After about a year of staring at piles of important documents needing to be digitized and wanting to learn more about ways to safely store them for the long run, I have finally started to look into my options. I started with learning about RAID 0, 1, and 10, and I thought I was sufficiently satisfied with the thought of using RAID 1. Then I learned about 5 and 6. Then I heard about Zfs and thought it sounded like the most amazing thing. Then I started reading about all the people hating Zfs. Then I read a key feature is that it can catch and fix corruption issues. Then I read someone say that no RAID or Zfs setup substitutes backups. What the fuck is that supposed to mean? How is RAID 1 not a backup, let alone any of the other options?
Please help me the fuck out. I want to digitize loads of shit. Business receipts, photos, medical documents, manuals, etc. my main concern after reading all the complaints is that all I want is to know that 10 years from now my data will be as good as it was the day I digitized it.
Redundancy complements backups but they're not a substitute. RAIDs are used to keep a system online in case of hardware failure but they cannot stop software or users from doing damage. A RAID setup cannot stop a malicious program from executing for example. Even users or programs that are not intentionally malicious can cause damage to important files or the filesystem.
>>52114880
RAID1 is just a copy, not a backup.
Oh my, i had some elaborate analogy with buttplugs in mind, but it wouldn't be too SFW...
Anyway, using RAID1 is very good but it is not backup, a bit less pricey solution would be RAID5 or 6. In the event of disk failure it let's you replace failed device without interrupting your work. But let's say, all drives fail, or you have some electrical surges, storm or flood. Now all your data is gone. That's where BACKUPs come in handy. They should be stored on a separate device, preferably far away fro your production environment. So whatever happens to your production machine wouldn't affect the backup.
> my main concern after reading all the complaints is that all I want is to know that 10 years from now my data will be as good as it was the day I digitized it.
All you can do is just minimise the risk. Also if you don't pay someone a lot of money you have to supervise everything yourself in order to be pretty sure your data is safe.
I'd go for RAID1 or RAID5 if your volume is small, plus regular backups to a separate USB drive (solid state is better for home use IMO - less prone to physical shock), plus offsite backup using backblaze or similiar service.
The best thing is raid 50 and zfs. but you end up replacing discs alot. zfs does work on crappy drives
>>52114880
Redundancy is against hw and sometimes software failure.
Backup means offsite copy/s of the data, offsite is just a saying for an separate system.
What you want is software or hardware RAID for your live system/data, software is preferred so that you don't rely on hardware OEMs since most RAID systems aren't compatible.
ZFS is a good choice.
Then you want to also backup your data, daily/weekly/monthly/yearly what ever suits your data the best, to another medium, like optical disks, HDs, tapes etc. which are stored "offline" and only accessed if a restore is needed.
Best case scenario you will never have to actually use your backups, but they are there for "just in case".
>>52114880
>using raid 1
>cryptolocker
>everything is fucked
>have backup
see? literally are not the same
If your data is extremely important you should store it on magnetic media and when you're not using it store it offline in a safe place, preferably off site. Magnetic media holds its information for a long time, tapes are the preferred method due to how inexpensive cartridges are relative to their size. Hard disks can work as well but they can break down and render them unreadable.
Is it possible to use RAID 5 and make a backup of the parities across all the disks (not RAID 6)?
>>52115364
What's the point?
>>52115445
To serve as my backup, I guess.
>>52115502
the parity alone wouldn't be enough to reconstruct the files in the event of a complete array failure, you'd be better off just backing up the original data. if you're concerned about failures you can also use a RAID level with higher fault tolerances.
>>52115122
Moar like
>zfs rollback my/shit@tuesday
Zero fucks.
Mirrored RAIDZ3 VDEVS.
>>52115122
bitches don't know about my snapshots
>>52114984
/thread
>>52115186
>Magnetic media holds its information
Can you give details on the longevity, cost, accessibility, and size of magnetic storage mediums?
https://www.youtube.com/watch?v=yAuEgepZG_8
https://www.youtube.com/watch?v=pv9smNQ5fG0
These videos are literally made for you, why haven't you watched them?
>>52117234
Tapes are known to last decades in storage. Hard disks can last quite a while as well probably multiple decades when stored in a safe place with low humidity but the drive motor and arm can break down and render them unreadable.
Flash media is not good for long term storage, it doesn't retain its charge indefinitly it needs to be refreshed periodically or the data will be corrupted after 5-10 years.
Writable CDs/DVDs last like a decade or two at most. The dyes in them break down after a while.
>>52117353
The thing with tapes is the cartridges are cheap but the drive can be pricey for a good one. Tapes are terrible for general storage because of the awful random read speeds but they have very fast sequential read/write speeds which makes them quite useful for backups
