On Tue, Mar 02, 1999 at 03:20:28PM -0500, David Mansfield wrote:
> 
> I am concerned that these stats for bonnie are in fact pretty poor for the
> hardware involved.  Could someone running SW RAID5 on similar hardware
> comment on your bonnie findings?
...
> with 6 UW Quantum Viking II 4.5GB drives, three on each channel,
> configured (by VA Research) into a Raid5 volume.  The system is a dual PII
> 450Mhz, running linux kernel 2.2.2 (a RedHat 5.1 + upgrades system), with
> 256MB ram.  Note, the bonnie file size is 500MB (about double phys. ram).

Kernel 2.2.1, latest RAID patch, Dual PII, 350 MHz, 256 MB RAM.
RAID-5 Configuration with seven 6 GB partitions.
One partition on IBM UltraStar 9ES,  this disk on one aic7990 controller
Six partitions on Quantum Fireball TM, three disks on each NCR53c810 (Asus SC200) 
controller

> Here is the bonnie output (slightly edited to fit well in 74 columns):
> 
> File './Bonnie.4392', size: 524288000
> Writing with putc()...done
> Rewriting...done
> Writing intelligently...done
> Reading with getc()...done
> Reading intelligently...done
> Seeker 1...Seeker 2...Seeker 3...start 'em...done...done...done...
>        -------Sequential Output-------- ---Sequential Input-- --Random--
>        -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
>    MB  K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec %CPU
>    500  3742 46.8  4623  6.7  2604  7.9  6614 72.9 13314 14.6 183.4 3.2

File './Bonnie.11838', size: 524288000
Writing with putc()...done
Rewriting...done
Writing intelligently...done
Reading with getc()...done
Reading intelligently...done
Seeker 1...Seeker 2...Seeker 3...start 'em...done...done...done...
              -------Sequential Output-------- ---Sequential Input-- --Random--
              -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine    MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec %CPU
          500  5825 91.5 18720 42.3  6353 19.4  4522 65.0 14453 16.2 150.0  2.7


A note though:  This machine was running a very lightly loaded SQL server,
MRTG for roughly 50 ports, www and user file-serving while I did this benchmark.
The bonnie was niced -5, but still this benchmark may or may not have been
influenced by other processes.
Also it would be preferable to run with a 1G file, but I can't do that with 2.2.1.

> 
> vmstat shows blocks in and out maxing out at 9000, but typically in th
> 4500-5500 range at any given time.

vmstat maxed out with bo=19309 but usually was in the 3K-4K range.  All this
during ``intelligent'' write.

fs blocksize is 4K, raid chunksize is 32K.

> If these are in fact poor numbers, what is wrong with my system.  I feel
> like I should be seeing numbers 5 times these.

I think your numbers look a little poor. Except ofcourse for the CPU usage.

...wonder if ext2fs options (stride) makes a difference...

-- 
................................................................
: [EMAIL PROTECTED]  : And I see the elder races,         :
:.........................: putrid forms of man                :
:   Jakob Østergaard      : See him rise and claim the earth,  :
:        OZ9ABN           : his downfall is at hand.           :
:.........................:............{Konkhra}...............:

Reply via email to