disk i/o

Hubbard, David dhubbard at dino.hostasaurus.com
Tue Jun 18 19:19:00 CDT 2002


Sounds like your testing tool did not use a large
enough data set.  It would be physically impossible
for four 10k rpm drives to output 381 MB/sec, that's
nearly 100 MB/sec per drive, it must have been
reading from cache.  Try some of the utilities
on this page:

http://www.acnc.com/04_02.html

bonnie++ is particularly useful, but use a data
set of at least twice your RAM to make sure cache
doesn't skew the results.

Besides that though, the 35 MB/sec is the normal
unacceptable rate you'll commonly see with newer
Dell (AMI and Adaptec) raid controllers.  I have
heard that the big daddy Perc3/QC is a different
beast altogether though, posting very impressive
results.  It's quite a bit more money...

David

> -----Original Message-----
> From: Duncan Morgan [mailto:dmorgan at gmi-mr.com] 
> Sent: Tuesday, June 18, 2002 7:36 PM
> To: linux-poweredge at dell.com
> Subject: disk i/o
> 
> 
> Hello,
> 
> can anyone tell me what are good numbers to shoot for when 
> tuning disks i/o?. Specs for the server are this:
> 
> - Dell 6450
> - quad 700 MHz Xeon processors
> - 2 GB RAM
> - 4 x 36 GB Ultra 3 SCSI drives
> - AMI Megaraid
> - Linux Red Hat 7.2 w/ 2.4.9-34smp redhat kernel
> 
> I did some benchmarking with a tool called 'hotio' and got 
> write throughput of average 35 MBytes/sec and read throughput 
> of average 381 MBytes /sec. These seem like reasonable 
> numbers compared to other servers I have benchmarked bit I'm 
> not sure. Is there a more accurate testing tool I can use? 
> Are these numbers in fact good?
> 
> Thanks,
> Duncan
> 
> _______________________________________________
> Linux-PowerEdge mailing list
> Linux-PowerEdge at dell.com 
> http://lists.us.dell.com/mailman/listinfo/linu> x-poweredge
> 
> Please read the FAQ at 
> http://lists.us.dell.com/faq or search the list archives at 
http://lists.us.dell.com/htdig/




More information about the Linux-PowerEdge mailing list