raidz2 a bit big

peterjeremy at optushome.com.au peterjeremy at optushome.com.au
Tue May 26 18:59:05 UTC 2009


On 2009-May-23 15:37:14 +0300, Artis Caune <artis.caune at gmail.com> wrote:
>2009/5/23 Randy Bush <randy at psg.com>:
>> a dozen 2tb drives in a raidz2
>Reads on such configurations are very slow.

Not really.  Assuming each disk is capable of X IOPS and assuming
non-degraded mode, you can still issue (N-2)*X random reads/sec
because the parity stripes are not needed/used.  (Compared to N*X
random reads/sec for a mirrored configuration).  Degraded reads _are_
very slow because you need to read most of the spindles (I'm not sure
of the exact recovery mechanism for RAIDZ2 but it's probably close to
X random reads/sec).

Write performance _is_ poor - a write requires at least 2 (and maybe
3) physical reads and 3 physical writes.  But you can still perform
multiple parallel writes across the RAIDZ2 set, giving you roughly
N*X/6 random writes/sec.  (Compared to N*X/2 random writes/sec for
a mirrored configuration).

In general, you would not choose a RAIDZ{1,2} configuration where
write performance was an issue - the benefit you get is greater
storage utilisation.

>with 4 disks in raidz2 (total 3 raidz2 vdevs) you get 300 IOPS

Actually 600 reads/sec if all the sets are non-degraded.  Unless you
are particularly worried by dual disk failures, you would probably be
better off using a 2+2 disk mirror configuration, rather than a 4-disk
RAIDZ2 setup.

-- 
Peter Jeremy
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 196 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-fs/attachments/20090526/0a3fc33a/attachment.pgp


More information about the freebsd-fs mailing list