* Life's too short ...(Raid 1 sync times)
@ 2004-02-29 19:27 Richard Scobie
2004-02-29 19:40 ` Jeff Garzik
` (2 more replies)
0 siblings, 3 replies; 4+ messages in thread
From: Richard Scobie @ 2004-02-29 19:27 UTC (permalink / raw)
To: linux-raid
I am in the process of benchmarking a 4 x 250GB on 3ware 7506 to try and
sort out the myriad of options for RAID and XFS, in order to get best
performance for my workload.
I have tested with software RAID5 and have just attempted to do the same
for software RAID 10. Unfortunately, it is going to take 31000 minutes,
or around 21 days to sync this array and I wish to test 3 different
stripe sizes. Is there some way to allow more bandwidth for the sync up?
The RAID 5 took 70 minutes and was allowed up 100MB/s of idle IO
bandwidth - RAID 1 seems to use < 200kB/s.
While I am very pleased with the hardware RAID 10 result (98261kB write,
97148kB read), I would prefer to use software RAID 10 for a few reasons,
if the performance is similar.
Regards,
Richard
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Life's too short ...(Raid 1 sync times)
2004-02-29 19:27 Life's too short ...(Raid 1 sync times) Richard Scobie
@ 2004-02-29 19:40 ` Jeff Garzik
2004-02-29 20:47 ` Gordon Henderson
2004-02-29 21:29 ` Måns Rullgård
2 siblings, 0 replies; 4+ messages in thread
From: Jeff Garzik @ 2004-02-29 19:40 UTC (permalink / raw)
To: Richard Scobie; +Cc: linux-raid
Richard Scobie wrote:
> I am in the process of benchmarking a 4 x 250GB on 3ware 7506 to try and
> sort out the myriad of options for RAID and XFS, in order to get best
> performance for my workload.
>
> I have tested with software RAID5 and have just attempted to do the same
> for software RAID 10. Unfortunately, it is going to take 31000 minutes,
> or around 21 days to sync this array and I wish to test 3 different
> stripe sizes. Is there some way to allow more bandwidth for the sync up?
>
> The RAID 5 took 70 minutes and was allowed up 100MB/s of idle IO
> bandwidth - RAID 1 seems to use < 200kB/s.
>
> While I am very pleased with the hardware RAID 10 result (98261kB write,
> 97148kB read), I would prefer to use software RAID 10 for a few reasons,
> if the performance is similar.
The resync speed defaults to "nice and slow" IIRC... There should be
some procfs tunable to increase the resync speed...
Jeff
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Life's too short ...(Raid 1 sync times)
2004-02-29 19:27 Life's too short ...(Raid 1 sync times) Richard Scobie
2004-02-29 19:40 ` Jeff Garzik
@ 2004-02-29 20:47 ` Gordon Henderson
2004-02-29 21:29 ` Måns Rullgård
2 siblings, 0 replies; 4+ messages in thread
From: Gordon Henderson @ 2004-02-29 20:47 UTC (permalink / raw)
To: Richard Scobie; +Cc: linux-raid
On Mon, 1 Mar 2004, Richard Scobie wrote:
> I have tested with software RAID5 and have just attempted to do the same
> for software RAID 10. Unfortunately, it is going to take 31000 minutes,
> or around 21 days to sync this array and I wish to test 3 different
> stripe sizes. Is there some way to allow more bandwidth for the sync up?
echo 999999 > /proc/sys/dev/raid/speed_limit_max
Gordon
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Life's too short ...(Raid 1 sync times)
2004-02-29 19:27 Life's too short ...(Raid 1 sync times) Richard Scobie
2004-02-29 19:40 ` Jeff Garzik
2004-02-29 20:47 ` Gordon Henderson
@ 2004-02-29 21:29 ` Måns Rullgård
2 siblings, 0 replies; 4+ messages in thread
From: Måns Rullgård @ 2004-02-29 21:29 UTC (permalink / raw)
To: linux-raid
Richard Scobie <richard@sauce.co.nz> writes:
> I am in the process of benchmarking a 4 x 250GB on 3ware 7506 to try and
> sort out the myriad of options for RAID and XFS, in order to get best
> performance for my workload.
>
> I have tested with software RAID5 and have just attempted to do the same
> for software RAID 10. Unfortunately, it is going to take 31000 minutes,
> or around 21 days to sync this array and I wish to test 3 different
> stripe sizes. Is there some way to allow more bandwidth for the sync up?
echo 200000 > /proc/sys/dev/raid/speed_limit_min
It's a known bug. The resync code is trying to be nice to other IO in
the system, but forgets that it is itself causing lots of IO in the
lower RAID layer.
--
Måns Rullgård
mru@kth.se
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2004-02-29 21:29 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-02-29 19:27 Life's too short ...(Raid 1 sync times) Richard Scobie
2004-02-29 19:40 ` Jeff Garzik
2004-02-29 20:47 ` Gordon Henderson
2004-02-29 21:29 ` Måns Rullgård
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox