Fwd: [Gluster-devel] Performance

[Steffen Grunewald]

On Mon, Oct 15, 2007 at 02:09:30PM +0530, Anand Avati wrote:
> Copying gluster-devel@
> 
> From: Brian Taber <btaber at diversecg.com>
> Date: Oct 15, 2007 3:59 AM
> Subject: Re: [Gluster-devel] Performance
> To: Anand Avati <avati at zresearch.com>
> 
> Now there's a beautiful thing....
> 
> NFS Write:
> time dd if=/dev/zero bs=65536 count=15625 of=/shared/1Gb.file
> 1024000000 bytes (1.0 GB) copied, 62.9818 seconds, 16.3 MB/s
> 
> Gluster Write:
> time dd if=/dev/zero bs=65536 count=15625 of=/mnt/glusterfs/1Gb.file
> 1024000000 bytes (1.0 GB) copied, 41.74 seconds, 24.5 MB/s
> 
> NFS Read:
> time dd if=/shared/1Gb.file bs=65536 count=15625 of=/dev/zero
> 1024000000 bytes (1.0 GB) copied, 44.4734 seconds, 23.0 MB/s
> 
> Gluster Read:
> time dd if=/mnt/glusterfs/1Gb.file bs=65536 count=15625 of=/dev/zero
> 1024000000 bytes (1.0 GB) copied, 42.1526 seconds, 24.3 MB/s
> 
> 
> this test is performed within a VMWare virtual machine, so network speed
> isn't as good.   I tried it from outsite, 1000MB network:
> 
> 
> 
> NFS Write:
> time dd if=/dev/zero bs=65536 count=15625 of=/shared/1Gb.file
> 1024000000 bytes (1.0 GB) copied, 27.619 seconds, 37.1 MB/s
> 
> Gluster Write:
> time dd if=/dev/zero bs=65536 count=15625 of=/mnt/glusterfs/1Gb.file
> 1024000000 bytes (1.0 GB) copied, 11.1978 seconds, 91.4 MB/s
> 
> NFS Read:
> time dd if=/shared/1Gb.file bs=65536 count=15625 of=/dev/zero
> 1024000000 bytes (1.0 GB) copied, 43.5323 seconds, 23.5 MB/s
> 
> Gluster Read:
> time dd if=/mnt/glusterfs/1Gb.file bs=65536 count=15625 of=/dev/zero
> 1024000000 bytes (1.0 GB) copied, 30.6922 seconds, 33.4 MB/s

What's not so beautiful is that the first dd (always nfs) does
include staging of the file from the input media into buffer
cache (/dev/zero means: filling memory with zero bytes, which
certainly is faster than reading from a physical disk).
I would have repeated the write tests to see whether ordering
is important:
- nfs write
- glusterfs write
- nfs write again
- glusterfs write again
Buffers are often able to fool the benchmarker.

Also some information about your machine is missing - but I suppose 1GB 
would easily fit into main memory. What about *several* GBs to effectively
trash the page cache?

Cheers,
 Steffen, always doubtful when it comes to benchmarks

-- 
Steffen Grunewald * MPI Grav.Phys.(AEI) * Am Mühlenberg 1, D-14476 Potsdam
Cluster Admin * http://pandora.aei.mpg.de/merlin/ * http://www.aei.mpg.de/
* e-mail: steffen.grunewald(*)aei.mpg.de * +49-331-567-{fon:7233,fax:7298}
No Word/PPT mails - http://www.gnu.org/philosophy/no-word-attachments.html