Questions about setting up a RAID Drive

[jblock22]

TLDR; What is the best way to set up a Raid array for 8 drives to give the most usable space, and also if a drive dies not lose any data?

---------------------

I'm new to the RAID game. I have a Plex server setup on my PC (pretty powerful) with four 8tb WD external drives connected (filled with media) that have no backup currently what-so-ever, so if a drive dies I'm SOL. I'm beginning to run out of room on them. I now have four 14tb WD externals sitting new in boxes that I'm trying to figure out the best way to set up with my existing configuration. I've ordered an 8-bay enclosure and plan on shucking the drives out of the external enclosures and housing them all in one enclosure.

My question now is, I have a LOT of data already on these four 8tb drives (nearly 30TB), linked up and through my plex server that's all nicely organized, etc. What I'm trying to figure out is a way to link all of these now eight total drives up into one backup system (in case a drive dies I don't lose a huge chunk of my media collection).

Does anyone have any suggestions on how best to do this? I've been looking into RAID and it seems a bit confusing. RAID 0 seems to be the best for 1-1 backup, but it eats up literally 50% of your available space, also it seems that all drives need to be formatted to work in a RAID setup... I don't have an ADDITIONAL four 8tb drives that I don't plan on using that I can transfer all of that data off of to format the four 8tb and four 14tb into a RAID 0 array. Maybe I'm looking at this wrong, because it doesn't seem to make sense. Is the only way to use a RAID 0 array is to start from scratch and wipe all of your data? What's the best way for me to get all of this current data (nearly 30TB) from these 8tb into a new backup system that will also secure and backup my data?

Also, is there a different RAID or some other system to use that will let me have more of my potentially available space? Just for back-of-the-hand math, I have four 8tb and four 14tb drives which would give me a hypothetical 88TB of space. RAID 0 cuts that in half by making a perfect duplicate of the data which would give me 44TB of space. Is there something that I can do to still give me a backup of my data but allow me to use more of my potential space? If a drive fails I don't need to access this stuff on-the-fly. It's movies, music, photos, backups of media I've created, etc. I'm thinking maybe there's something that will compress the backed up files and put it somewhere where it will take up less space since it doesn't need to be accessed, and then if a drive fails, it can uncompress the data and restore it. I also want this to be something that is pretty hands off. Set to backup once a day, week, etc and it just runs on a schedule without my having to physically initiate it.

I know there are certain types of software that will take on RAID duties (I don't personally need it to be a physical process), but not sure which is best, nor which would be best for my situation.

Sorry if the explanation of what I'm trying to do is a bit clunky. Please let me know if you need more details or something isn't quite clear. I appreciate your guys help! I

 

[r9]

RAID 5 if your system supports it.
In four drive RAID 5 configuration you have 75% space available and for redundancy you can have one drive fail without loosing data.
Build RAID 5 with the the 4 new 14tb drives.
Copy the data over from the 8TB drives.
Make another RAID 5 with the 8TB drives.

 

[claes]

Hate to be this person, but RAID5 is pointless on drives of that size and is not a backup solution.

If you’re married to an external enclosure then you want at least RAID6, ideally 60, with disks that size.

If you can, I’d build a new system and move to ZFS or similar. Hardware RAID has reached its limitations with larger disks, and external enclosures and their controllers are often buggy and quick to reach EoL/go out of business.

Still, you’ll need a proper, offsite backup solution. I use backblaze for their price, but there are plenty of options out there.

 

[newtekie1]

No RAID is a backup solution, neither is ZFS. And ZFS isn't really any better of a solution than RAID5 or 6 with large drives.

Talking about RAID5/Single Drive redundancy, the result is still the same, when a drive fails, you have to wait for the replacement drive to rebuild/resilver, both take about the same amount of time. If another drives dies during the process, you data is gone. The same applies for RAID6, but you have to loose 3 drives, instead of 2, before your data is dead.

The way I would do it is put 5x14TB drives in RAID6(or 2 drive redundancy if you aren't using RAID). That will give you 42TB of usable space. Copy everything from the 8TB drives onto the 42TB array. Then put 6x8TB drives in a RAID5, giving you 40TB of usable space. Then I'd use the array of 8TB drives as a backup of the 42TB array. You can either use a Robocopy script to do the backups, or something like Fbackup which allows you to compress the backup. You just might run into an issue with backup space once you hit 40TB of data.

 

[jblock22]

First off, thank you for the replies!

Next, it seems there is no consensus of how to best do this? @newtekie1 that doesn't sound like a bad plan but I'm already over my spending budget with the wife and I don't think picking up another 14tb as well as 2 more 8tb's are in the equation at the moment (plus they won't fit in the new 8-bay enclosure I have coming). Is there anything I can do with what I currently have?

I'd rather not use a cloud-based service for backup at the moment (hence purchasing all of the physical drives). Perhaps in time when unlimited data and internet speeds come to follow, then it may be a viable solution for me. It also seems like there's contrasting opinion on @r9 and @claes solutions will work?

@r9 's solution seems to be the easiest to do / most viable, except that others are saying it won't work? Any reason why?

Again, this is a lot of media stuffs. Whereas I don't want to lose any of it, it's not my crypto keys or anything like that. I think if I can have it so as long as 1 drive dies I can retrieve all of the data I should be able to sleep better at night. I can roll the dice on 2 drives not dying at the exact same time.

Thoughts?

 

[thesmokingman]

RAID is just asking for trouble. I used to use RAID back in the day, RAID 5 then 6, still shitty. After years of the issues from overpriced RAID cards, to bit errors, to sudden failures, rebuilding, god forbid you get an error in the middle of a rebuild, and all that crap it just isn't worth.

 

[Steevo]

The questions to be asked are.

Are you OK with data loss? If you have a limited amount of data you are OK losing that will fit on an external drive do that.

If yes, put them in RAID 5, one drive fails and you have time to rebuild or perform a backup of essential data.

If no, RAID 6 which will most likely require a dedicated card. But will tolerate 2 failures.

If you have been using JBOD or other multi-drive into one configurations and are OK with that, I would suggest sticking with it, moving to RAID essentially multiplies the chance of failure by the number of drives in the array, where if you have 6 drives runnign the chances of complete data loss become 6X higher for RAID 0, 3X greater for RAID 5 and at least 2X higher for RAID 6 due to the probability of failure.

I have had 2 3TB drives running in RAID 0 for years with no issues, I only backup my pictures, music and important documents. My downloads of things can be gotten again, my games in Steam and other places can be downloaded again and for the storage space a large external drive connected once a month is fine with me.

 

[jblock22]

RAID is just asking for trouble. I used to use RAID back in the day, RAID 5 then 6, still shitty. After years of the issues from overpriced RAID cards, to bit errors, to sudden failures, rebuilding, god forbid you get an error in the middle of a rebuild, and all that crap it just isn't worth.

What would your suggestion be, then? Just not back up the data?

The questions to be asked are.

Are you OK with data loss? If you have a limited amount of data you are OK losing that will fit on an external drive do that.

If yes, put them in RAID 5, one drive fails and you have time to rebuild or perform a backup of essential data.

If no, RAID 6 which will most likely require a dedicated card. But will tolerate 2 failures.

If you have been using JBOD or other multi-drive into one configurations and are OK with that, I would suggest sticking with it, moving to RAID essentially multiplies the chance of failure by the number of drives in the array, where if you have 6 drives runnign the chances of complete data loss become 6X higher for RAID 0, 3X greater for RAID 5 and at least 2X higher for RAID 6 due to the probability of failure.

I have had 2 3TB drives running in RAID 0 for years with no issues, I only backup my pictures, music and important documents. My downloads of things can be gotten again, my games in Steam and other places can be downloaded again and for the storage space a large external drive connected once a month is fine with me.

JBOD is a new term for me. I have only been looking at RAID. Is that something worth checking out? Will it work with what I currently have or will I need to purchase additional hardware and/or drives?

 

[thesmokingman]

What would your suggestion be, then? Just not back up the data?

What do you mean not back up? You back up anything you can't afford to lose whether you have RAID or not.

JBOD is a new term for me. I have only been looking at RAID. Is that something worth checking out? Will it work with what I currently have or will I need to purchase additional hardware and/or drives?

JBOD is even worse no redundancy just LOSE LOSE.

Just get a bunch of externals use something simple or advanced to schedule backups or do it manually to said externals. I use syncback free for ex. to back up like 60TB to externals, 1:1 ftw.

 

[jblock22]

What do you mean not back up? You back up anything you can't afford to lose whether you have RAID or not.


JBOD is even worse no redundancy just LOSE LOSE.

Just get a bunch of externals use something simple or advanced to schedule backups or do it manually to said externals. I use syncback free for ex. to back up like 60TB to externals, 1:1 ftw.

Thanks again for the quick reply.

I was looking up JBOD (just a bunch of disks). It seems that there a software called stablebit drivepool that you might be able to pair with a JBOD setup that could create some redundancy.

My issue with scheduling "straight backups" is that it seems you would lose a lot of drive space that way? Basically the same idea as RAID 0 at the point, right? a 1:1 backup ratio? E.G. if i had an 8tb drive full of data, I would need another fully empty 8tb drive to "back up" that data. Whereas if I did something like RAID 5, I could have four 8tb drives, and use 24tb of that space... am I correct in this thinking?

 

[thesmokingman]

Thanks again for the quick reply.

I was looking up JBOD (just a bunch of disks). It seems that there a software called stablebit drivepool that you might be able to pair with a JBOD setup that could create some redundancy.

My issue with scheduling "straight backups" is that it seems you would lose a lot of drive space that way? Basically the same idea as RAID 0 at the point, right? a 1:1 backup ratio? E.G. if i had an 8tb drive full of data, I would need another fully empty 8tb drive to "back up" that data. Whereas if I did something like RAID 5, I could have four 8tb drives, and use 24tb of that space... am I correct in this thinking?

Nah, with RAID 5 in a 4 drive setup you lose 25% of the space. You really need to research this more. There's a ton of downsides and high cost. It only makes sense in enterprise environments.

If you really must have large size... look at unraid.

 

[jblock22]

Nah, with RAID 5 in a 4 drive setup you lose 25% of the space. You really need to research this more. There's a ton of downsides and high cost. It only makes sense in enterprise environments.

If you really must have large size... look at unraid.

Hmmm, I'm guessing either i'm confused or I'm not explaining what my wishes are clearly enough.

It seems like your suggestion is to just do a "simple backup" or any files / folders I can't live without, and then if a drive dies, go back and try to redownload everything that's missing. (I could be incorrect, but that seems to be the gist of what I'm getting).

What I want to achieve is a backup of ALL files, while retaining as much usable drive space as possible. To me, it sounds like an absolute nightmare if a drive dies to try and re-rip 8TB worth of movies to recover them. I'd rather just pop a new drive in and have it recreate them, bada bing bada boom. However, I have a limited budget and limited space. I can't just buy 20 drives and do 1:1 backup. That's why I was thinking something like RAID 5 might be nice because I'm not losing 50% of my usable space, I'm only losing 25% but if a drive dies I can still retrieve the data with relatively little fuss.

However, I see some people are definitely against RAID configurations. So I'm wondering if there's something (like stablebit drivepool) that will allow me to use a JBOD backup, but also create redundancy without using a 1:1 ratio (basically using software to have a JBOD setup but the do distribute data so I can still use... 25% or however much of my total hard drive size).

I am definitely researching, but having actual conversations about my thought process is helping. Without this thread I'm not sure I would have even researched JBOD (or heard of it to look for it) and now I do, for instance.

 

[newtekie1]

@r9 's solution seems to be the easiest to do / most viable, except that others are saying it won't work? Any reason why?

It will work, it just doesn't provide that great of data redundancy. You can still survive a single drive failure, but two drives will cause you to loose all your data on the array.

There is always a trade off with data protection. RAID isn't a backup. Anything that doesn't totally reproduce the data in a 1:1 fashion in a different storage space is NOT a backup.

It sounds like what you want isn't a data backup, but just some data protection, which isn't an uncommon goal in the consumer space when dealing with huge data volumes. It gets financially difficult to have 1:1 backups when talking about 20TB+ storage spaces. This is where you have to weigh how important the data really is to you.

It sounds like in your situation a RAID5 with the 14TB drives is what you want. That will give you 42TB of usable space. You can't add the 8TB drives to this, RAID is best used with drives of the same size. With the 14TB drives in RAID5, if one drives fails, your data will survive. It sounds like you are OK with that level of redundancy. But just know that during a rebuild is the most likely time for a drive to fail, so it often is not as easy as just popping in a new drive and bada bing bada boom. Often times, when you pop that new drive in, another drive will fail and poof, all your data is gone.

r9's solution sounds like it will achieve the goal you want. It isn't ideal by any means, but if you financially are at your limit, it's probably the best option. Create the RAID5 array with the 14TB drives, copy the data from the 8TB drives. Then create another RAID5 array with the 8TB drives and use the smaller array to backup up your important files 1:1. You won't be able to backup the entire 42TB, but you can back up your important files. The files that you can reproduce just by, say, re-ripping the movies off the discs you can skip backing up, but those family pictures you back up, etc.

 

[thesmokingman]

Haha... try rebuilding 30tb array or larger, will take days to complete and all you need is a bit error and you are screwed. And on consumer systems without ecc you are asking for trouble.

 

[jblock22]

@newtekie1 wow, thank you for that detailed write-up, gives me a lot to consider.

Okay so that does sound like it could be a potential nightmare with RAID 5. Just to verify, if in a 4 drive array, ONE of the drives fails, and then (with 3 drives in the array) i install a new 4th drive, and another drive fails while trying to write that new data, I will lose ALL of my data in that 4 drive array? Or would I just permanently lose the data from that failed drive and now be unable to recover it? If that's true that I'd lose 4 drives worth of data, and the stress of writing data to the new drive is highly likely to cause another drive to fail, that does sound like it could (really) suck. It would take you from losing data from a single drive to losing ALL of your data on 4 drives permanently? Ugh. Question at that point is it better to just not backup your data at all? Heh

What about RAID 6, then? Wouldn't that require 2 drives to fail to lose all your data? Although does that eat up 2 drives worth of space to do instead of one? Or are there other reasons raid 6 isn't a great idea?

 

[newtekie1]

Okay so that does sound like it could be a potential nightmare with RAID 5. Just to verify, if in a 4 drive array, ONE of the drives fails, and then (with 3 drives in the array) i install a new 4th drive, and another drive fails while trying to write that new data, I will lose ALL of my data in that 4 drive array? Or would I just permanently lose the data from that failed drive and now be unable to recover it? If that's true that I'd lose 4 drives worth of data, and the stress of writing data to the new drive is highly likely to cause another drive to fail, that does sound like it could (really) suck. It would take you from losing data from a single drive to losing ALL of your data on 4 drives permanently? Ugh. Question at that point is it better to just not backup your data at all? Heh

If one drive fails, and you replace it, and another drive fails while the data is being rebuilt onto the new drive, Yes you lose the data on all of the drives. It seems like you are still thinking of RAID as a backup, it isn't. A backup is a 1:1 copy of your data. If you don't have that, you don't have a backup. RAID is a layer of protection from data loss, but it is not a backup.

What about RAID 6, then? Wouldn't that require 2 drives to fail to lose all your data? Although does that eat up 2 drives worth of space to do instead of one? Or are there other reasons raid 6 isn't a great idea?

RAID6 does eat 2 drives but it allows for 2 drives to fail and no data to be lost. There is no other reason to use RAID6 other than when a single drive fails, you are still protected and can have another drive fail during the rebuild and not lose data. RAID6 is recommended over RAID5 when using large drives because the larger the drives the longer it takes to rebuild a degraded array. The longer it takes to rebuild an array the more chance there is of a second drive failing before the rebuild is complete.

 

[thesmokingman]

In either case once you lose a single drive in R5 or two in R6, you are pretty much up shit creek and are teetering on the brink. In both situations anytime a drive drops you will be severely impacted on speed as it will suck resources to provide any data accessed by computing it from parity on fly. OP doesn't seem to have the deference for the seriousness that this implies. Not to mention the cost of a true hardware card, lol trying R5 in software is insane.

 

[dogwitch]

Haha... try rebuilding 30tb array or larger, will take days to complete and all you need is a bit error and you are screwed. And on consumer systems without ecc you are asking for trouble.

and yet with enterprise lvl. ecc can still fail. nothing is perfect

 

[thesmokingman]

and yet with enterprise lvl. ecc can still fail. nothing is perfect

Oh yea that is possible but it's a whole different ball of wax. Losing a storage server isn't the end of the world when you have triple backups, offsite, etc and corporate money and resources to throw at it. Doing this for home use is just stupid imo. The shit will fail, it always does and doing it just for the luxury one large ass archive folder/drive is ludicrous.

 

[Aquinus]

Others have said this, but RAID is not intended to be a replacement for backups. It's to mitigate downtime due to a hardware failure.

Not to mention the cost of a true hardware card, lol trying R5 in software is insane.

Modern software RAID is pretty quick when configured properly, at least in Linux. This isn't 20 years ago when CPU resources were constrained. CPUs these days are really good at doing a lot of XORs quickly. I would have zero issues with using mdadm for RAID-5 or 6 with a modern CPU. In fact I have and performance actually wasn't that bad, and that was on a Phenom II chip, so hardly modern either.

 

[Hachi_Roku256563]

Just use a backup disk
thats how you backup

 

[A Computer Guy]

TLDR; What is the best way to set up a Raid array for 8 drives to give the most usable space, and also if a drive dies not lose any data?

---------------------

I'm new to the RAID game. I have a Plex server setup on my PC (pretty powerful) with four 8tb WD external drives connected (filled with media) that have no backup currently what-so-ever, so if a drive dies I'm SOL. I'm beginning to run out of room on them. I now have four 14tb WD externals sitting new in boxes that I'm trying to figure out the best way to set up with my existing configuration. I've ordered an 8-bay enclosure and plan on shucking the drives out of the external enclosures and housing them all in one enclosure.

My question now is, I have a LOT of data already on these four 8tb drives (nearly 30TB), linked up and through my plex server that's all nicely organized, etc. What I'm trying to figure out is a way to link all of these now eight total drives up into one backup system (in case a drive dies I don't lose a huge chunk of my media collection).

Does anyone have any suggestions on how best to do this? I've been looking into RAID and it seems a bit confusing. RAID 0 seems to be the best for 1-1 backup, but it eats up literally 50% of your available space, also it seems that all drives need to be formatted to work in a RAID setup... I don't have an ADDITIONAL four 8tb drives that I don't plan on using that I can transfer all of that data off of to format the four 8tb and four 14tb into a RAID 0 array. Maybe I'm looking at this wrong, because it doesn't seem to make sense. Is the only way to use a RAID 0 array is to start from scratch and wipe all of your data? What's the best way for me to get all of this current data (nearly 30TB) from these 8tb into a new backup system that will also secure and backup my data?

Also, is there a different RAID or some other system to use that will let me have more of my potentially available space? Just for back-of-the-hand math, I have four 8tb and four 14tb drives which would give me a hypothetical 88TB of space. RAID 0 cuts that in half by making a perfect duplicate of the data which would give me 44TB of space. Is there something that I can do to still give me a backup of my data but allow me to use more of my potential space? If a drive fails I don't need to access this stuff on-the-fly. It's movies, music, photos, backups of media I've created, etc. I'm thinking maybe there's something that will compress the backed up files and put it somewhere where it will take up less space since it doesn't need to be accessed, and then if a drive fails, it can uncompress the data and restore it. I also want this to be something that is pretty hands off. Set to backup once a day, week, etc and it just runs on a schedule without my having to physically initiate it.

I know there are certain types of software that will take on RAID duties (I don't personally need it to be a physical process), but not sure which is best, nor which would be best for my situation.

Sorry if the explanation of what I'm trying to do is a bit clunky. Please let me know if you need more details or something isn't quite clear. I appreciate your guys help! I

There's a lot of good info here I won't repeat it but here is what I do in case it is similar to your needs.

Mine are pretty simple so this kind of setup may not be for you depending on your needs.

I use two 4 bay Synology NAS since it's pretty easy, plug and play, and power efficient. (although 5 bay might be nicer for a hot spare)

Primary in RAID6 (4 drives, 2 drive redundancy with spare on shelf) so I can withstand drive failure and replacement without having to restore from backup unless everything fails.

Secondary RAID6 (identical config) as a spare device in case the primary unit fails. If the primary fails I can just swap the drives out of the primary into the secondary and be back up and running again in no time.

2 standalone backup disks. One active daily incremental backup, One manual monthly synchronized backup. Occasionally I test that the backups can actually restore on the Secondary unit.

 

[Aquinus]

To be honest, if I were going to put any 8 disk configuration into RAID, I'd probably do RAID-6 (regardless of software, fakeraid, or hardware implementations.) That's two disks of tolerance and means 48TB of disk space available with 8TB disks. Reads should be amazing, writes should be okay with modern hardware

...but:

Spoiler

YOU STILL NEED TO BACK IT UP!!!!​

...and have fun backing up 48TB+.

 

[A Computer Guy]

...I've ordered an 8-bay enclosure and plan on shucking the drives out of the external enclosures and housing them all in one enclosure.

Depending on the shucked drives, you might NOT want to put those in RAID if you don't know the answer to the following questions.
- Do you know if they are SMR or CMR drives?
- Do you know if they are RAID compatible drives?

 

[Chrispy_]

Haha... try rebuilding 30tb array or larger, will take days to complete and all you need is a bit error and you are screwed. And on consumer systems without ecc you are asking for trouble.

Rebuild times with modern >10TiB drives means that I wouldn't even expect a RAID6 to recover unless the server had plenty of ECC RAM and the drives were different makes/models; Drives die in batches. If anyone buys 8 identical drives with close manufacturing dates, when one drive dies they are all on their last legs. Mix and match drives for safety if you must run a single array.

To be honest, if I were going to put any 8 disk configuration into RAID, I'd probably do RAID-6 (regardless of software, fakeraid, or hardware implementations.) That's two disks of tolerance and means 48TB of disk space available with 8TB disks. Reads should be amazing, writes should be okay with modern hardware

At 14TB I'd be building a 7-disk RAID6 with one dedicated hot-spare. You want the rebuild from parity to happen immediately and automatically when a drive fails and you don't want to power-cycle the unit or mess with anything until it's done.

At that point though, you're only using 5/8 of your drives for data, and 3 are dedicated to integrity. That's not far off just have two separate storage systems and syncing from a primary/production to a secondary/cold system.