Question DAS: Which RAID Configuration to Minimize Bit Rot Error Risk?

[chane]

I'm dangerously overdue for a revamped back up system; I just use old JBOD among two xeon PCs, though both have ECC RAM. I'm so badly overdue for new storage hardware it's terrifying!!!! But I've got to be sure to buy what I need most to protect my data.

Capacity wise, total data is ~ 350 gig, but I will eventually want to not only do hundreds more uncompressed CD track rips but also dozens rips of TV episodes and rare movies from DVDs (though rarely from not BDs). Those videos I'd be saving as MKV files, not much larger ISO files. The good news is that I mostly watch (re-watch) lots of vintage stuff, and when I do buy more movies I do so on BDs, which are less likely than DVDs to get hit with bit rot. Thus, I'd be less inclined to rip (with RedFox) and back them up. So for for these back ups, should each RAIDed HDD drive be 5 or 7TB?

My plan is to have a multi drive DAS in my living room and bedroom. My internet connection is via my iPhone's wifi so I have no router that could have been otherwise used with two NAS instead. So for every monthly backup session, I'd have to grab the bedroom DAS, bring it to the living room DAS and use my PC's Windows Explorer to share and/or update any document, videos, music files between them via USB.

But got to make sure I don't trip during trips! It would have helped if at least Qnap or Synology made at least one DAS with a safety handle! Why not a case like this? https://www.silverstonetek.com/en/product/info/server-nas/CS01-HS/

AND how best to protect data from however unlikely bit errors?? My data is my life. Qnap and Synology use btrfs, zfs or other non-Windows file management systems offering powerful bit error prevention and self-healing between RAIDed drives. But the learning curves of those file systems-let alone the initial building of them-are way more for the enthusiasts than newbie laypeople like me, and I don't want to get stuck with hardware/software that's a nuisance to learn, use and maintain.

Instead, will the right RAID configuration in both DAS boxes and the ECC RAM in both my PCs get me reasonably good error prevention? If yes, which RAID config to use with two to four 5TB or 7 TB HDDs in each DAS? https://www.seagate.com/products/hard-drives/barracuda-hard-drive/

And which Qnap or Synology model DAS?

 

[Tech Junky]

I run my own diy setup along with other functions in a PC build.

Raid is simple if you do it right and there's no need for ECC unless it makes you feel better. I use my setup as a router as well which was the intent from the start. Initially it also was the AP for wifi as well and had a quad tuner inside for Plex OTA recording. It doesn't take much and is a better deal than dealing with a bunch of different expensive devices that do the same functions.

People tend to get tied up with the whole idea of things sometimes. It doesn't have to be complicated and even if you don't have a traditional ISP connection setup you can still host your files in your home with this approach.

I'm in the middle of a redo / uplift of the system converting it from HDDs to NVME based storage so.... simplifying and collapsing it into a smaller case and reliable drives. I picked up a new MOBO for under $200 and 4 x 4TB NVME drives for ~$800 and then just porting over most of the stuff needed to make it run.

Any system will work though as long as you have enough power and cooling inside the case. Setting up Raid on a DY setup doesn't take too much effort when it comes to the commands and setting up weekly data checks / syncs to automate the health of the data is easy as well.

Current setup I'm using is 5 x 8TB drives in a Raid 10 running in the OS which is Linux Ubuntu. No special cards / HBA / Raid / etc. needed with the drives just connected to the MOBO. To setup the raid you just use MDADM commands to build the array and then add a line to the fstab file to mount it on boot.

Taking one of your PCs and configuring it as a "homebrew router" and adding a quad port 1GE NIC to it for $50 shouldn't be a huge deal. From there you can either wire clients to it directly or hook an AP up to one of the ports and use a streamer stick to get the video to your rooms / TVs or even phones. For media management I use Plex but there are several different options out there. Since I do OTA recordings Plex works well for managing that with the EPG like you might have on your cable TV setup. A couple of clicks and recordings are scheduled / updated automatically.

Anyway... back to the PCs.. I would need to know a bit more about the HW inside and plans to expand to allow you to RIP more to the system. Of course in this instance you'll need a BR device to pull the info from the disks which is fairly easy to do and maybe just go with an external one that connects over USB to the PC for your rip sessions. I don't use a monitor with my setup and manage it from my laptop 99% of the time through apps that have a web GUI and also through SSH command line. Webmin has most of those functions though as well and works with any browser.

If you want to stick with a DAS though I would go with https://www.amazon.com/dp/B07KY73BNQ/ - $190 and gives you full speed of the disks unlike most other options. for HDD setup it allows 400MB/s+ w/ a 10gbps USB connection. If you upgraded to SSD drives or cache you should be able to top out around 1GB/s

For me I would just set it as JBOD and control it from my server setup but the box does have riad functions built into it as well.

There's a few different ways to hit the same goal though as you can see. It depends on personal preferences and aesthetics if you're putting it in a viewable place. When it comes down to it though buying devices is a waste of money since they all use Linux and that costs $0 so, you're paying for something that's free and some over priced HW to hold some drives. Might as well get more than storage out of the setup right?

 

[gea]

Bitrot means datacorruptions by chance at a statistical rate without a hardware defect that cannot be fixed by disk internal error corrections. This means the more data and the longer you wait the more errors are to be expected.

The only way to prevent bitrot are checksums on data and metadata to get informed about errors and redundancy to repair when it happens. This means that you need a filesystem with checksums like btrfs, ReFS or ZFS. As these are not usual desktop filesystems it is normally not possible to use DAS storage with bitrot prevention. Typically you have NAS that you access via SMB. Only option for DAS alike behaviour is iSCSI what means that parts of aZFS NAS storage are offered as iSCSI targets that can be connected by clients like a local disk.

Btrfs are avaiable on Synology, ZFS on higher end Qnaps. Main problem on Synology is that Raid redundancy is not achieved via btrfs raid but traditional raid. That means that btrfs sees just a single disk with checksum errors are detected and bitrot can be repaired only on metadata errors as they are stored twice but not on data errors as this would require btrfs or ZFS software raid.

If you really want to be save, either use a Qnap NAS with ZFS or use any PC with a ZFS distribution as NAS and a dual or triple mirror depending your security need + regular backups to be save on a fire, flash or theft.

ECC is more than a nice to have for a storage server as undetected ram errors can then lead to corrupted datablocks that are written with correct checksums. No chance to ever detect such errors even with ZFS.

 

[Tech Junky]

btrfs, ReFS or ZFS

Good points and all are available in Linux since that's where they're developed.

 

[gea]

Good points and all are available in Linux since that's where they're developed.

Btrfs was mainly developped by Oracle for Linux. ReFS is Microsoft Windows. ZFS was developped by Sun (now Oracle) for its Solaris Unix.

 

[chane]

Anyway... back to the PCs.. I would need to know a bit more about the HW inside and plans to expand to allow you to RIP more to the system......

If you want to stick with a DAS though I would go with https://www.amazon.com/dp/B07KY73BNQ/ - $190 and gives you full speed of the disks unlike most other options. for HDD setup it allows 400MB/s+ w/ a 10gbps USB connection. If you upgraded to SSD drives or cache you should be able to top out around 1GB/s

For me I would just set it as JBOD and control it from my server setup but the box does have riad functions built into it as well.

There's a few different ways to hit the same goal though as you can see. It depends on personal preferences and aesthetics if you're putting it in a viewable place. When it comes down to it though buying devices is a waste of money since they all use Linux and that costs $0 so, you're paying for something that's free and some over priced HW to hold some drives. Might as well get more than storage out of the setup right?

Btrfs was mainly developped by Oracle for Linux. ReFS is Microsoft Windows. ZFS was developped by Sun (now Oracle) for its Solaris Unix.

Thanks for all of this, not that my knowledge base can process more than half of it, nor would I have any hope of configuring the hardware and software to run any of these solutions alone. In any case, if a local (Long Island) provider can't build me a basic redundant back up (and hopefully bit error preventing) system that virtual dummies like me can operate, then my ECC RAM PCs and a DAS or two is probably the best that I can have. https://www.lidatasolutions.com/

For more on my specific data capacity needs and applications please check out my little thread.
https://www.diyaudio.com/community/...imize-bit-rot-error-risk.401936/#post-7418595

phofman there makes a sterling case for TrueNAS, though I've no time and no wish to go anywhere near anything Linux, or risk fiddling with the required VM even if it could run on Windows.

I'm simply too far below the knowledge and experience level of any of you. But what really stinks is that by now you'd think that providers like TrueNAS, Qnap and/or Synology would do a lot more to help those like me who will spend the rest of their lives as perhaps just slightly above average Windows users.

But on that note, gea mentioned Qnap. True OR False: Would this ZFS based NAS with ECC RAM be "easy" for a simple Windows (NON-Coder!!) GUI person to use once a professional configures the ZFS for me??

And how is the fan and ~ 4 HDD acoustical noise from ~ 5 ft away?

Also, if I've ripped a DVD (not 1080p BD) movie and saved it to an uncompressed MKV file, would my PC have any trouble playing it with this QNAP NAS feeding it via USB 2.0?

https://ark.intel.com/content/www/u...w1390-processor-16m-cache-up-to-5-20-ghz.html

https://www.gigabyte.com/Motherboard/W480M-VISION-W-rev-10/support#support-dl-driver

https://www.anandtech.com/show/14071/nvidia-gtx-1660-review-feat-evga-xc-gaming

32GB ECC RAM

Last but hardly least: IF Qnap's flavor of ZFS would be at all useable by my newbie brain, what are the chances of that NAS crashing, perhaps from the sheer complexity of the ZFS itself? In which case, would it make sense to keep a simple two drive mirrored DAS that I normally keep disconnected until it's due to be updated by the NAS?

If yes, what 5, 7 or 10TB HDDs would you recommend for that DAS?

 

[Tech Junky]

There's a lot going on here....
ZFS is going to be more pick your size and stick with it or at least it used to be.

As for bit rot / corruption... anything is possible even with the best laid out solution to prevent it. I've been running my setup for 7-8 years now and haven't seen any issues w/o fancy controllers or complicated FS's other than just EXT4.

Once a NAS/DAS/File Server is setup it's as easy as mapping the location in Explorer. Nothing complicated about it. Sure, off the shelf options give you a pretty GUI and you pay dearly for what it does. Google is your friend or forums as we are here now.

The idea of housing your OD inside the storage platform makes it run faster too as it's using the same bus to store to the disks. You don't need a whole lot of CPU to RIP. Playback is dependent on the target system whether it can do MKV w/o transcoding will make a difference.

As for drives.... If you go with higher capacity now it means less maintenance to swap / upgrade later. Noise? Recent drives don't make much noise like in the old days where they chattered if you even looked at them. Considering I have 5 WD Red drives running in my living room 24/7 should say something about how quiet the setup is. The big decision is what kind of redundancy you want R1 - mirror / R5 - parity / R10 - stripe \ mirror or something else.

It's all up to you though. I like to KISS things and for me R10 does that w/o calculating changes for parity R5. Switching up to NVME based Raid though has me considering it though since it shouldn't have much impact on speed with the calculations.

 

[chane]

There's a lot going on here....

Yes there is, and as if I didn't have enough decisions to make, since everyone at these sub forums seems to rave loudly about ZFS, why did you opt for EXT4, not that I would know of any difference between them?

But from what I'm hearing it's like ZFS won't really let you increase your drive array capacity down the road, but EXT4 will? But with either FS, if you eventually needed greater storage capacity, wouldn't you have to "unbuild" the array first, then do a primary backup to one way bigger drive, then build the new array with drives identical to that new big one?

If yes, then why not go with ZFS, especially IF it may do error prevention better than EXT4?

Or might ZFS be somehow more failure prone than EXT4?

In any case, while I'm forever thankful for places like this, there's absolutely no way that I alone could select, order and slap together hardware and software fast and reliably enough to DIY a NAS, no matter how much expert help I got. I have far too many deadlines piled on me, my crazy day job and a kazillion daily errands and hoops to jump through. It's bad enough that I had to push back other overdue work and spend my entire Saturday posting and brainstorming about this nightmare. la dolce vita non è.

Synology says that EXT4's for Linux. https://kb.synology.com/en-us/DSM/tutorial/Which_file_system_should_I_use_to_create_a_volume

So unless BTRFS is better for Windows users like me then ZFS wins by default?

So if it's to be ZFS then Qnap

https://www.qnap.com/en-us/product/ts-473a/specs/hardware (though I thought that AMD processors didn't officially support ECC RAM) or https://www.qnapworks.com/TS-h886.asp

As DIYing the NAS is not a realsitic option and as no optical drive will fit in a Qnap (or Synology) NAS box, it's my LG BD drive in my pc that will do the DVD video ripping, saved to uncompressed MKV files and temporarily stored on a 500GB internal 2.5" HDD. Then I'll play it back to check for disc based A/V errors befoare sending it to the NAS via USB rather than CAT cable-assuming only USB provides bit error protection via ECC RAM.

Video playback will be with JRiver or VLC player, so no lossy transcoding needed.

So, for my vintage (and thus limited number) DVD ripping needs, this https://www.qnapworks.com/TS-h886.asp with how many HDDs and what RAID type?

Or get two of these https://www.compsource.com/buy/TSh686D16028GUS/Qnap-2640/?src=F or two of these

https://www.qnap.com/en-us/product/ts-473a/specs/hardware (though I thought that AMD processors didn't officially support ECC RAM).
And with what RAID type?

 

[Tech Junky]

With ext4 and using mdadm I can grow on the fly just add additional disks to the party. The system goes into sync mode and redistributes the data. It's just a little bit of prep work on the disk and a couple of commands to add it to the array.

Ext4 is Linux native. You have two systems right now right? Take one and make your own setup before wasting money on something prepackaged. If you don't like it then convert it back and copy the data back. Only thing you lose is a bit of time. Once you have it working then if you want different disks or want to expand the capacity you know how to. You could get a bigger case for it like a Meshify 2 that holds up to 13 disks and add the disks from the second PC to it. The only costs might be a couple of hba cards for more SATA connectors. Even an 8 port SATA hba is cheap.

It's up to you though and if you get a NAS you're stuck to the number of bays you buy or you have to replace it to make more room. When you get beyond 4 drives they get expensive. Plus you rip directly to the drives vs a temp disk and waste time copying to the drives.

 

[bhelhokie]

ZFS won't really let you increase your drive array capacity down the road

My understanding is that you can increase the capacity of your drive pool (zpool) by adding a new virtual device (vdev) of the same raid type.

For example, in my current 4-disk NAS project, I'm planning on creating 1 zpool with 2 mirrors (so 2 vdevs) of 2 10TB disks each. This vpool has a capacity of 20TB acts as a JBOD (just a bunch of disks). If I decide in the future to increase the capacity of this zpool, I can simply buy 2 more drives (of any disk size, I believe) and create a new mirror vdev and add it to the zpool.

I found this tutorial on arstechnica very helpful:

ZFS 101—Understanding ZFS storage and performance

Learn to get the most out of your ZFS filesystem in our new series on storage fundamentals.

arstechnica.com

Gurus of the forum, please correct me if my understanding is incorrect.

 

[chane]

With ext4 and using mdadm I can grow on the fly just add additional disks to the party. The system goes into sync mode and redistributes the data. It's just a little bit of prep work on the disk and a couple of commands to add it to the array........Ext4 is Linux native.

You Linux people fight like tigers to promote that OS. And that's fine as it's likely a way better built OS than Windows may ever be. And it's a shame that Windows rules the world because I like Gates about as much as I do Musk and the rest of what passes for "leadership" in most countries. But that's how it is, and even if Linux ever did morph into the seamlessly functional Windows like GUI that I would require, it probably still couldn't run Samplitude DAW, JRiver, paid version MS Office 2019, Audiolense and/or other software I need.

Beyond any such OS issues, getting anything accomplished, especially in a timely fashion, invariably involves some degree of compromise and money. And it's not just another "What If?" so say that your place may get hit by a flood, fire or whatever and you'll then have to try vacating with your most precious stuff-plus your ~ 55 lb 12 drive NAS tower.

It's bad enough that no pre-packaged NAS have handles, save for those 2U chassis types, with their tiny high rev noisy as hell fans.

So, if I got one or two of the Qnap Xeon 4 bay ZFS NAS, and made two mirrored pairs out of four of these 6TB drives
https://www.westerndigital.com/products/internal-drives/wd-red-pro-sata-hdd#WD6003FFBX that gives me 12TB per box, which is more than twice what I thought I'd have to settle for.

Yes, it all amounts to serious money. But my priceless data will be much safer with drive redundancy and ZFS and ECC RAM data protection, I'll have capacity for lots of choice DVD ripping and the time I save on a prebuilt NAS is well worth the money for me. And the box is small and light enough to stuff at least one of the two NAS into a duffle bag, grab my HP Zbook and escape my doomed apartment when disaster strikes.

So what RAID to use for these four 6TB drives? It surely shouldn't be anything like the "striping" type, as if one drive dies all data is lost, and I would hardly need the higher speeds that RAID type offers.

Would not the best be to simply have two separate mirrored pairs, 6TB capacity each, and for backing up non-identical data? They would be pair A and pair B. Then do the same thing with the other NAS box, where pair A and pair B store the same data as their counterparts do in the other box?

 

[gea]

Last but hardly least: IF Qnap's flavor of ZFS would be at all useable by my newbie brain, what are the chances of that NAS crashing, perhaps from the sheer complexity of the ZFS itself? In which case, would it make sense to keep a simple two drive mirrored DAS that I normally keep disconnected until it's due to be updated by the NAS?

If yes, what 5, 7 or 10TB HDDs would you recommend for that DAS?

There is no Qnap ZFS flavour. It is a regular Open-ZFS on Linux derived from OpenSolaris/Illumos like on any other Linux. Beside Open-ZFS there is native ZFS on Oracle Solaris, the creator of ZFS. Only on Illumos/OmniOS (Solaris fork) Open-ZFS is slightly different (but compatible) due the deeper integration of ZFS into the OS with sharing services like SMB and Windows ntfs alike ACL permissions with Windows SID security references instead Unix uid/gid for SMB.

ZFS is not complicated and very stable as it s used in production since 2005. In the easiest create a datapool from a mirror of two disks, create a filesystem and share it via SMB. If your systemdisk dies, install a new one and import the datapool even with another controller. No special configuration needed as raid config is stored on raid member disks.

If you want to increase ZFS capacity, replace disks by larger ones or add one or more mirrors (or any other ZFS raid like Z1-3). This is then a (multiple) Raid 10 alike stripe setup (or with Z1-3 like a single or multiple Raid 50/60 stripe)

Why the additional DAS? Every mass data move from secure NAS to a more unsecure DAS can produce data errors during the transfer or on a DAS that lacks bitrot protection. Why not keep all data on a checksum protected NAS with snap versioning beside local file processing on your desktop or laptop? Only for backups (best to a second removeable ZFS pool ex on USB disks where transfers are also checksum protected) you want to duplicate data.