[mythtv-users] OT Hardware RAID 5 Recommendations

Blammo blammo.doh at gmail.com
Tue Dec 12 21:40:55 UTC 2006

On 12/12/06, Paul Hamm <paulhamm at verizon.net> wrote:
> Interesting ran this on my machine.  See the bottom for PC specs.  The
> system is running 2 instances of FAH.  Not a very consistent test.

Here's the specs on my main backend:

tyan mpx dual motherboard, athlon 2400's, 2gig's ram, 3ware 9500S-12
(256M cache) in PCI-X 66mhz slot, 11(+1)x160gig SATA drives, HW Raid5,
XFS file system. Looks like this:

/dev/sda1              1.6T   1.2T   427G  74% /raid

write	1024	11.789	86.8
read	1024	11.648	87.9
write	2048	30.056	68.1
read	2048	20.27	101.0

(note, both raw & bonnie benchmarks were run while the system was
fairly busy --- 2 tuners going, 1 remote frontend playing, 1 remote
commercial flagging running)

For raw disk performance, I find a much better tool to use is
Bonnie++, which tests things like writes a second / reads a second,
CPU load while this is happening, etc. I run it with the following

[root at backend1 ~]# bonnie++ -d /raid/temp -u0 -g0 -n0 -s 2048

Version  1.03       ------Sequential Output------ --Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
backend1         2G 37034  96 67223  34 39536  19 38805  94 140993  43 442.5   1

Here's the very best part... I could have left those benchmarks
running for 60 minutes straight (done it as a test) and I would not
have dropped a video frame, chugged out a recording, or caused any
instability in the box. The HW Raid card (true HW Raid, not psuedo
rocket cards) was worth every penny.

More information about the mythtv-users mailing list