I can resonate with your point of view, photos are also my main reason for a backup strategy. Or multiple strategies for that matter.
Are absolutely not a valid backup, not a backup at all. First, content can vanish as soon as licensing changes. Sometimes not coming back. And, even if it does return it might not be the same.
The streaming services will only provide what the record labels license. So, if a new remix of an album is released then THAT version is the only one you have access to. Again, a backup implies you have access to the original, not a replacement that is most likely not as good as the original.
2 Likes
Yes, @Tim_Rhodes is absolutely correct.
Just as RAID is not backup, synchronization is not backup.
Among the problems that backup is supposed to protect against — disk failure, theft, fire, nuclear — near the top is human error. If I delete or otherwise mess up a file, sync/replication will rapidly propagate the mistake to all replicas, even off-site replicas. Backup is different: once a backup is written, nothing will alter or delete that backup. (This is one of my complaints with Synology, when I looked for backup functionality they offered sync.)
I know this well, in my recent job we had a group that built data protection (backup) and a group that built sync, nobody thought they had anything in common.
1 Like
Yeah, sometimes through secondary effects.
When hurricane Sandy flooded parts of Manhattan, some wine storage facilities we inundated and the labels on the wine bottles floated off. Anonymous wine bottles have less value.
In my youth we were out sailing one summer, had stored cans of food in the bottom of the (wooden) boat where there was always water, and the labels were lost. Dinners were very adventurous.
1 Like
You are mostly right, but depends of the point of view. For example I’m spinning the drives very rare these days (I keep the storage on the separate machine which is most of the time switched off) as long as all I want to listen at the moment is available via streaming… At one (limited) extend you can almost consider a turned off computer backup enough 
(joke, I don’t want to start a discussion about what’s the best for a hard drive, continuous spinning or countless power cycles!)
Except for another human error! 
I have seen the damage a lightening strike nearby can do to such HW. On prem alone is NOT a back up.
I run on prem with a Innuos ZENmini AND a PC) which is replicated to cloud storage with history recovery.
I used to have a server here but that’s long gone as not secure.
Allway Sync is excellent and worth the cost, provides sync from the Zen to the PC, automated cloud sync from there.
FWIW
1 Like
Fair comment; I don’t think that this is so much of a problem here in the UK though. 
One option for folk that want to keep a local ‘offsite’ clone of their NAS is to buy a second NAS from the same manufacturer but basic unit, then connect via ethernet to a point in a garage, summerhouse etc.
Use the built in NAS to NAS sync, then in the unfortunate event of a break-in or partial fire, you’ve got a complete clone of the NAS.
We’ve done this for a few folk where they have a physically separate garage or outbuilding that’s unlikely to affected unless it’s a major fire, at which point i’m guessing your music store is the least of your worries. Obviously, you could then leave a USB drive plugged into it for an addition backup.
Avoid hardware RAID (disk controller or NAS system) because it locks you in - instead build your own RAID/NAS - not for the fainthearted but lots of good documentation if you don’t mind building a PC. Most of the dedicated NAS systems are just a custom motherboard, with integrated software controller and a low end CPU, a really nice case and then run Linux or BSD. Having said that, I have definitely (and catastrophically) screwed that up once losing most of the data, but in 15 years, every drive, the controller, the motherboard etc. have all failed or become too decrepit and have been swapped, and with some caution, everything will always be fine.
I use Linux and and specifically mdadm for the software RAID and a relatively low cost JBOD controller. RAID 1 and RAID-6. Things I really really care (photos, documents) about are on the RAID-1 and I use three disks to survive any two failures. Things I care about (music, videos) I put on a 7 drive RAID-6 using 2 TB drives, again to survive two drive failures. I plan for two drive failures, as failures during rebuilds are more common than during normal operation - and you have to rebuild after a single drive failure.
mdadm assumes all drives are the same size as the smallest drive, which makes growing the RAID a case of adding a new drive, or being patient and swapping one out at a time, until all are replaced, and then growing the RAID.
For backup - separate drive in a fire safe at home, and a second drive at another house in a fire safe. Frequent backups to local backup drive, and then every now and again swap that drive with the one at the other house.
RAID-1 is a thing for live data that you need with high availability (think day to day work data), a scenario in which a RAID-10 is even more suitable (and just one disk away from your setup).
For archived data (think occasional access, I suppose you don’t watch your pictures 100 times per day) you are better with the same (or any other) number of disks in a non raid setup. This way you keep the resiliency (the only advantage of the raid 1 setup) but you also add the backup factor.
Totally - RAID is not backup! And, yes no one looks at the photos that often, but we do access documents daily. And I got on this path as the RAID-1 discs were upgraded out of the RAID-6 for size reasons, so why not.
But there are a few reasons why I prefer 3 in a RAID-1. I don’t want to have to restore files if I don’t have to, so despite having a backup strategy (two disks, two locations miles apart), I still want my RAID to be robust. For me, when a drive fails, the replacement needs to be ordered and delivered and I’m “paranoid” that I’ll suffer a second failure - with two spares, I’m protected from that scenario without paying a penalty for rushed delivery. Unlike RAID-6, which too can survive two disc failures, a RAID-1 can survive those failures and any subsequent disc corruption is limited to the specific files it impacts - RAID-6 will be gone if both parity drives fail and there is corruption during a restore - hence my notional concept of RAID-1 with 3 drives being more reliable than RAID-6.
RAID-10 I spent a lot of time thinking about as there’s a lot of commentary about RAID-5/6 becoming less relevant with very large drives - as I read it, statistically if one drive fails, then a second is likely to fail even if they’re from separate manufacturing batches - due to the size of the drives. I haven’t considered RAID-10 to replace my RAID-1 as it cannot survive any two discs failing, so it really is a single failure away from running with no backup, and as I don’t need the performance boost it offers, its not an upgrade yet.
Having said all that, thanks for sharing your thoughts - there is always more to be learnt about data protection.
RAID-6 parity is spread over all disks;
Somehow correct, but they are having very different uses : RAID-1 is for redundancy only, RAID-6 is for redundancy, performance and overall capacity.
RAID-10 has a better failure tolerance than RAID-1: in a 6 disks RAID-10 array you can recover after 3 disks failure. In a 3 disks RAID-1 setup you can only recover after 2 disks failure.
We’re getting off topic, so my finally thoughts on RAID types.
You are correct, RAID-6 of course writes parity across all drives, sloppy statement on my behalf, thank you for adding the clarification.
Agreed, RAID-1 is for redundancy, which is my stated goal - and, in my scenario of 3 discs, RAID-6 has the same capacity as RAID-1, so it has no benefits for my use in this scenario, only increased risk and complexity.
RAID-10 as a stripe across mirrored discs can unconditionally survive 1 disc failure, and only 1. A second disc failure on the same mirrored pair will lead to the entire array being lost along with all data - it doesn’t matter how many discs are in a RAID-10. The striping, which is solely for performance, ensures nothing is recoverable if both drives in a mirrored pair are lost. In my case, the added performance benefit of RAID-10 doesn’t warrant the reduced storage capacity and increased risk of data loss as the RAID-6 I have can already generate double the throughput of the GigE port on my server.
Again, I appreciate your thoughts, thanks.
Not common no, but it does happen. A few years ago, lightening hit very close to the house. Killed my old Arcam AV amp, TV and the Virginmedia router. Everything else plugged into the mains (including Synology NAS) survived. It took Virginmedia about 10 days to replace the router. The tech said he had spent 2 days in the neighbourhood replacing their routers due to the strike damage. He fitted some sort of lightening arrestor inline with the cable feed.
I always prefer offline backups where possible. The family digital photo archive is the only truly irreplaceable thing, and that has several offline disk copies. Cloud is an option for that (<500GB).
1 Like
The problem with any RAID scheme, not just RAID6 and especially if RAID is being thought of as backup, is that RAID actually increases the odds of failure. Only with the type of redundancy that RAID1 implements is this not a problem.
The more discs involved the greater the chance of any disc failing.
RAID’s purpose is to recover from failures, not to prevent them.
FWIW I use backblaze B2 which integrates very nicely with Synology. You can encrypt the cloud storage (keep your keys!! - i use 1Password for that), and you can also encrypt the comms. Schedule the sync for out of hours, or leave on 24/7 (depends on your data volatility). It’s pretty much fire and forget. Do it from a new BB2 account and take advantage of the early free/cheap ingress.
I’m paranoid of media loss as well. It’s why I have a master copy on a mac where I do editing as needed, a replicated copy on ssd in my nucleus for playing, a backup copy on a NAS, and a backup of the backup on a cloud service…
Last year I bought an 8TB disk that also fitted all my music and I wisely decided to postpone buying one for backup, since the disk was so new that in principle nothing could go wrong with it.
So even though I had no backup, I did save money and that was smart.
Then the disk died. The control board / power source failed and I replaced it with a docking station. My Mac OS offered to reformat the disk to make it work again but that seemed a bit drastic since a reformat is known to sometimes delete all that’s on disk.
I looked around and bought DiskWarrior which seemed to be excellent for the job. Alas, the app did not connect to the HD as well.
I contacted the DiskWarrior helpdesk and the man there logged into my laptop remotely, entered a list of long command line stuff, studied the results, started a bit by bit copy to a replacement HD I had ready by then, he looked at the result, added some more command line magic, started DiskWarrior and boom, all content was back!
Since then I did make backup copies. Storing them off site would be the next smart step.
Disk space is cheap…compared to recovery costs it’s priceless
2 Likes