Re: Software vs Hardware RAID 10?
On Thu, Aug 30, 2007 at 10:03:34AM +0200, Goswin von Brederlow wrote:
> You can test this with for example:
>
> dd if=/dev/md0 of=/dev/zero bs=1M count=1024 &
> dd if=/dev/md0 of=/dev/zero bs=1M count=1024 skip=1024
The machine at my office (kernel is vanilla 2.6.22.1, i386):
# (dd if=/dev/md1 of=/dev/zero bs=1M count=4096 iflag=direct &); dd if=/dev/md1 of=/dev/zero bs=1M count=4096 skip=1024 iflag=direct
4096+0 records in
4096+0 records out
4294967296 bytes (4.3 GB) copied, 81.1718 seconds, 52.9 MB/s
4096+0 records in
4096+0 records out
4294967296 bytes (4.3 GB) copied, 142.024 seconds, 30.2 MB/s
That's about what "hdparm -t" says for the disks (yes, it sucks when one
of the disks is significantly slower...). I added "iflag=direct" to
ensure that the read requests really hit the disk and are not served
from the page cache.
On a somewhat saner hardware (kernel is Debian 2.6.20-1-amd64):
# (dd if=/dev/md1 of=/dev/zero bs=1M count=4096 iflag=direct &); dd if=/dev/md1 of=/dev/zero bs=1M count=4096 skip=1024 iflag=direct
4096+0 records in
4096+0 records out
4294967296 bytes (4.3 GB) copied, 73.1666 seconds, 58.7 MB/s
4096+0 records in
4096+0 records out
4294967296 bytes (4.3 GB) copied, 75.9613 seconds, 56.5 MB/s
In both cases both disks are over 99% utilization during the test
according to iostat.
Gabor
--
---------------------------------------------------------
MTA SZTAKI Computer and Automation Research Institute
Hungarian Academy of Sciences,
Laboratory of Parallel and Distributed Systems
Address : H-1132 Budapest Victor Hugo u. 18-22. Hungary
Phone/Fax : +36 1 329-78-64 (secretary)
W3 : http://www.lpds.sztaki.hu
---------------------------------------------------------
Reply to: