Some considerations on different default settings
As stated above, comparing two Linux distributions is not an easy task. One key point to remember is that different distributions can have different default setting – and there is often no right setting, but many reasonable ones. Realizing (and accepting) this is a key factor in correctly comparing two different operating systems.
However there a some settings that are too much different to simply leave it to default value, because in this case they can skew the comparison – for example, they can disable some important security checks in order to give artificially higher performances.
In light of these considerations, while I tried to leave much things to their default values, in order to produce a meaningful comparison I changed some settings so that the results can be compared. More specifically, I manually settled on the following fixed settings:
-
use LVM to manage the disk/partitions
-
use a very simple partition scheme, with only a /boot partition and a / (root) partition
-
run two benchmark sessions for each distribution, one with EXT3 and one with EXT4 for root partition (Red Hat use EXT4 by default, while Debian use EXT3)
-
manually enable write barriers for the EXT3 test (note that Red Hat enable barrier by default, while Debian not).