LINUX vs FreeBSD mysql performance using a large RT database

[Mike Tancsa]

At 04:06 PM 10/20/2006, Ed Maste wrote:

>On Fri, Oct 20, 2006 at 02:57:46PM -0400, Mike Tancsa wrote:
>
> > With all the threads about poor FreeBSD performance, I wanted to test
> > it out myself to see how 64bit LINUX would compare using the same hardware.
>[ snip ]
>
>It seems your message ended up with some unfortunate line wrapping,
>which made it a little hard to see at a glance what the results were.

Sorry about that, my 21 inch Radiation Master has a very wide res ;-)


>Scott sent me them reformatted as:
>
>                                                   BSD     LINUX
>time mysql rt3 all.out (full import)              106m    123m
>first full content search                         35.92   45.29
>Second content search after first full content
>   search                                          24.66   26.14
>Subject search                                    0.48    0.35
>4 different select scripts run at the same time
>   on different tables                             62      59
>create index Subject_IDX on Tickets (Sub          8.29    7.12
>
>If this is what you measured, the results look fairly competitive.
>Thanks for performing this real-world test and posting this info.


As I was saying to gnn offlist, you can look at these numbers all sorts of ways

e.g. you could look at these results and say

"so the import takes a long time, big deal. you only do that 
once!  Look at the index, its 15% faster!"

or

"the index creation is barely a second... a second, big whoop.  Look 
at the time to import, thats a big difference!"


or... You are just measuring the quality of the driver or the speed 
of the disks etc etc... or "try 40 queries at the same time and you 
will see a BIG difference".... well, when my requirements get to 40 
queries at once, great, I will focus testing on those requirements!

For me, this exercise was very much driven by my own need to 
understand if there really would be a big (I mean > 50% in *my* world 
tests) to make it worth while to use LINUX that is (for me anyways) 
more cumbersome and less stable to administer (read more $$$$ and 
headache).  I have to be practical about the boxes I manage.  Put 
another way, if the same gear under FreeBSD is 10% slower than LINUX, 
the one time cost for me to buy hardware that will be 10% faster is 
laughably trivial.... The on going cost (let alone GNU vs BSD for our 
OEM products we put out) is WAY more.


The next set of comparisons I want to run is in our spam 
scanners.  The boxes which operate in round robin make heavy use of 
mysql, DNS lookups, perl, net io etc etc.  That one I am quite 
interested in and will post the results in a week or two.

         ---Mike