On 14 Apr 08, at 1140, tkimball wrote:
Ian, in regards to your question - our own concern is with ZFS in
general. Though we're happy with its performance in a Development
environment, we're leery of implementing it in production until we
have it more widely deployed on the more 'stressful' (i.e. everyone
runs something all the time) test servers and see what happens.
There's a performance hit at >95% occupancy under high load, otherwise
it's been solid for us for some years.
We also don't like the idea that a corruption of the 'master
indexing file' held on the boot disk will effectively wipe our
entire configuration without recovery (we use our adv_file disks as
a sort of 'data warehouse' for certain types of backup, so this is a
showstopper).
I don't follow. Are you referring to /etc/zfs/zpool.cache? That only
changes when you change the configuration of the zpool (ie adding and
removing disk devices), is only a few kilobytes in size and is
trivially easily backed up.
# pwd
/etc/zfs
# ls -ltr
total 16
-rw-r--r-- 1 root root 8024 Feb 29 07:56 zpool.cache
# zfs create pool1/throwaway
# ls -ltr
total 16
-rw-r--r-- 1 root root 8024 Feb 29 07:56 zpool.cache
# zfs destroy pool1/throwaway
#
Once you have the pools, the filesystems within them are recorded in
the pools.
Were you to lose this file, `zpool import' searches all the devices
in /dev/dsk and attempts to stitch back together the pools that it
finds there. You can even import pools that have been destroyed.
Read, for example:
http://docs.huihoo.com/opensolaris/solaris-zfs-administration-guide/html/ch04s06.html
Lastly, from our own testing here, RAID-5 on hardware is usually
more efficient on our SPARC servers than any software-based one (the
sole exception, paradoxically, being Veritas VxVM on a FC JBOD).
You can (indeed, we do) use ZFS on hardware RAID-5. We just create
LUNs and then catenate them. pool1 and pool2 span 48-disk DotHill
boxes, each sub-device being a RAID 5 group:
# zpool status
pool: onboard
state: ONLINE
scrub: none requested
config:
NAME STATE READ WRITE CKSUM
onboard ONLINE 0 0 0
mirror ONLINE 0 0 0
c0t0d0s4 ONLINE 0 0 0
c1t0d0s4 ONLINE 0 0 0
mirror ONLINE 0 0 0
c0t1d0s4 ONLINE 0 0 0
c1t1d0s4 ONLINE 0 0 0
errors: No known data errors
pool: pool1
state: ONLINE
scrub: none requested
config:
NAME STATE READ WRITE
CKSUM
pool1 ONLINE 0
0 0
c7t600C0FF0000000000992DD1E11EA7D00d0 ONLINE 0
0 0
c7t600C0FF0000000000992DD2D599A5300d0 ONLINE 0
0 0
c7t600C0FF0000000000992DD02D3066400d0 ONLINE 0
0 0
c7t600C0FF0000000000992DD39CBE66800d0 ONLINE 0
0 0
c7t600C0FF0000000000992DD152E5BE700d0 ONLINE 0
0 0
errors: No known data errors
pool: pool2
state: ONLINE
scrub: none requested
config:
NAME STATE READ WRITE
CKSUM
pool2 ONLINE 0
0 0
c7t600C0FF0000000000992DB2E5C0D7200d0 ONLINE 0
0 0
c7t600C0FF0000000000992DB4F24F7AE00d0 ONLINE 0
0 0
c7t600C0FF0000000000992DB6B49A6CB00d0 ONLINE 0
0 0
c7t600C0FF0000000000992DB037012C900d0 ONLINE 0
0 0
c7t600C0FF0000000000992DB149204DE00d0 ONLINE 0
0 0
errors: No known data errors
#
ian
To sign off this list, send email to listserv AT listserv.temple DOT edu and type
"signoff networker" in the body of the email. Please write to networker-request
AT listserv.temple DOT edu if you have any problems with this list. You can access the
archives at http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
|