AF:
NF:0
PS:10
SRH:1
SFN:
DSR:
MID:<20090128142052.3e28e44d@ripper.onstor.net>
CFG:
PT:0
S:andy.sharp@onstor.com
RQ:
SSV:exch1.onstor.net
NSV:
SSH:
R:<narayan.venkat@onstor.com>,<patrick.haverty@onstor.com>,<dl-Leopard@onstor.com>,<dl-performance@onstor.com>
MAID:1
X-Sylpheed-Privacy-System:
X-Sylpheed-Sign:0
SCF:#mh/Mailbox/sent
RMID:#imap/andys@onstor.net@exch1.onstor.net/INBOX	0	2779531E7C760D4491C96305019FEEB51851D41497@exch1.onstor.net
X-Sylpheed-End-Special-Headers: 1
Date: Wed, 28 Jan 2009 14:23:35 -0800
From: Andrew Sharp <andy.sharp@onstor.com>
To: Narayan Venkat <narayan.venkat@onstor.com>
Cc: Patrick Haverty <patrick.haverty@onstor.com>, dl-Leopard
 <dl-Leopard@onstor.com>, dl-Performance <dl-performance@onstor.com>
Subject: Re: Leopard Initial NFS Performance Results
Message-ID: <20090128142335.741b487e@ripper.onstor.net>
In-Reply-To: <2779531E7C760D4491C96305019FEEB51851D41497@exch1.onstor.net>
References: <2779531E7C760D4491C96305019FEEB51851D413F5@exch1.onstor.net>
	<2779531E7C760D4491C96305019FEEB51851D41497@exch1.onstor.net>
Organization: Onstor
X-Mailer: Sylpheed-Claws 2.6.0 (GTK+ 2.8.20; x86_64-pc-linux-gnu)
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

On Wed, 28 Jan 2009 14:02:44 -0800 Narayan Venkat
<narayan.venkat@onstor.com> wrote:

> Pat,
> 
> What is the recordsize of the ZFS filesystem?  If it is 128K, then it
> makes sense to use that for the sequential write & read numbers. 

Only for marketing numbers.  For comparative testing, you should use
the same size for all, regardless.

> As for random read and write, we should test with a recordsize that is
> less than 128K.  Something like 16K or 32K.  Otherwise, we are not
> leveraging the file systems capability to handle different workloads.

Hmm, how about 8K?  ~:^)

> Thanks.
> 
> Narayan Venkat
> Vice President, Marketing
> ONStor Inc. (www.onstor.com<http://www.onstor.com>)
> Tel: (408) 963-2404
> Cell: (408) 221-4297
> ________________________________
> From: Patrick Haverty
> Sent: Wednesday, January 28, 2009 11:07 AM
> To: dl-Leopard; dl-Performance
> Subject: Leopard Initial NFS Performance Results
> 
> The lower chart shows the results I'm getting now that the four
> clients have all been updated to a more recent Red Hat version.
> Interesting is that sequential write performance exceeds read
> performance with multiple clients.  Since the clients have been
> updated, results are much more repeatable than I was seeing
> previously.
> 
> Cheers,
> 
> Pat
