[Steve Bosell]

I was really disappointed to learn of OS X god awful mysql/apache performance. I would have liked to pitch one to my work, but I can not in good conscience now

http://www.anandtech.com/mac/showdoc...i=2436&p=6

I wonder if Apple is doing anything about it

[Millennium]

Given that Apple does not make Apache or MySQL, one wonders what they could do. It's up to the makers of those apps, is it not?

AnandTech's tests were deeply flawed, in any case. They should have tested Darwin/X86 and Linux/PPC in addition to the other OS/platform combinations they tested, so that resders could compare the different operating systems running on the same hardware. As things currently stand, there are too many variables for their numbers to be truly meaningful.

[wadesworld]

Read these two:

http://ridiculousfish.com/blog/?p=17
http://lists.apple.com/archives/darw.../msg00072.html


Additionally, there's some question of what was used to compile MySQL. It sounds as if they may have used stock gcc, rather than Apple's gcc.

In short, I think there's a lot of questions regarding the validity of the tests. It'd be interesting to see the test rerun with the F_FULLFSYNC behavior disabled, as that would be much more equivalent to what Linux does.

Wade

[Steve Bosell]

The tests seemed like a pretty good "real world" simulation, and it is hard to believe F_FULLFSYNC is responsible for such a huge performance hit.