Disk access mode under Linux

View: New views
4 Messages — Rating Filter:   Alert me  

Disk access mode under Linux

by Velu Erwan :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Hey,

I'm using "iometer-2008-06-22-rc2.src.tgz".
<http://www.mirrorservice.org/sites/download.sourceforge.net/pub/sourceforge/i/io/iometer/iometer-2008-06-22-rc2.src.tgz>

Reading the changelog I can see the following :
   - Patch by Christensen ReedX K to use O_DIRECT as default for Linux

Grepping the sources for the "open()" calls and grepping for O_DIRECT
let me think that IOmeter doesn't use the O_DIRECT mode.

Does IOmeter really use the O_DIRECT access mode under Linux ?

Cheers,
Erwan,

------------------------------------------------------------------------------
_______________________________________________
Iometer-user mailing list
Iometer-user@...
https://lists.sourceforge.net/lists/listinfo/iometer-user

Re: Disk access mode under Linux

by Ming Zhang-3 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

we disable and enable this back and forth because of various problem
under linux with O_DIRECT...



On Fri, Mar 27, 2009 at 8:50 AM, Erwan Velu <erwan@...> wrote:

> Hey,
>
> I'm using "iometer-2008-06-22-rc2.src.tgz".
> <http://www.mirrorservice.org/sites/download.sourceforge.net/pub/sourceforge/i/io/iometer/iometer-2008-06-22-rc2.src.tgz>
>
> Reading the changelog I can see the following :
>   - Patch by Christensen ReedX K to use O_DIRECT as default for Linux
>
> Grepping the sources for the "open()" calls and grepping for O_DIRECT
> let me think that IOmeter doesn't use the O_DIRECT mode.
>
> Does IOmeter really use the O_DIRECT access mode under Linux ?
>
> Cheers,
> Erwan,
>
> ------------------------------------------------------------------------------
> _______________________________________________
> Iometer-user mailing list
> Iometer-user@...
> https://lists.sourceforge.net/lists/listinfo/iometer-user
>

------------------------------------------------------------------------------
_______________________________________________
Iometer-user mailing list
Iometer-user@...
https://lists.sourceforge.net/lists/listinfo/iometer-user

Changes for IOMETER 2006 to 2008 and impact for SSD's

by Mark_Lindholm :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Following is from the change list for IOMETER 2008:

     Patch by Kris Murray and Allen Wayne from Intel on randomizing the
written
     data. The previous implementation of writing 0's causes an
un-desired side-effect
     on some media, especially flash-based devices. The writing of
random data
     creates a realistic workload for these new devices. (Also data
patten will have impact
     on protocol like iSCSI as well).

I thought IOMETER always wrote random data.  Can I get some more info on
the change and what impact it has on SSD performance.

Mark Lindholm
Enterprise HDD Engineering
512 725 3195
mark_lindholm@...


------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Iometer-user mailing list
Iometer-user@...
https://lists.sourceforge.net/lists/listinfo/iometer-user

Re: Changes for IOMETER 2006 to 2008 and impact for SSD's

by Mark_Lindholm :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

I would really like to understand more about this change.  I thought
IOMETER 2006 wrote pseudo-random data, is that true or not?

Mark Lindholm
Enterprise HDD Engineering
512 725 3195
mark_lindholm@...


-----Original Message-----
From: Lindholm, Mark
Sent: Wednesday, August 19, 2009 5:04 PM
To: iometer-user@...
Subject: Changes for IOMETER 2006 to 2008 and impact for SSD's

Following is from the change list for IOMETER 2008:

     Patch by Kris Murray and Allen Wayne from Intel on randomizing the
written
     data. The previous implementation of writing 0's causes an
un-desired side-effect
     on some media, especially flash-based devices. The writing of
random data
     creates a realistic workload for these new devices. (Also data
patten will have impact
     on protocol like iSCSI as well).

I thought IOMETER always wrote random data.  Can I get some more info on
the change and what impact it has on SSD performance.

Mark Lindholm
Enterprise HDD Engineering
512 725 3195
mark_lindholm@...


------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Iometer-user mailing list
Iometer-user@...
https://lists.sourceforge.net/lists/listinfo/iometer-user