set up and configure a ZFS storage pool under Solaris 10

This is a brief example of setting up a ZFS storage pool and a couple of filesystems ("datasets") on Solaris 10. You'll need the Solaris Update 2 release (dated June 2006) or later, as ZFS was not in the earlier standard releases of Solaris 10.

Once you have the Solaris system up and running, this is how to proceed:-

1. You'll need a spare disk partition or two for the storage pool. You can
use files (see mkfile) if you're stuck for spare hard partitions, but only for experimentation. The partitions
will be OVERWRITTEN by this procedure.

2. Login as root and create a storage pool using the spare partitions:-
# zpool  create  lake  c0t2d0s0 c0t2d0s1
(The # indicates the type-in prompt)

If the above partitions contain exiting file systems, you may need to use
the -f (force) option:-
# zpool  create -f  lake  c0t2d0s0 c0t2d0s1

The pool(called lake) has been created.

3. Use zpool list to view your pool stats:-
# zpool list
NAME    SIZE            USED    AVAIL   CAP     HEALTH          ALTROOT
lake        38.2G           32.5K   38.2G   0%      ONLINE          -
 A df listing will also show it:-
# df  -h  /lake
Filesystem      size    used    avail   capacity  Mounted on
lake            38G     8K      38G     1%              /lake

4. Now create a file system within the pool (Oracle call these datasets):-

# zfs create lake/fa
# zfs list
NAME            USED            AVAIL   REFER   MOUNTPOINT
lake                43.0K           38.0G   8.50K   /lake
lake/fa             8K               38.0G   8K      /lake/fa

To destroy a storage pool (AND ALL ITS DATA)
# zpool destroy lake
To destroy a dataset:-
# zfs destroy lake/fa

5. zpool can also create mirror devices:-
# zpool  create  lake  mirror  c0t2d0s0   c0t2d0s1
and something called RAID Z (similar to RAID 5):-

# zpool  create lake  raidz  c0t2d0s0 c0t2d0s1 c0t2d0s3

6. To add further devices to a pool (not mirrors or RAIDZ - these must be extended only with similar datasets):-
# zpool  add  lake  c0t2d0s3  c0t2d0s4
# zpool   status
  pool: lake
 state: ONLINE
 scrub: none requested
config:

        NAME                    STATE                   READ WRITE CKSUM
        lake                    ONLINE                  0       0       0
        c0t2d0s0                ONLINE                  0       0       0
        c0t2d0s1                ONLINE                  0       0       0
        c0t2d0s3                ONLINE                  0       0       0
        c0t2d0s4                ONLINE                  0       0       0

Note that datasets are automatically mounted, so no more
updating of /etc/vfstab!

You can also offline/online components, take snapshots (recursive since update 3), clone
filesystems, and apply properties such as quotas, NFS sharing, etc.
Entire pools can be exported, then imported on another system, Intel or SPARC.

There are even built-in backup and restore facilities, not to mention
performance tools:-

# zpool  iostat  -v