ZFS or Raid 5 for backup - Only used/on 1-2 hours a week

[Mysteriouskk]

What is the best way to store a backup that makes it the easiest to recreate a drive if it fails?

I have 5x 3TB drives and need a way to backup static data. Write once and no more changes ever.

Is ZFS the best way? Does ZFS verify the data continuously? How does it know when a hard drive is bad or a sector on the hard drive fails?

 

[Frozen.Shots]

Hello Mysteriouskk,
If your goal is to backup individual Files or Projects, then copying these to your ZFS storage pool is a good start.
Please keep in mind that a backup should consist of at least three copies.
1 original, 1 local and 1 offsite / offline and secured against elemental damage / theft.

ZFS performs regular integrity checks, called "scrubs".
Here's something to read: http://arstechnica.com/staff/2005/12/2049/

If you need help setting up a fileserver to backup to, you've surely found the right forum!

/edit: Drat! The article ars refers to is offline. This should be as informative: https://www.youtube.com/watch?v=CN6iDzesEs0 (Depending on your technical background, you might want to investigate further...)

 

[Mysteriouskk]

If the ZFS server is only on for 2-3 hours a week, will it still be able to verify the entire hard drive pool?

 

[Frozen.Shots]

Depending on the amount of used (allocated) data in one zpool, the scrubbing (verification) process can take up several hours.
For this scenario I'd recommend using a script that powers off the server after finishing a scrub and sending a status notification via email.
I'd estimate that a raidz2 filled to 80%, thus containing ~~6 TB of data, will scrub for at least 6 hours though...
Scrubbing can occur while the system is online, you have full read/write access to the array.

 

[brutalizer]

Either you do a "scrub" which is similar to a "fsck" or "chckdsk" and all data and metadata will be scanned for errors. This takes hours, but you can use the pool meanwhile. Normally, you can not use your raid while scanning for errors ("fsck", "chckdsk").

Or, when you access a file in some way, it will be automatically scanned for errors, and automatically repaired.

 

[FlangeMonkey]

What are you backing up? just files I assume? but your reference to 'recreate a drive' throw's me.

What comes to mind is snapshots, as they can be useful if you need to role-back, which is good for backups if something goes wrong. ZFS has many benefits, but I'd suggest you look into what's referred to as a write hole. ZFS does quite a nice job of consistency, but at the cost of usable space.

Your question "What is the best way to store a backup that makes it the easiest to recreate a drive if it fails?" is loaded, that has more questions than answers...

Some more detail on what your backing up would really help? And what from, is this NAS, a workstation with the 5x Drives running on RAID. Is this Windows, are you backing up by just copying files from time to time? Are you needing to use Time Machine? Is this for archive?

You can do some really intelligent things with ZFS based systems, but backup's are so widespread that without knowing the details people cannot recommend solutions.