[Gluster-users] Why so bad performance?

[Kirby Zhou]

I have tried.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
volume writebehind
  type performance/write-behind
  option aggregate-size 8MB # default is 0bytes
  option flush-behind off   # default is 'off'
  subvolumes unify0
end-volume
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[@123.25 ~]# dd if=/opt/xxx of=/mnt/xxx
100353+0 records in
100353+0 records out
51380736 bytes (51 MB) copied, 129.343 seconds, 397 kB/s

[@123.25 ~]# dd if=/opt/xxx of=/mnt/xxx bs=16M
16+0 records in
16+0 records out
268435456 bytes (268 MB) copied, 23.5073 seconds, 11.4 MB/s

[@123.25 ~]# dd if=/opt/xxx of=/mnt/xxx bs=2M
128+0 records in
128+0 records out
268435456 bytes (268 MB) copied, 23.4631 seconds, 11.4 MB/s

[@123.25 ~]# dd if=/opt/xxx of=/mnt/xxx bs=1k
123905+0 records in
123905+0 records out
126878720 bytes (127 MB) copied, 54.4253 seconds, 2.3 MB/s

[@123.25 ~]# dd if=/opt/xxx of=/mnt/xxx bs=16k
16384+0 records in
16384+0 records out
268435456 bytes (268 MB) copied, 23.6589 seconds, 11.3 MB/s

-----Original Message-----
From: anand.avati at gmail.com [mailto:anand.avati at gmail.com] On Behalf Of
Anand Avati
Sent: Saturday, December 06, 2008 2:24 AM
To: Kirby Zhou
Cc: a s p a s i a; gluster-users at gluster.org
Subject: Re: [Gluster-users] Why so bad performance?

Kirby,
 please try by loading write-behind xlator on the client side.

avati

2008/12/5 Kirby Zhou <kirbyzhou at sohu-rd.com>:
> OK, They are HP 380 G4 rack server boxes.
> P4 Xeon 3.0GHz*2 with Hyper-Thread, 4G RAM.
> scsi 72g*5 raid5 for /exports/ns, sata 250G*7*2 raid5 for /exports/disk1
and
> /exports/disk2.
> All of the 4 boxes are connected with gigabit Ethernet switcher.
> 10.10.123.21 and 10.10.123.22 are servers, while 10.10.123.25 and
> 10.10.65.64 are clients.
>