[OmniOS-discuss] virtualbox installed on omnios latest stable

[Jared Morrow]

The only performance numbers I really have lying around are from when I
changed over a centos 5 VM to the virtio driver. I just did a simple bonnie
test with a 2gb file and limited it to use 1gb of ram (on two cpu's). I
don't have numbers offhand for some other uses.

Here are the bonnie numbers from Centos 5 before virtio and after (and
centos 5 is fairly slow anyway, so your results may vary).
Before

[buildbot at riak-test-centos-5 ~]$ /usr/sbin/bonnie++ -d fstest/ -r 1000 -s 2000
Writing with putc()...done
Writing intelligently...done
Rewriting...done
Reading with getc()...done
Reading intelligently...done
start 'em...done...done...done...
Create files in sequential order...done.
Stat files in sequential order...done.
Delete files in sequential order...done.
Create files in random order...done.
Stat files in random order...done.
Delete files in random order...done.
Version  1.03       ------Sequential Output------ --Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
riak-test-cen 2000M 24003  30 22200   3 10379   1 67092  15 3253044
99 159.1   0
                    ------Sequential Create------ --------Random Create--------
                    -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                 16 18542  32 +++++ +++ +++++ +++ 21890   2 +++++ +++ +++++ +++
riak-test-centos-5,2000M,24003,30,22200,3,10379,1,67092,15,3253044,99,159.1,0,16,18542,32,+++++,+++,+++++,+++,21890,2,+++++,+++,+++++,+++

After

[buildbot at riak-test-centos-5 ~]$ /usr/sbin/bonnie++ -d fstest/ -r 1000 -s 2000
Writing with putc()...done
Writing intelligently...done
Rewriting...done
Reading with getc()...done
Reading intelligently...done
start 'em...done...done...done...
Create files in sequential order...done.
Stat files in sequential order...done.
Delete files in sequential order...done.
Create files in random order...done.
Stat files in random order...done.
Delete files in random order...done.
Version  1.03       ------Sequential Output------ --Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
riak-test-cen 2000M 33582  64 66872   8 24244   3 75227  24 3723163
99 546.8   0
                    ------Sequential Create------ --------Random Create--------
                    -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                 16 15421  26 +++++ +++ +++++ +++ 18583  55 +++++ +++ +++++ +++
riak-test-centos-5,2000M,33582,64,66872,8,24244,3,75227,24,3723163,99,546.8,0,16,15421,26,+++++,+++,+++++,+++,18583,55,+++++,+++,+++++,+++

As you can see, the change was quite dramatic.


Hope that helps at all,

-Jared


On Tue, May 28, 2013 at 1:01 PM, Geoff Nordli <geoffn at gnaa.net> wrote:

>  On 13-05-28 11:45 AM, Jared Morrow wrote:
>
>  What are you using to auto manage the startup/shutdown of KVM
>> instances.  I use
>> http://sourceforge.net/projects/vboxsvc/ with vbox.
>
>
>  In SmartOS I use their vmadm tool, and on OmniOS I just started it up by
> hand and haven't shut it down since.  There is a project that came up on
> this list to replicate vmadm on OmniOS, but the name escapes me right now.
>
>  Regarding high performance, I do some testing of Riak (our distributed
> data store) on SmartOS->KVM->Linux and as long as I use the virtio driver I
> get pretty decent performance.  For reference, FreeBSD and Centos5 need a
> little bit of work to get virtio to work, but it isn't that hard.
>
>
> That sounds encouraging.  Can you ballpark the performance numbers you are
> seeing ?
>
> thanks,
>
> Geoff
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://omniosce.org/ml-archive/attachments/20130528/828094f7/attachment-0001.html>