-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

kashani wrote:
> Sean Cook wrote:
> 
>> This is not necessarily the case, typically if you are CPU bound then
>> the above statement is true.  If you are IO bound, then software raid
>> offers, overall, better throughput to the disk and allows for more
>> flexibility.
>>
>> Back in the day... we were running the LSI MegaRAID controllers with
>> RAID 1 on 9gig SCSI disks, we were choking our application
>> (mysql/mod_perl) we took the MegaRAID controller out and performance
>> increased by about 25%.  (note: this is a real world situation)
>>
> 
> Then you my friend had some seriously broken hardware. I've never seen
> anything get slower by moving to dedicated RAID card and usually it got
> 2-3x faster due to having a dedicated local cache on the card.

No... there are plenty of people who have experienced the same thing...

infact, http://linas.org/linux/raid-reviews.html

shows that the through put is more that 5x's the speed of the hardware
raid.

Machine    MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU
K/sec %CPU
(DPT hardware RAID5, 3 disks)
DPT3x4G  1000  1914 20.0  1985  2.8  1704  6.5  5559 86.7 12857 15.6
97.1  1.8

(Linux soft RAID5, 3 disks)
SOF3x4G  1000  7312 76.2 10908 15.5  5757 20.2  5434 86.4 14728 19.9
69.3  1.5


Redhat In their manual:
"With today's fast CPUs, Software RAID performance can excel against
Hardware RAID."

As I stated before, it depends on where your bottleneck is... if you are
not cpu bound, software raid is great! and will boost IO through put on
comparable hardware.  If you are already CPU bound, forget software
raid, it will degrade your system to a crawl...

Sean
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (MingW32)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFD4R0Uy9wPyZpnL2URApqnAJ9t4K297aTbMXPaOWTE3T+E/jrC6wCaAlDY
fjFntqWwbG6oQIyGHpsntw8=
=RtK2
-----END PGP SIGNATURE-----
-- 
[email protected] mailing list

Reply via email to