Re: RAID1, a failed disk and performance

2005-01-28 Thread Christian Hiris
On Thursday 27 January 2005 23:12:23, Chad Morland wrote:
  http://members.chello.at/freebsd-5.3/bonnie-gmirror/summary
  http://members.chello.at/freebsd-5.3/bonnie-gmirror/detail

 I expect to see data transfer rate increase when you break the mirror.
 RAID1 has the higest disk overhead of all RAID configurations and is
 very inefficient in that regard. It would be interesting to see
 performance results with the bad disk still attached to the mirror.
 Unfortunately I am not able to break disks on a whim so I can't test
 it out. :P

I also run the bechmarks with the gnop class invoked. It seems that the 'gnop 
-f nnn' option doesn't work, so that I couldn't observe any significant 
performance differences. I own several really nice broken IBM SCSI disks, but 
they are broken in such a way, that the mirrors break immediately after 
hitting the damaged areas. Maybe it's the better solution to run smartd from 
ports/smartctl and replace flaky disks asap. 

-- 
Christian Hiris [EMAIL PROTECTED] | OpenPGP KeyID 0x3BCA53BE 
OpenPGP-Key at hkp://wwwkeys.eu.pgp.net and http://pgp.mit.edu


pgpQMiRLIbs6i.pgp
Description: PGP signature


RAID1, a failed disk and performance

2005-01-27 Thread Chad Morland
What happens in terms of performance when a drive in a RAID1 system
fails? Will disk access be slower because it attempts to read/write to
a failed disk or will performance be faster because it doesn't need to
do half the work it usually does? I couldn't really find any online
resources that deal with performance levels when there are failed
drives present in a RAID array.

-CM
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: RAID1, a failed disk and performance

2005-01-27 Thread Doug Poland
On Thu, Jan 27, 2005 at 03:14:46PM -0500, Chad Morland wrote:
 What happens in terms of performance when a drive in a RAID1 system
 fails? Will disk access be slower because it attempts to read/write to
 a failed disk or will performance be faster because it doesn't need to
 do half the work it usually does? I couldn't really find any online
 resources that deal with performance levels when there are failed
 drives present in a RAID array.
 
I recently set-up and tested a geom-based RAID1.  When I pulled a
hot-swap SATA drive from the server, I didn't notice any performance
degradation.  However, I must note that I wasn't running any monitoring
software nor gathering empirical data.  It's just my observation of the
responsiveness of the system while one drive was gone.  YMMV.

-- 
Regards,
Doug
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: RAID1, a failed disk and performance

2005-01-27 Thread Christian Hiris
On Thursday 27 January 2005 21:14:21, Chad Morland wrote:
 What happens in terms of performance when a drive in a RAID1 system
 fails? Will disk access be slower because it attempts to read/write to
 a failed disk or will performance be faster because it doesn't need to
 do half the work it usually does? I couldn't really find any online
 resources that deal with performance levels when there are failed
 drives present in a RAID array.

If you are interested in gmirror software-raid performance, I put some bonnie 
benchmark data online. I run the benchmark on a cheap none-raid 
Promise-Ultra-133-TX2 aka PDC20269, which costs about 25 Euros:

http://members.chello.at/freebsd-5.3/bonnie-gmirror/summary
http://members.chello.at/freebsd-5.3/bonnie-gmirror/detail

Cheers,
ch 

-- 
Christian Hiris [EMAIL PROTECTED] | OpenPGP KeyID 0x3BCA53BE 
OpenPGP-Key at hkp://wwwkeys.eu.pgp.net and http://pgp.mit.edu


pgpGoqCAnqt5g.pgp
Description: PGP signature


Re: RAID1, a failed disk and performance

2005-01-27 Thread Chad Morland
 http://members.chello.at/freebsd-5.3/bonnie-gmirror/summary
 http://members.chello.at/freebsd-5.3/bonnie-gmirror/detail

I expect to see data transfer rate increase when you break the mirror.
RAID1 has the higest disk overhead of all RAID configurations and is
very inefficient in that regard. It would be interesting to see
performance results with the bad disk still attached to the mirror.
Unfortunately I am not able to break disks on a whim so I can't test
it out. :P
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]


Re: RAID1, a failed disk and performance

2005-01-27 Thread Chuck Swiger
Chad Morland wrote:
What happens in terms of performance when a drive in a RAID1 system
fails? Will disk access be slower because it attempts to read/write to
a failed disk or will performance be faster because it doesn't need to
do half the work it usually does?
Read access will become slower.  Write access will become faster.
--
-Chuck
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to [EMAIL PROTECTED]