FC4 kernel performance

[Peter Backlund]

ons 2005-06-22 klockan 09:01 -0400 skrev Paul A Houle:
> Peter Backlund wrote:
> 
> >
> >The debugging should be taken out of the release kernels, yes. But RAID
> >support is modularized, and I'd imagine that the SELinux overhead is
> >incredibly small when it's turned off. The same goes for exec-shield. I
> >don't really know what the other 19 security features are
> >though :-)      
> >
> >  
> >
>     I run RAID 1 on my Linux machines at home.  Perhaps Seagate is 
> benefitting from my bad experiences with 1999-2001 vintage Maxtor 
> drives,  but it's cheap protection from problems that can waste a lot of 
> time.  (As compared to $700 tape backup drive that needs $250 worth of 
> tapes to back up a $100 drive and takes an hour to do it.)
> 
>     RAID 1 speeds up the boot process considerably and helps with random 
> access read I/O,  which gives a noticable performance boost for ordinary 
> desktop tasks;  I know PC enthusiasts who swear by RAID 0 for pure 
> performance aspects...  One of these days I'm going to build a RAID 0+1 
> machine for database work.

Hmm...note that I didn't suggest that RAID was not useful for desktop
users. I merely stated that RAID support is built as modules, and
therefore does not have any effect on kernel performance when they are
not loaded.

The reply makes more sense to the original posting, but maybe that was
the intention?    

/Peter