Using HDD's for ZFS: 'desktop' vs 'raid / enterprise' -edition drives?

[Charles Richards]

Just FWIW, the "proper" way to do "online expansion" with ZFS is quite  
different than that of a RAID controller, as doing a regular OCE  
process is quite difficult in ZFS
(due to the variable-size stripes that ZFS uses, which makes the  
simple expansion of regular RAID-5 or RAID-6 not workable - or so I've  
read on the ZFS mailing lists)

The way that I believe you can add capacity to an existing RAIDZ /  
RAIDZ2 is as follows:

(this is based directly off of their documentation: http://docs.sun.com/app/docs/doc/819-5461/gcvjg?a=view)

Your ZFS "pool" is made up of "vdevs", which  are in turn made up of  
raw disks / partitions / unicorns, other block devices.

To increase the capacity of your existing pool, you simply add another  
"vdev" to it, and ZFS will intelligently start distributing the data  
between the two vdevs.

The example, straight from their documentation (with some commentary  
from me) is an existing ZFS pool named "rpool", which as 3 disks and  
is in a RAIDZ configuration

# zpool status
   pool: rpool
  state: ONLINE
  scrub: none requested
config:
         NAME         STATE     READ WRITE CKSUM
         rpool           ONLINE       0     0     0
           raidz1       ONLINE       0     0     0
             c1t2d0   ONLINE       0     0     0
             c1t3d0   ONLINE       0     0     0
             c1t4d0   ONLINE       0     0     0

errors: No known data errors

Now, we want to add three *new* disks to the storage pool - to do  
this, we simply add a new RAIDZ device comprised of our three new  
disks into the existing pool:

# zpool add rpool raidz c2t2d0 c2t3d0 c2t4d0
# zpool status
   pool: rpool
  state: ONLINE
  scrub: none requested
config:
         NAME         STATE     READ WRITE CKSUM
         rpool           ONLINE       0     0     0
           raidz1       ONLINE       0     0     0
             c1t2d0   ONLINE       0     0     0
             c1t3d0   ONLINE       0     0     0
             c1t4d0   ONLINE       0     0     0
           raidz1       ONLINE       0     0     0
             c2t2d0   ONLINE       0     0     0
             c2t3d0   ONLINE       0     0     0
             c2t4d0   ONLINE       0     0     0

errors: No known data errors


If you're looking for more info on ZFS, I suggest looking at the Sun  
ZFS administration guide: http://docs.sun.com/app/docs/doc/819-5461/gavwn?a=browse
Particularly, the section #5, on ZFS Storage Pools: http://docs.sun.com/app/docs/doc/819-5461/gavwn?a=view

Charles Richards
richardsc at gmail.com
charlesrichards.net





On Jan 3, 2009, at 5:19 AM, Frederique Rijsdijk wrote:

> After some reading, I come back from my original idea. Main reason  
> is I'd like to be able to grow the fs as the need develops in time.
>
> One could create a raidz zpool with a couple of disks, but when  
> adding a disk later on, it will not become part of the raidz (I  
> tested this).
>
> It seems vdevs can not be nested (create raidz sets and join them as  
> a whole), so I came up with the following:
>
> Start out with 4*1TB, and use geom_raid5 to create an independent  
> redundant pool of storage:
>
> 'graid5 label -v graid5a da0 da1 da2 da3'  (this is all tested in  
> vmware, one of these 'da' drives is 8GB)
>
> Then I 'zpool create bigvol /dev/raid5/graid5a', and I have a / 
> bigvol of 24G - sounds about right to me for a raid5 volume.
>
> Now lets say later in time I need more storage, I buy another 4 of  
> these drives, and
>
> 'graid5 label -v graid5b da4 da5 da6 da7'
> and
> 'zpool add bigvol /dev/raid5/graid5b'
>
> Now my bigvol is 48G. Very cool! Now I have redundant storage that  
> can grow and it's pretty easy too.
>
> Is this OK (besides from the fact that graid5 is not in production  
> yet, nor is ZFS ;) or are there easier (or better) ways to do this?
>
> - So I want redundancy (I don't want one failing drive to cause me  
> to loose all my data)
> - I want to be able to grow the filesystem if I need to, by adding a  
> (set of) drive(s) later on.
>
>
>
> -- FR
> _______________________________________________
> freebsd-questions at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-questions
> To unsubscribe, send any mail to "freebsd-questions-unsubscribe at freebsd.org 
> "