HP ProLiant DL385 G2 Server

[Edwin Groothuis]

On Thu, Aug 30, 2007 at 09:16:48AM +0300, Paulius Stakauskas wrote:
> Hi,
> 
> I'm running FreeBSD 6.2-RELEASE on this server.
> 
> # dmesg|grep ciss
> ciss0: <HP Smart Array P400> port 0x4000-0x40ff mem
> 0xfde00000-0xfdefffff,0xfddf0000-0xfddf0fff irq 18 at device 0.0 on pci6
> ciss0: [GIANT-LOCKED]
> da0 at ciss0 bus 0 target 0 lun 0
> da1 at ciss0 bus 0 target 1 lun 0
> 
> # camcontrol devlist -v
> scbus0 on ciss0 bus 0:
> <COMPAQ RAID 1  VOLUME OK>         at scbus0 target 0 lun 0 (pass0,da0)
> <COMPAQ RAID 5  VOLUME OK>         at scbus0 target 1 lun 0 (pass1,da1)
> scbus1 on ciss0 bus 32:
> scbus-1 on xpt0 bus 0:
> <  >                               at scbus-1 target -1 lun -1 (xpt0)
> 
> For first everything seems to be fine, but the problem is while copying 
> files from da0 to da1, take a look at the statistics:
> 
> File: 635 MB
> 
> copy from da0 to da1 = 58sec.
> copy from da1 to da0 = ~9sec.
> 
> Is this normal or not? I know that it's RAID-5 but.., I think it takes 
> too long to copy from da0 to da1.

That's indeed what I had too: (a posting from an internal mailinglist
at february 2006)

On a RAID0 disk, 16384 blocks of 16384 bytes:

    [~] root at cleo>dd if=/dev/zero of=/mnt/foo bs=16384 count=16384
    16384+0 records in
    16384+0 records out
    268435456 bytes transferred in 2.167275 secs (123858514 bytes/sec)

On a RAID1 disk, 16384 blocks of 16384 bytes:

    [~] root at janus>dd if=/dev/zero of=/usr/foo bs=16384 count=16384
    16384+0 records in
    16384+0 records out
    268435456 bytes transferred in 4.020692 secs (66763495 bytes/sec)

On a RAID5 disk, 16384 blocks of 16384 bytes:

    [~] root at janus>dd if=/dev/zero of=/mnt/databases/foo bs=16384 count=16384
    16384+0 records in
    16384+0 records out
    268435456 bytes transferred in 20.688034 secs (12975397 bytes/sec)

On a RAID5 disk, 16384 blocks of 16384 bytes:

    [~] root at cleo>dd if=/dev/zero of=/usr/foo bs=16384 count=16384
    16384+0 records in
    16384+0 records out
    268435456 bytes transferred in 17.556463 secs (15289837 bytes/sec)

RAID0:          2.2 secs
RAID1:          4.0 secs
RAID5:          17.6 / 20.7 secs




A little bit bigger: 65535 64K byte blocks: (yes that's 16x)

RAID0:
    [~] root at cleo>dd if=/dev/zero of=/mnt/foo bs=64k count=65535
    65535+0 records in
    65535+0 records out
    4294901760 bytes transferred in 35.805593 secs (119950583 bytes/sec)

RAID1:
    [~] root at janus>dd if=/dev/zero of=/usr/foo bs=64k count=65535
    65535+0 records in
    65535+0 records out
    4294901760 bytes transferred in 68.338178 secs (62847765 bytes/sec)

RAID5:
    [~] root at janus>dd if=/dev/zero of=/mnt/databases/foo bs=64k count=65535
    65535+0 records in
    65535+0 records out
    4294901760 bytes transferred in 337.764817 secs (12715658 bytes/sec)

    [~] root at cleo>dd if=/dev/zero of=/usr/foo bs=64k count=65535
    65535+0 records in
    65535+0 records out
    4294901760 bytes transferred in 271.828036 secs (15800069 bytes/sec)

RAID0:          35.8 secs
RAID1:          68 secs
RAID5:          271 / 337 secs


At the end we just throw our databases on RAID1 partitions

Edwin

-- 
Edwin Groothuis      |            Personal website: http://www.mavetju.org
edwin at mavetju.org    |              Weblog: http://www.mavetju.org/weblog/