Is one buffered and the other not? Perhaps if you mount the
4500
filesystem with -forcedirectio to see what that does ?
-----Original Message-----
From: Peter Santos [mailto:psantos cheetahmail.com]
Sent: Monday, April 16, 2007 1:25 PM
To: suse-oraclesuse.com
Subject: [suse-oracle] Suse 10 vs SUN E4500 - ORION io tool
does not
match iostat output
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Folks,
I'm wondering if someone can help me understand the output
of oracle's
ORION I/O calibration tool.
I'm comparing a SuSE 10 Dell 6800 /IBM DS4400 (4 dual core
Xeon 2.60GHz)
vs Solaris 5.8 E4500/IBM DS4400 (12 x 400Mhz
cpus).
Both the devices used are a RAID 10 configuration of 24 x
36GB 15000 rpm
drives.
I've made sure the controllers are setup the same, and
tested only small
random reads (8k reads).
The problem I have is that the IOPS rate for the Solaris
machine is much
greater than Suse, but when I compare
iostat outputs it shows that the SuSe throughput is far
better... which
is what I expect, since the E4500 is a 10yr+
machine with slower bus speeds.
I'm trying to understand why the results are conflicting,
and/or if this
is just a reporting issue by Orion and/or
iostat differences between Sun/SuSe. You will notice below
that ORION
reports the IOPS for the Sun test at
29,462 vs 2,539, but the iostat output shows differently.
Does anyone have an idea of why the ORION IOPS does not
match up with
the iostat output on the Sun test?
- -peter
iostat on linux
===============
Device: r/s rsec/s rkB/s avgrq-sz avgqu-sz await
svctm %util
sde 2519.20 40307.20 20153.60 16.00 55.85 22.10
0.40 99.80
sde 2527.07 40433.17 20216.58 16.00 56.87 22.52
0.40 99.86
sde 2535.10 40561.60 20280.80 16.00 56.92 22.47
0.39 99.96
sde 2534.70 40555.20 20277.60 16.00 56.92 22.37
0.39 99.96
sde 2534.80 40556.80 20278.40 16.00 56.93 22.51
0.39 99.96
sde 2539.16 40626.57 20313.29 16.00 59.87 23.49
0.39 99.86
sde 2523.50 40376.00 20188.00 16.00 56.93 22.45
0.40 99.96
sde 2508.10 40129.60 20064.80 16.00 58.72 23.47
0.40 99.80
sde 2517.90 40286.40 20143.20 16.00 59.92 23.79
0.40 99.96
sde 2528.10 40449.60 20224.80 16.00 59.93 23.71
0.40 99.96
ORION report: Maximum Small IOPS=2539 Small=48
iostat on Sun
==============
r/s kr/s wait actv asvc_t %w %b device
559.4 4475.2 0.0 0.2 0.4 0 22 sd225
575.4 4603.2 0.0 0.3 0.4 0 23 sd225
568.7 4549.5 0.0 0.2 0.4 0 22 sd225
578.9 4631.2 0.0 0.3 0.4 0 23 sd225
528.9 4226.5 0.0 0.2 0.4 0 21 sd225
537.9 4303.2 0.0 0.2 0.4 0 21 sd225
533.4 4267.2 0.0 0.2 0.4 0 21 sd225
546.5 4372.0 0.0 0.2 0.4 0 21 sd225
ORION report: Maximum Small IOPS=29462
Small=48
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFGI9uaoyy5QBCjoT0RAnXjAKCLhZ0o/4NAlov/nPhCzvg+NZWy6gCb
Bhbh
osEImLKFOdDyheaESgM19UU=
=YHwV
-----END PGP SIGNATURE-----
--
To unsubscribe, email: suse-oracle-unsubscribesuse.com
For additional commands, email: suse-oracle-helpsuse.com
Please see http://www.suse.com/oracl
e/ before posting
--
To unsubscribe, email: suse-oracle-unsubscribesuse.com
For additional commands, email: suse-oracle-helpsuse.com
Please see http://www.suse.com/oracl
e/ before posting
|