single disk pool?

[evilted]

I am sure this has been asked many times, however, I dont see clear advice for this. When i read about a single disk, the first thing everyone always says is use raid. (period). look, thats great if you have spare cash, or you live in the western world and can walk down to the local pc shop and buy disks, cheaply, off the shelf. If its a business, then yes, raid is not even a question. thats not me, i have an old pc - 3rd gen intel, and i would like to turn the machine into centralised storage. the machine is very old and has had a great life, it runs like a champion and actually its only this last year i stopped using it for day to day activities,

I have 5 disks, 180gig ssd - this is where the os is installed. 500gig ssd - wanted to put a windows or linux vm on it. 2tb rotational no idea what to do with it and 2 x 4tb rotational drives - these i will mirror and put photos and personal documents on it. So yes, i get that important data should be mirrored or backed up at the very minimum.

my question is simply to know which 'layout' method to use with a single disk? the only options that exist are for raid, unless im not looking in the correct place? i will never buy another hard drive, and I am very much ok with the potential of data loss in the event a non-redundant disk dies, this is not my question.

please before someone jumps in and lays it out pretty thickly that i must use a mirror or raid (as thats what all the other posts have said/done): i am only asking for advice on which laytout to use with a single disk.

 

[Ericloewe]

There is no layout, a single disk is a single disk. I'm not sure what else you'd want.

 

[chuck32]

So yes, i get that important data should be mirrored or backed up at the very minimum.

I corrected that for you ;)

RAID (or mirrors for that matter) are not a backup solution, they are an availability solution.

 

[Etorix]

That's called "single drive vdev", and is equivalent to a "1-wide" mirror. Just declare the disk as the sole member of the vdev.
It works—until it doesn't because there is no redundancy, but you know that already. (You may use "copies=2" as a way to have at least resilency against read errors, short of having resiliency against disk failure, but that, of course, costs half of the capacity.)

 

[Arwen]

...
(You may use "copies=2" as a way to have at least resilency against read errors, short of having resiliency against disk failure, but that, of course, costs half of the capacity.)

One nice thing about "copies=2", is that it is per ZFS Dataset. If 90% of the data is okay with one copy, but if the other 10% is more important, you can put that data in a separate data set, with "copies=2".