On Wed, Feb 09, 2005 at 11:29:22PM -0800, Andrew P. Lentvorski, Jr. wrote: > > On Feb 9, 2005, at 10:36 PM, Wade Curry wrote: > > >On Wed, Feb 09, 2005 at 05:25:02PM -0800, Andrew P. Lentvorski, Jr. > >wrote: > >>Rattle the disks a little on that Pea Sea and the system performance > >>turns to crud. > > > >I confess to not being _entirely_ sure what you mean here.. may be > >terminology, or may be lack of familiarity with the hardware. > > Run multiple Linux kernel compiles on a standard x86 box. The > performance degrades much faster than one would expect given the > processor and disk utilization. > > The problem is the I/O subsystem. Even today, one can see a > significant difference between the performance of an x86 box with an > ATA subsystem rather than a SCSI subsystem. And that is a very weak > configuration relative to the balancing that goes on when companies > make big iron. > > Big iron is basically all about transactions per second, not seconds > per transaction. >
I'm starting to get it. The truck can't accelerate like the sports car, but man, can it haul stuff! -- Lan Barnes [EMAIL PROTECTED] Linux Guy, SCM Specialist 858-354-0616 -- [email protected] http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
