

If you think you need the separate disks to separate your data, that's not the smartest way to do that on a NAS. Instead you do that with separated shared folders and user permissions.

- the 2-slot NAS will be having an identical drive which I frequently copy the files over to.
The last disk is also for backup files, and will also be using one drive on each NAS and will be my partners backup.

Personally I'm no fan of single disk configurations in NASes either but to get 2*10 TB of storage, single disks is the best configuration in a 2-bay NAS. As long as you're aware of and accept the issues when the first disk fails, I think it's okay.

The 2-slot NAS will be located at my parents house and copy will go over internet.

Great!

[camaxide]

* https://www.memset.com/tools/raid-calculator/ shows that a 8 disk raid 6 has more than twice the chance of failing than a 6 disk one - as there are more disks that can go down. and I felt that keeping the backup file disk seperate makes not only the process of "rebuild" much easier - as it is simply to get a new disk and copy the still working disk over again - is better than if the raid 6 goes down - where you then need all the other disks to survive the rebuild cycle to avoid disaster.
Also, with files simply stored at a disk, a company can easily recover the files even if a mechanical error should make the disk unreadable and in the odd event both backups should go dead - while I'm not sure that is as easy with a setup where 8 drives has data striped and shared amongst them through a raid controller.

* About the 2 drives in a NAS running single - if drive 1 with the os goes down that will not affect disk2 right? I mean - sure I cant access disk 2 until the os is reinstalled on disk 1 and access is back up - but the files wont be changed or so? as soon as disk 1 is reinstalled from scratch I can read disk2 as before?

[camaxide]

I like pointing people to this site for backup basics. It's written as marketing material for an old solution (Veracity) but the principles are sound: http://www.taobackup.com/

I've recently bought 2 8TB Seagate USB drives in the US. In the week between buying the first and the second it went from $169 to $189 but those two drives will backup all my stuff with plenty of room to grow.

Thanks, I was a bit unsure in the start as it was an odd presentation, but the message was still quite clear. I guess the last slide was quite ad-oriented like you said - but else it summed up most of the good points I've gotten through here

[P3R]

* https://www.memset.com/tools/raid-calculator/ shows that a 8 disk raid 6 has more than twice the chance of failing than a 6 disk one - as there are more disks that can go down.

You're still at a fantastic reliability with 8 disks. For example this RAID reliability calculator tells us an 8-disk RAID 6 have a Mean Time to Data Loss of 28079.42 YEARS using very conservative estimates on the disks used. If you're worried about that, then you should really also start looking at things to protect from of all the other single point of failure components in the NAS failing.

and I felt that keeping the backup file disk seperate makes not only the process of "rebuild" much easier...

You'll be accessing the disk through a shared folder so in reality not any easier than having that shared folder on a larger RAID array.

...is better than if the raid 6 goes down - where you then need all the other disks to survive the rebuild cycle to avoid disaster.

No the beauty of RAID 6, and what makes it so reliable is that a second disk is also allowed to fail and you'll still be fine. Used together with normal best practise of having notifications properly set up, doing rapid SMART testing daily, extended SMART testing weekly, RAID scrubbing monthly and having a cold spare disk close by I consider an 8-disk very reliable. That said, if we rely on RAID to keep our data safe we're in very dangerous territory.

RAID is for keeping the system available and to make life easier for the administrator.
Backups on separate systems is the only thing that really protect our data!

Also, with files simply stored at a disk, a company can easily recover the files even if a mechanical error should make the disk unreadable and in the odd event both backups should go dead - while I'm not sure that is as easy with a setup where 8 drives has data striped and shared amongst them through a raid controller.

In my opinion emergency recovery shouldn't be a part of a proper data storage strategy but technically you're correct, a single disk would be easier. Still very expensive though.

* About the 2 drives in a NAS running single - if drive 1 with the os goes down that will not affect disk2 right? I mean - sure I cant access disk 2 until the os is reinstalled on disk 1 and access is back up - but the files wont be changed or so? as soon as disk 1 is reinstalled from scratch I can read disk2 as before?

If you do things exactly correct yes (and it's not clearly documented, as it's not really a recommended strategy). You need to be very careful, as it's not as straight forward as you probably think.

[camaxide]

Thanks a lot for all the inputs - it helps a lot to get closer to a decision - and with black friday closing in I might be getting a setup soon

* I'm thinking about in the event the raid controller messes something up - or if I simply want to pull a disk out of a dead NAS and access it in a computer etc. - for those cases a disk running "vanilla" will simply just work, while a raided disk needs the entire system to work in order to be usable at all. And even if last way out emergency recovery is expensive, it's well worth not losing your lifetime data storage of projects and memories. I guess the calculators also only calculate chance of drive failures, and not the chance of raid controllers messing the thing up etc.? thats just my assumption though - but I read that larger raids with more disks tends to get messed up by the controller easier than simpler raids - and a non-raid simply cant be messed up by any controller as a disk can usualy be rebuilt even if the table goes corrupted (as long as new data is not saved over the old files).

* I do see the beauty of Raid 6 though, which is why I consider running that for all the drives that are not the priority backup data. As even the rest of the files landing on the NAS are files I'd rather see not disapearing - but still the two backup disks are the most important ones.

* Hm, how come it's not straight forward with one disk? does the NAS do anything special or out of the box to a drive running outside raid? like storing the file table differently or something? can't you just stuff a single non-raid drive into the nas and start reading files on it? (as long as disk1 has an os installed that can read that disk)
in a normal pc there is no issues with this at least. and since I've got plenty of experience with pc's I assume it to be the same, but would like to know from someone here who knows it in NAS

ps: is there a simple autoquite here? or do I have to type each quote manually?

[dolbyman]

no raid controller in qnaps..all software raid

reading internal qnap drives requires extensive knowlage of linux and lvm structure ...so consider this a non option

[camaxide]

no raid controller in qnaps..all software raid

reading internal qnap drives requires extensive knowlage of linux and lvm structure ...so consider this a non option

I didn't quite catch which of the parts is not an option? Recovering the backup from a raid6 if it fails - or also to recover a nonraid drive? or did you mean something different?

I also found: https://helpdesk.qnap.com/index.php?/Kn ... t-can-i-do

[dolbyman]

simply put .. you will not be able to read a qnap drive with another computer

[camaxide]

Ah ok, but with another Qnap NAS it will be fine? what about other NASs?

[dolbyman]

other qnap would work (check manual under system migration)

other nas vendors...unknown

[camaxide]

hm, and what about the link I posted?
https://helpdesk.qnap.com/index.php?/Kn ... t-can-i-do
does not work?

[dolbyman]

that was for old cat1 devices..does not work anymore

[P3R]

...but I read that larger raids with more disks tends to get messed up by the controller easier than simpler raids...

Unless you can give us a reliable source presenting a technical explanation for that statement, I would consider that FUD talk by people that for one reason or another don't like RAID. Almost all of the people that don't like RAID are non-professionals that have been burnt by doing unsafe RAID configurations, using cheap incompatible desktop disks, not been following the best practices that I mentioned earlier and, most importantly, not having proper backups for the cases when the used RAID redundancy wasn't enough.

I've been working with (both hardware and software) RAID since the early 1990's and have never heard of RAID controllers messing disks up being a general problem. Yes there could be an occasional bug but that's extremely rare. As far as I can remember Qnap have had a single RAID bug in the last 10-11 years since I started using them.

If RAID was generally unsafe, hi-end professional storage simply wouldn't use it but they do.

To be clear, Qnap don't use hardware RAID controllers but use software RAID in the OS used. That doesn't really matter though as there's not really much difference between hardware and software RAID today.

- and a non-raid simply cant be messed up by any controller as a disk can usualy be rebuilt even if the table goes corrupted (as long as new data is not saved over the old files).

Bugs are bugs and they could corrupt data regardless of underlying hardware and disk configuration used. The thing with RAID 5 and 6 is that several disks are necessary to do any data recovery so slightly more complicated but that's not really a problem for a professional data recovery service.

* I do see the beauty of Raid 6 though, which is why I consider running that for all the drives that are not the priority backup data. As even the rest of the files landing on the NAS are files I'd rather see not disapearing - but still the two backup disks are the most important ones.

It's funny, having a long RAID experience, I would definitely put the most important files on RAID.

With your strategy, you make sure that every time your main disk fail you'd have your most important data only on a backup disk. Though unlikely to happen at that time, backups can also fail. With my strategy of keeping the important data on a RAID, I'd have pretty good odds of never having to rely on the backup as the only copy.

...in a normal pc there is no issues with this at least. and since I've got plenty of experience with pc's I assume it to be the same, but would like to know from someone here who knows it in NAS

Many do that mistake but don't assume the NAS is the same as a PC! Not even a Linux PC. Qnaps are embedded systems which are in several important ways are different from other Linux computers.

[P3R]

other qnap would work (check manual under system migration)

The documented procedure though is moving all disks over from one Qnap NAS to another. It's not moving a Qnap disk holding data into an already configured Qnap!

The latter should be possible if doing it correctly I think but I've never tried it myself as I like to play it more safe and rely on RAID and backups instead.