A caveat for bonnie: on an ideal system, with inifinitely-fast I/O,
the %CPU numbers should all be 100%. Anything less means the CPU
is being held up waiting for I/O.
One can see it the other way around: for CPU usage higher than, say 85%,
your bonnie numbers are probably CPU b
how would one use these statistics to diagnose how to improve raid
performance??
hdparm -t -T /dev/md1
/dev/md1:
Timing buffer-cache reads:64MB in 1.36 seconds =47.06 MB/sec
Timing buffered disk reads:32 MB in 4.01 seconds = 7.98MB/sec
Seeker 1...Seeker 2...Seeker 3...start 'em...done...
On Wed, 2 Dec 1998, Mark Lord wrote:
> Here are a pair of 7200rpm Seagate IDE drives as RAID0, using UDMA0:
>
> ---Sequential Output ---Sequential Input-- --Random--
> -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
> MachineMB K/sec
And lo, Mark Lord saith unto me:
>
> Michael Shields wrote:
> >
> > Here is a bonnie (tested with -s of 3x physical RAM) on my new home
> > system:
> >
> > ---Sequential Output ---Sequential Input-- --Random--
> > -Per Char- --Block--- -Rewrite-- -Per Cha
Michael Shields wrote:
..
> I have a Celeron 333, which should not be dramatically slower. Are
> you sure these numbers are right -- 85 MB/s? How much RAM is in that
ooops. Yeah, ignore the 85MB values.
Here it is again, using "bonnie -s 256" on my 128MB P2-400 w/raid0 (2 IDE):
Seeker 1...Se
