I have the chance to set up my system again (better 8TB drives x 4) and add an extra two 8GB for RAM usage over the 2x4GB I have. I’ve noticed that my limitation is more CPU speed than RAM, which 8GB seems enough but I’ll get the 16GB if needed. Having a 453D, I cannot install Quts Hero and I’m told that it won’t be powerful enough anyway.
Ext4 > RAID-5?: Less overhead. Able to recover data from other volumes if 1+ drives fail and don’t need to chance rebuild on a new drive that will cost $$$$ and take a long time.
Thick v. Thin: I don’t need to have drive space flexibility for pools as it will be set to the max with snapshots set in advance that I should not need to adjust. (Willing to hear thoughts on this.)
My Goal: Store my personal digital photos and videos and use QuMagie as a Google Photos alternative. I do not need drive redundancy for uptime. I will keep a crc32 hash file for each photo and consider doing par32 archives/zip or rar archives of folders for backup purposes and data bad file restoration. Having a separate backup with the same crc32/zip/rar mirror allows me to address corrupt files if/when I find them.
One additional concern: I can see having about 100,000 photos and wonder whether performance will be a serious problem regardless of the RAM installed. I’m not sure how much of a difference it will make anyway since QuMagie will be the only app I will likely be running.
Actually - I think you helped answer this question earlier. Snapshots are supposed to address this issue by running them periodically. But originally I set up the NAS with RAID-5 since this was the ‘recommended’ set up when I began. Quite frankly, I’m not sure what benefit it provides other than, theoretically, being able to rebuild the drive (which will likley take a long time, perhaps a day) v. taking a periodic backup. The latter can work if you don’t add files often, as i will not. So now I’m rethinking using Ext 4 with snapshots v. RAID-5.
I just did a deeper dive into understanding what each does and now I get it. RAID-5 checks against the entire disk and doesn’t actually do file integrity checks against a checksum. RAID is the bottom layer - in a way it’s meant to create just one huge disk system that can then be rebuilt. If one disk fails it needs to be rebuilt for the system to work again. You can have layers that reside on top for a file system too, but the ability to read the ext4 is lost with the RAID as the bottom layer as it’s striped across various disks, etc.
Without RAID you can replace a drive by copy data from a backup onto that drive (if you have it identical) and just pop in the drive without a rebuild needed. That should be ideal as my data will be copied across 1-2 backups as I add additional photos to the main nas. All I actually need are separate disks that have file checks for data integrity for each file, not for the entire pool that is the sum of all the disks.
Filesystem scrubbing can check for file integrity. There isn’t any actually done with qts the way I hoped (with checksum) and I’ll stick with that to cross bridges when I get there. Will need to look into what scrubbing does on qts but I recalled it wasn’t available when I set up the first time.
So this will be a bit of a pain to reinstall but I don’t have much on there at all. I can actually install ext4 and not worry about all data being lost if I can’t rebuild. I can use snapshots and also take crc32 checksums as I do for file integrity checks on both the NAS and backup. (I try to keep 3 copies of the data.)
Thin is fine as this data set is not expanding beyond 4x8TB (and certainly not at current prices.) Thanks for the return to discussion.
RAID5 provides hardware redundancy for your disks. The whole point of RAID5 is spreading the risk of a single drive failing across multiple drives. So if a drive fails you can replace it with another one w/o losing your data. Without a RAID5 configuration if a drive fails, you basically lost your data.
There are also speed advantages to RAID arrays as well. And RAID gives you a much larger storage space than a single drive will give you. But RAID0 (basically just adding drives together) is very risky as yeah, you get a lot of storage space, but if one drive goes, you lose everything on all drives. What will take longer? Restoring a backup or rebuilding a raid while you can still access your data.
QTS does have a scrubbing option. So you can do that periodically.
How you want to configure things is up to you and your risk tolerance.
But don’t rely on RAID as a “backup.” It is not that.
Snapshots are great but they aren’t really a backup unless that are stored on some other drive. They will be useless if they are on your drive and the drive fails. Where they are useful is in individual file recovery. Say you accidentally delete or modify a file in a way you don’t want. Well, with snapshots you can go back in time and recover a previous version of that file.
I’ve decided to set up 4 single drives, no RAID. This is merely for photo storage and display top level, with 2 layers of backups. Photos (RAW + JPG) and Video will be uploaded occasionally. I’m not a professional photgrapher so this will be maybe once every 3 months from phone and maybe camera. I have cloud backup for the camera, but photos need to be removed or they will be too unwieldy - so they will go here and to backups.
I created 4 single drives. RAID-5 from my understanding is (1) availability/risk tolerance; and (2) bigger storage pool. #1 doesn’t appear to be a benefit for me using 8TB size disks which I’ve heard can take days to try rebuild which, to me, is a substantial risk of failure. I’d rather have 3 layers of backups with crc32 files to confirm data integrity. I may even do well with 16TB of files on 2HDD and replicating that manually on the other 2HDD.
I’m creating 3 folders on each drive - photo, video, raw. I won’t be using qumagie to see or index the raw since the photos are already in jpg and this is far more burdensome on the system.
For the moment I have years folders (2025, 2026) and each has quarters (Q1, 2, 3, 4) for now but haven’t yet assigned the folders to qumagie. I may divide them into 12 months for general photos and, if I have trips, I may include them individually start with month to keep them sorting, followed by a description. I haven’t yet figured out the optimal way since going through photos and trying to pull out individual trips will be difficult and time consuming. If they are substantial trips, I’ll separate them, with the rest going into one big month folder.
In each month folder I have the source medium (phone or camera model) .
I have not yet decided what to do with media photos I’ve saved from apps such as sms, chat apps, etc. This is difficult but I’ll get there. For now, these are separate from Qumagie until I figure out the best way.
Eventually I’ll use the internal folder assignments in Qumagie, if feasible. Hope this makes sense and we’ll see how we progress.
RAID5 lowers your risk and guarantees availability, so besides the “loss” of one disks worth for distributed parity, there is not really a downside (of course no substitute for backups)
Good points - but the downsides are losing a full 8TB volume in addition to losing data on all 3 disks (21TB) in the event the recovery isn’t successful. Also less wear on the drives. These days the price of HDDs are like gold… For a big photo and video library that isn’t business, I’ll take that over the availability and keep the backups.
No risk..that’s why you have backups, RAID + Backups …easy peasy.
By duplicating data manually between disks you are increasing the wasted storage, and if the NAS dies,..are you prepared with external backups? (you cannot easily remove the NAS drives and read them outside the NAS)
just speaking from tons of experience over the years with other users, many of them ignored the advice and paid with money (recovery services) or tears (data gone)
I completely appreciate this comment. But my understanding is that this is true only if you use RAID. Once striped, those disks can only be read inside the NAS and never outside - but not drives formatted as Single (which use ext4 from my understanding.) See the below article too. I saw instructions on how it may even be possible to put these drives into USB enclosures and potentially recover data that way as well. Using RAID on USB enclosures (like the 4 drive bays for RAID) results in the same problem as the NAS. If one of those drives die and you don’t have an exact replacement enclosure, they are all gone. In many cases I’ve read, even the same make and model will not guarantee that RAID striped drives will be usable. Hence using single + ext4 seems the right way to go. Even if it was striped + ext4, I’m not sure what other benefits I’m receiving with a major loss of storage (8TB.) If I had a 6+ drive bay and/or used smaller drives, I’d consider the RAID-5 or RAID-6 (better) approach. Hope this makes sense.
I don’t consider RAID-5 with 8TB+ hard drives a comfortable solution if one of the drives die because it will require all 3 drives to spin for days to recreate the missing drive. Huge risk. Furthermore, there isn’t any way to know whether data integrity is a problem because my understanding is that nothing is done on the file level to ensure this.
Let’s say I use 2 drives as single and 2 as backup of the four I have. If there is any file degradation, I can quickly identify which file is complete since I have checksums. I can then replace the corrupted file with the original file. This doesn’t mean I will ONLY use backup on the NAS - most certainly not. I will have at least 1 if not 2 copies of everything on the NAS as a backup in the same manner.
I’m not the expert here so I fully appreciate all of you who have greater knowledge than myself inserting your comments and explaining why my logic and understanding may be erroneous or short sighted. Completely appreciated.
From my understanding RAID is ideal for availability with some measure of integrity being maintained. Integrity is best managed on the file system level (like ZFS). I recall considering using RAID-1 mirroring and it was raised that it would absolutely keep the files identical between drives. But RAID doesn’t know which file is actually the correct file to keep because it doesn’t have a checksum to match it with. This is what I understand filesystems like ZFS and what other competitors sell on their more advanced drives (like Btrfs from some other manufacturer I don’t use.) Here that is available on NAS boxes that offer QuTS, which my TS-453D (with 16GB of RAM and expandable unofficially to 32GB) will not. Hence I’m stuck with this approach.
Yeah, using 4 individual drives is just a bad idea. And slow.
I’m not much concerned with speed for this setup. It will be used occasionally to view photos and videos from an archive. I’d feel differently if this was a huge library that I’d regularly access, such as a photographer who may need to have it on daily and accessing files from various folders. I may use this NAS once every 2 weeks, at best. When I do, it will be solely to drop and save new media from my phone/camera, and find, view/retrieve them on the occasions when I’d like to access them. Most of the time I’m looking at Google Photos to look through my photos taken over the past year.
@slinky, I think you are WAY over thinking this. Yes, in a RAID5 configuration, there is a risk of losing it all if a new drive dies while you are rebuilding the raid after a failure. Sure that’s a risk. But that’s why we say to have backups!
You say:
it will require all 3 drives to spin for days
They are spinning anyhow! Hard drives never stop spinning once you power them up. So it’s not like you are doing something to the drives in a rebuild that they were not designed to do.
I would not be so sure that a drive formatted in EXT4 for a QNAP is going to be readable by other systems. I would double check that if I were you.
Also the stuff you have “read” about drives dying and not having exact replacements is BS. Drives don’t need to be the exact same models. And you can move a RAID5 from one QNAP model to another (so long as they are both compatible models). I’ve done it. I moved my RAID5 from my TS451 to my TS873A with zero problem. Just took the drives out of the one and put them in the other.
But you go do you. You seem to know better than all of the experts in the industry that have created and developed concepts such as RAID5 or RAID6, etc. You seem like you are much smarter than them. So go have at it. Do your single discs…It’s not my data that is being stored…
Thanks for the extended response! The challenge with RAID-5 is better explained here.
As to the use of EXT4, RAID-5 is out due to the above. RAID-6 I’d consider using but would loose 2 drives at what cost? For my purposes, 16TB of storage is probably plenty for a long time. I could have 2 drives storing what I access and 2 drives as backup. Failure of any drive makes for an easy HD replacement, copy the files from the backup, hash check to confirm no corruption. Any that don’t check, restore from a backup on an external HDD.
Regarding compatibility, I was talking about between NAS manufacturers and models. I remember the Drobo failures that caused major panic, even trying to move to a replacement from the same manufacturer. While I could get a new QNAP NAS model if I used RAID like you say - that is good! But I’m still seeing being better off without RAID and recovering the extra 8TB of space for this usage.
I respect all the experts and what you say. But is this my use case? I don’t think so. After hearing the greatness of RAID and its primary applications, it seems RAID6 is where I’d want to be with drives 8TB and higher. The truth is that my use is NOT what anyone would probably use a NAS to do - they would use an archival drive at 5400RPM, big, slow, designed for occasional use. But in my case, there is one reason to use the NAS - QuMagie as my Google Photos replacement. Hope this makes sense.
I’m not going to keep going round and round with you on RAID or not RAID. But one writer’s opinion in one article in one tech rag doesn’t make RAID5 invalid. It just doesn’t.
Yes, it takes a long time to rebuild a large array. I’ve done it. And yes there is a slim chance of a drive failing during that process. But that’s what multiple backups are for both on-site and cloud. The odds are that the array is going to rebuild fine. But in case something goes oops you have a backup.
In your case you don’t care. Fine. It sounds like you are doing identical backups to each drive. Basically you are making your own RAID1 in reality. If it works for you then great. Have at it and enjoy.
For me I’ve got > 10 TB worth of data backed up on that NAS and in multiple places. I just had my MacMini go wacko and possibly die (I can’t get the drive reformatted and the system installed no matter what I do), but I have peace of mind that I have over a year’s worth of backups of that machine sitting in multiple places and when I either do get it working or buy a new machine I can just get back up and running from my backups. If my RAID dies tonight it’s OK as I have backups online and in house as well. I’ve lost enough data over the years relying on single drives and a single backup. Never again.
