filebench 1.4.8 - file allocation time added to final result ?

[przemolicc-IjDXvh/]

Hi,

we are using filebench 1.4.8 to test some storage and we got some strange results.
My collegue found that filebench add file creation time to the final result.
So the first run gives really bad result (since all the files need to be created)
but following runs give much better results because all needed files are
already created.
Is this expected behaviour or we do some mistake ?

Regards
przemol





























----------------------------------------------------------------------
Zobacz najwiekszy samolot na swiecie!
Kliknij >>> http://link.interia.pl/f238f

[Vasily Tarasov]

This is not true. Filebench does not take creation time into account. What most probably happens
is that after the first run your caches (page/buffer caches) are warm (i.e. all the files are already
in the RAM), so Filebench runs much faster. On Linux you need to run sync and echo 3 > /proc/sys/vm/drop_caches to drop the caches. In this case you won't see performance differences.

Vasily

[przemolicc-IjDXvh/]

Post by Vasily Tarasov
This is not true. Filebench does not take creation time into account. What most probably happens
is that after the first run your caches (page/buffer caches) are warm (i.e. all the files are already
in the RAM), so Filebench runs much faster. On Linux you need to run sync and echo 3 > /proc/sys/vm/drop_caches to drop the caches. In this case you won't see performance differences.

The test was done on Solaris with direct mounted fs.

Regards
Przemyslaw Bak (przemol)
--
http://przemol.blogspot.com/






























----------------------------------------------------------------------
Sprawdz warunki na drogach!
Kliknij >>> http://link.interia.pl/f259f