Re: read and write speed.

[Fyodor Ustinov <ufm@ufm.su>]

On 05/31/2011 05:51 PM, Gregory Farnum wrote:
> Sage created an issue to track this:
> http://tracker.newdream.net/issues/1122
>
> Based on the differences between the FUSE and kernel clients, it's
> probably just readahead settings. :)
> -Greg
I'm not sure.

rsize 10 times more than the default:

root@gate0:~# mount -t ceph -o rsize=5242880 10.5.51.230:/ /mnt
root@gate0:~# cd /mnt/
root@gate0:/mnt# dd if=bbb of=/dev/null bs=1024000
10000+0 records in
10000+0 records out
10240000000 bytes (10 GB) copied, 126.168 s, 81.2 MB/s
root@gate0:/mnt# cd
root@gate0:~# umount /mnt

rsize 100 times more

root@gate0:~# mount -t ceph -o rsize=52428800 10.5.51.230:/ /mnt
root@gate0:~# cd /mnt/
root@gate0:/mnt# dd if=bbb of=/dev/null bs=1024000
10000+0 records in
10000+0 records out
10240000000 bytes (10 GB) copied, 123.432 s, 83.0 MB/s
root@gate0:/mnt# cd
root@gate0:~# umount /mnt

cfuse

root@gate0:~# cfuse -m 10.5.51.230:/ /mnt
  ** WARNING: Ceph is still under development.  Any feedback can be 
directed  **
  **          at ceph-devel@vger.kernel.org or 
http://ceph.newdream.net/.     **
cfuse[1686]: starting ceph client
cfuse[1686]: starting fuse
root@gate0:~# cd /mnt
root@gate0:/mnt# dd if=bbb of=/dev/null bs=1024000
10000+0 records in
10000+0 records out
10240000000 bytes (10 GB) copied, 93.3285 s, 110 MB/s
root@gate0:/mnt#

WBR,
     Fyodor.