RAID is not a backup- digital music archiving discussion

I always hear the old saying "RAID is not a backup" for reasons like it does not protect against data corruption (corrupt file will be copied over numerous times) or hardware failure which could take out the entire array.

But I rarely hear of solutions, so I am curious what do people use as a legitimate well thought out backup system for your digital music collection? I have read one interesting system of using hard drives partitioned to ZFS which has inherent error correction/checking and using those drives in 1:1 mirror arrays. However it requires a special protocol to write to and it's over my non-techie head

My current system (after losing many CDs, it can happen) is to backup to hard drives and Backblaze. And while CD rot may have been limited to a certain era/CD manufacturers, I have noticed at least two dozen or more classical CDs that I was able to rip without issue many years ago now cause EAC and dbpoweramp to go wild. These are pristine CDs (mostly from the 90s/early 2000) that were either only played in home CD players or ripped once.

I'm still not entirely sold on using Backblaze as my backup, so I'm looking for other ideas. Besides my photos and documents my music collection is my most valuable digital data.

 

ZFS-raid(raid-5) + external hdd + blu-ray media.

One solution is never enough!

 

What hardware are you using with your ZFS Raid5?

 

I have a completely separate NAS at a friend's house with an internet connection that allows for back up over the internet, plus, the content of the entire collection is held in another location (though not backed up as often, so it does not contain the latest additions). This is the way I hope to protect 4,400 CDs ripped in WAV files, with meta data that has been edited for almost all of the ripped CDs. I don't want to even think about having to rip the collection all over again.

 

RAID is not backup, as the OP said. RAID is availability. My solution:

An 8TB NAS (RAID) backs up to an identical NAS that is on prem, different circuit. And to AWS. And to a set of external hard drives that I rotate to an off site location.

Because both the OP and Larry above are right - it's not the hardware expense, it's that you never want to have to rip all that stuff 2x. Because you can never recover that time. I have 7k discs ripped and tagged. There is no way in the world I want to repeat that exercise.

Back it up people. Back it up.

 

I keep all of my music and my customer's files backed up on multiple 1.5 TB hard drives on different computers, plus 3 hard drives and around 100 DVD's kept off-site in case my house burns down. You can't be too backed up. Hard drive real estate has never been cheaper. It'll take a good bit of time to put all your files on multiple hard drives. But in the end, the peace of mind you'll have is worth it.

 

3 identical external USB 3.0 drives.....free file sync software.....monthly syncs

done

 

rsync or robocopy to an external drive. And then cycle that drive with another drive off-site, for extra peace of mind.

 

I use a 8TB NAS as my music and video server (about 50% filled up).
Then I back up my Music to an old external 2TB backup, and my Video to another old 2TB external backup.

I bet NAS's and external HD's will go on sale for black Friday and Cyber Monday.
I got my 8TB NAS on BF for $399 a couple years ago, and will keep my eyes open for another if cheaper.

I ever trust an external that is older than 5 years old.

 

RAID is protection against drive failure - it has nothing to do with "backups".

https://en.wikipedia.org/wiki/RAID

A RAID drive can be wiped, by accident, in a matter of seconds - never mind data corruption.

A "backup" is a duplicate. Totally different concept. That's why people say "RAID is not a backup".

 

It's important to keep an asynchronous backup (e.g. external hard drive) in addition to any synchronized backups so that any data corruption, accidentally deleted or modified files, etc don't propagate from your original to your backup.

 

Let's also not confuse the different levels (or types) of RAID. (https://en.wikipedia.org/wiki/RAID#Standard_levels)

RAID is NOT synonymous with "protection against drive failure."

Most people just mean 'RAID 5' when they said RAID. It provides redundancy (which is good) but not a backup.

But for example, RAID 0 provides no redundancy, but more drives for faster data access (useful for very data-hungry applications like video editing). So, for example, RAID 0 is not just not a backup, it's actually WORSE than having no redundancy, because your data is spread across multiple drives, and any single drive failure wipes you out completely. This is raw performance only. Not usually what people mean when they talk RAID, though, again unless you're in film production or something like it and you're actually hampered by drive-read speed.

Unless you have a minimum of 3 complete copies, in a minimum of 2 different geographic locations, you don't have a backup.

 

Absolutely correct. My previous comment was a misleading generalization.

 

I've used Backblaze for three years now. I couldn't be happier. It only provides the disaster recovery part of my backup plan. If the worst happens, I can get the last 30 days of my data back, including all my music files. I still rely on Time Machine for local archival backups, which allow me to recover files from weekly backups as far back as I want to go.

No matter what you implement on-site, you aren't protected until your data resides in another physical location. It's not either or, it's BOTH.

 

This ^^^^^^

 

Just backup to two external drives and keep one in a closet and one in your desk at work. Every few months copy anything new to the hard drives and check the integrity of the backups.

 

At the rate at which hard drives seem to die or start to corrupt files, if you're making multiple backups to other hard drives isn't it more wise to use something other than NTFS?

 

All of my music is duplicated on 2 desktops, 1 laptop, optical disc, and 4 external hard drives.
So, 8 copies total.

 

One of the reasons why RAID1 (aka a hot disk 'mirror') should only be used in very limited cases, or in conjunction with add'l asynchronous backups.

One of the biggest 'failures' in the industry I saw was the human 'oops' - files got deleted, entire directories or even volumes. Then begins the restore process.

And I'll chime in - be sure to check that your restore process 'works'. It's one thing to confidently say - oh I have backups, and another to verify that you can actually restore.

 

As crazy as this sounds I have zero backups. For ANY files, let alone the thousands of CDs I've ripped. None. Zilch. Somewhat broke at the moment and praying I'm safe for the foreseeable future, but as soon as I can I plan on investing into something resembling the following.

One network attached drivebay with at least 2 4tB drives in a RAID1 configuration.

One off-site drive kept up to date 2-4 times a year.


Is this the most foolproof solution? No but realistically speaking this should be fine. In the future when SSDs get big enough for mass storage and cheaper I'd put all my data on one of those too.

I would recommend using NTFS for any backup drives. It's a common filesystem, and data recovery is pretty trivial these days - in the case of electronic or mechanical failure of the drive, a professional will be able to recover your data without too much headache and likely without any corruption.


I would also recommend having 4.7 gb of your most critical and irreplaceable data burned to an M-Disc DVD and kept somewhere off-site. Or more. The discs aren't really that expensive nor are the burners, and we can assume that there will be a relatively easy way to access data stored on DVDs for decades to come.

 

Absolutely. I worked for one of the very first disk-based network backup products. The ugly little secret about tape backups is that more than 50% of them won't restore.

 

Without a doubt. I have burned so many DVD discs using "quality/best" media like Ritek and Taiyo Yuden. And a bunch of them now return CRC errors when I try and copy their contents using any number of drives.