Xen users - Jun 2008 - Re: Re lease 0.9.9 of GPL PV Drivers for Windows

Thank you, the network driver is working perfectly.
But i have a strange issue; after the installation, the disk performance is
slow : 20MB/s write, 45MB/s on read.
Without /GPLPV (qemu hard disk), the result is 130MB/s read/write.
I''m physically connected on lvm device (SAN storage)

If my virtual machine is connected to a img file (/dev/loop0 mount on ext3
filesystem), with GPLPV driver i have the best performance : ~230MB/s

On the same virtual OS , fresh install 2003 server x64 r2.

sorry for my bad english..
-- 
View this message in context:
http://www.nabble.com/Release-0.9.9-of-GPL-PV-Drivers-for-Windows-tp17839379p17862198.html
Sent from the Xen - User mailing list archive at Nabble.com.


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

> Thank you, the network driver is working perfectly.
> But i have a strange issue; after the installation, the disk
performance
> is
> slow : 20MB/s write, 45MB/s on read.
> Without /GPLPV (qemu hard disk), the result is 130MB/s read/write.
> I''m physically connected on lvm device (SAN storage)
> 
> If my virtual machine is connected to a img file (/dev/loop0 mount on
ext3
> filesystem), with GPLPV driver i have the best performance : ~230MB/s
> 
> On the same virtual OS , fresh install 2003 server x64 r2.
You need to tell me what tool you are using to measure performance.

James

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

James Harper wrote:
> 
> You need to tell me what tool you are using to measure performance.
> 
> James
> 
Yes, i''m using bench32 (ATTO Disk Benchmark), this result is for a 1024
to
4096KB with 32MB total length (Direct I/O and Overlapped I/O, queue depth 4)

-- 
View this message in context:
http://www.nabble.com/Release-0.9.9-of-GPL-PV-Drivers-for-Windows-tp17839379p17863630.html
Sent from the Xen - User mailing list archive at Nabble.com.


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

> James Harper wrote:
> >
> > You need to tell me what tool you are using to measure performance.
> >
> > James
> >
> 
> Yes, i''m using bench32 (ATTO Disk Benchmark), this result is for a
1024 to
> 4096KB with 32MB total length (Direct I/O and Overlapped I/O, queue
depth
> 4)
> 
I haven''t tested that one. The poor performance may be due to unaligned
buffers. Could you please try testing with iometer and compare the pv
drivers vs the qemu drivers?

Thanks

James

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Ok , i''m sending you the results with iometer, but I don''t
know that one.
Thank you.

James Harper wrote:
> 
>> James Harper wrote:
>> >
>> > You need to tell me what tool you are using to measure
performance.
> 
> I haven''t tested that one. The poor performance may be due to
unaligned
> buffers. Could you please try testing with iometer and compare the pv
> drivers vs the qemu drivers?
> 
> Thanks
> 
> James
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
> 
> 
-- 
View this message in context:
http://www.nabble.com/Release-0.9.9-of-GPL-PV-Drivers-for-Windows-tp17839379p17864587.html
Sent from the Xen - User mailing list archive at Nabble.com.


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Hi Antolien,

please use 
	- a new & separate disk for http://www.iometer.org test 
	  (because of disk fragmentation issue)
	- a HDD-Size = 2 x (your RAM in your phy. Dom0 domain) 
	  (because of Caching issue''s !)

Maybe the following test will be interesting:

- 4k Block, 100% read, 100% sequential
- 4k Block, 100% write, 100% sequential
- 16k Block, 100% read, 100% sequential
- 16k Block, 100% write, 100% sequential
- 128k Block, 100% read, 100% sequential
- 128k Block, 100% write, 100% sequential
- 1 MB Block, 100% read, 100% sequential
- 1 MB Block, 100% write, 100% sequential
- 4k Block, 80% read, 20% write, 100% random

Regards
Danny

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

This is a bit off topic; but it would be nice if you could post the results 
of the test to the list.  I am in the process of benchmarking our systems 
with iometer and will learn a lot from seeing the discussion.

Thanks in advance,
Chris....

Daniel Schwager wrote:
> Hi Antolien,
> 
> please use 
> 	- a new & separate disk for http://www.iometer.org test 
> 	  (because of disk fragmentation issue)
> 	- a HDD-Size = 2 x (your RAM in your phy. Dom0 domain) 
> 	  (because of Caching issue''s !)
> 
> Maybe the following test will be interesting:
> 
> - 4k Block, 100% read, 100% sequential
> - 4k Block, 100% write, 100% sequential
> - 16k Block, 100% read, 100% sequential
> - 16k Block, 100% write, 100% sequential
> - 128k Block, 100% read, 100% sequential
> - 128k Block, 100% write, 100% sequential
> - 1 MB Block, 100% read, 100% sequential
> - 1 MB Block, 100% write, 100% sequential
> - 4k Block, 80% read, 20% write, 100% random
> 
> Regards
> Danny
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
> 
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Hi,
Sorry, some strange results with iometer; sometimes negative ?
and an error message : performance counter percentage is less than zero...
i''ll be back tomorow
Regards
Antolien


Daniel Schwager wrote:
> 
> Hi Antolien,
> 
> please use 
> 	- a new & separate disk for http://www.iometer.org test 
> 	  (because of disk fragmentation issue)
> ...
> Regards
> Danny
> 
-- 
View this message in context:
http://www.nabble.com/Release-0.9.9-of-GPL-PV-Drivers-for-Windows-tp17839379p17873524.html
Sent from the Xen - User mailing list archive at Nabble.com.


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

> Hi,
> Sorry, some strange results with iometer; sometimes negative ?
> and an error message : performance counter percentage is less than
zero...
> i''ll be back tomorrow
Are you using SMP? I see strange things happen with time under SMP,
things like ping reporting 1ms, 1ms, -22000ms, 1ms, etc. If you are
using SMP, try just a single processor and see if it makes any
difference.

James

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Monday June 16 2008 05:21:05 pm Antolien Hetzel
wrote:
> Sorry, some strange results with iometer; sometimes negative ?
> and an error message : performance counter percentage is less than zero...
> i''ll be back tomorow
Yeah, I get that all the time. I stop the test, then restart till I get a 
non-negative value for Total MB/s after the first results update.

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Mon, Jun 16, 2008 at 03:41:20PM +0200, Daniel Schwager
wrote:
> Hi Antolien,
> 
> please use 
Hi,

to compare the tests (i will also measure on our system), please tell
us some information about the configuration - without this info - the test
will be useless ..

- Dom0 Controller: Controllername / SAN-Name, Controller Cache-Size 
  RAID-Level, number of disks in RAID
  Diskname, Kind of disk (SATA, SAS...), Size of disk, Speed of disk (7.2k, 10k,
15k)

Also usefull will be a native iometer test in Dom0. Therefore, you can
install iometer (dynamo) in Dom0 and connect another instance of iometer running
on any windows (only the GUI) connecting to the "worker"-dynamo.

So, you can compare the results of the native iometer with the windows-xp-hvm
iometer results.

My tests will follow ..

Thx
Danny



_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Hi Daniel,

Dom0 controller: Emulex Corporation Zephyr-X LightPulse Fibre Channel Host
Adapter
SAN Name : MSA 2000 FC (HP),  cache size 2x1024MBytes
RAID 6 with 12x300GB SAS 15k rpm.

I think the performance on dom0 should be good :)
I will forward the test results as soon as possible.


Daniel Schwager wrote:
> 
> On Mon, Jun 16, 2008 at 03:41:20PM +0200, Daniel Schwager wrote:
>> Hi Antolien,
>> 
>> please use 
> 
> Hi,
> 
> to compare the tests (i will also measure on our system), please tell
> us some information about the configuration - without this info - the test
> will be useless ..
> 
> - Dom0 Controller: Controllername / SAN-Name, Controller Cache-Size 
>   RAID-Level, number of disks in RAID
>   Diskname, Kind of disk (SATA, SAS...), Size of disk, Speed of disk
> (7.2k, 10k, 15k)
> 
> Also usefull will be a native iometer test in Dom0. Therefore, you can
> install iometer (dynamo) in Dom0 and connect another instance of iometer
> running
> on any windows (only the GUI) connecting to the "worker"-dynamo.
> 
> So, you can compare the results of the native iometer with the
> windows-xp-hvm iometer results.
> 
> My tests will follow ..
> 
> Thx
> Danny
> 
> 
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
> 
> 
-- 
View this message in context:
http://www.nabble.com/Release-0.9.9-of-GPL-PV-Drivers-for-Windows-tp17839379p17882233.html
Sent from the Xen - User mailing list archive at Nabble.com.


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Hi,

I don''t know if the results is realy correct, but finaly GPLPV seems to
be
ok...

http://www.nabble.com/file/p17916466/benchmark1.png 


Antolien Hetzel wrote:
> 
> Hi Daniel,
> 
> Dom0 controller: Emulex Corporation Zephyr-X LightPulse Fibre Channel Host
> Adapter
> SAN Name : MSA 2000 FC (HP),  cache size 2x1024MBytes
> RAID 6 with 12x300GB SAS 15k rpm.
> 
> I think the performance on dom0 should be good :)
> I will forward the test results as soon as possible.
> 
> 
> Daniel Schwager wrote:
>> 
>> On Mon, Jun 16, 2008 at 03:41:20PM +0200, Daniel Schwager wrote:
>>> Hi Antolien,
>>> 
>>> please use 
>> 
>> Hi,
>> 
>> to compare the tests (i will also measure on our system), please tell
>> us some information about the configuration - without this info - the
>> test
>> will be useless ..
>> 
>> - Dom0 Controller: Controllername / SAN-Name, Controller Cache-Size 
>>   RAID-Level, number of disks in RAID
>>   Diskname, Kind of disk (SATA, SAS...), Size of disk, Speed of disk
>> (7.2k, 10k, 15k)
>> 
>> Also usefull will be a native iometer test in Dom0. Therefore, you can
>> install iometer (dynamo) in Dom0 and connect another instance of
iometer
>> running
>> on any windows (only the GUI) connecting to the
"worker"-dynamo.
>> 
>> So, you can compare the results of the native iometer with the
>> windows-xp-hvm iometer results.
>> 
>> My tests will follow ..
>> 
>> Thx
>> Danny
>> 
>> 
>> 
>> _______________________________________________
>> Xen-users mailing list
>> Xen-users@lists.xensource.com
>> http://lists.xensource.com/xen-users
>> 
>> 
> 
> 
-- 
View this message in context:
http://www.nabble.com/Release-0.9.9-of-GPL-PV-Drivers-for-Windows-tp17839379p17916466.html
Sent from the Xen - User mailing list archive at Nabble.com.


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

> I don''t know if the results is realy correct, but finaly GPLPV
seems
to be
> ok...
> 
> http://www.nabble.com/file/p17916466/benchmark1.png
> > Dom0 controller: Emulex Corporation Zephyr-X LightPulse Fibre
Channel Host
> > Adapter
> > SAN Name : MSA 2000 FC (HP),  cache size 2x1024MBytes
> > RAID 6 with 12x300GB SAS 15k rpm.
Whow - 
	DomU, 1MB 117MB/sec read,  85MB/sec write ... 

Very impressive !

Whow big was your iometer-file ?

Regards
Danny


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Realy ?
my iometer-file is 2xdom0 memory = 36GB


Daniel Schwager wrote:
> 
>> I don''t know if the results is realy correct, but finaly GPLPV
seems
> to be
>> ok...
>> 
>> http://www.nabble.com/file/p17916466/benchmark1.png
> 
> 
>> > Dom0 controller: Emulex Corporation Zephyr-X LightPulse Fibre
> Channel Host
>> > Adapter
>> > SAN Name : MSA 2000 FC (HP),  cache size 2x1024MBytes
>> > RAID 6 with 12x300GB SAS 15k rpm.
> 
> Whow - 
> 	DomU, 1MB 117MB/sec read,  85MB/sec write ... 
> 
> Very impressive !
> 
> Whow big was your iometer-file ?
> 
> Regards
> Danny
> 
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
> 
> 
-- 
View this message in context:
http://www.nabble.com/Release-0.9.9-of-GPL-PV-Drivers-for-Windows-tp17839379p17921341.html
Sent from the Xen - User mailing list archive at Nabble.com.


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Tuesday June 17 2008 12:02:54 pm Antolien Hetzel
wrote:
> I don''t know if the results is realy correct, but finaly GPLPV
seems to be
> ok...
>
> http://www.nabble.com/file/p17916466/benchmark1.png
One concern - how long did you let the test run? My tests are for 5 min., and 
the MB/s always goes down (steadily) from the initial number.

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users