WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-devel

[Xen-devel] measuring cpu utilization under i/o workload

To: xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: [Xen-devel] measuring cpu utilization under i/o workload
From: tim benke <tbenke@xxxxxxxxxxxx>
Date: Thu, 22 Jan 2009 23:01:35 -0500
Delivery-date: Thu, 22 Jan 2009 20:02:38 -0800
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
List-help: <mailto:xen-devel-request@lists.xensource.com?subject=help>
List-id: Xen developer discussion <xen-devel.lists.xensource.com>
List-post: <mailto:xen-devel@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=unsubscribe>
Sender: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx
Hello,

I run I/O-bound experiments and I would like to measure CPU utilizations in Dom0 and DomU. Though these experiments were only executed in Dom0 to find out which tool reports consistent numbers.
Xentop reports much higher utilization when compared to mpstat or top in a xen-capable kernel. 
In a kernel without xen mpstat reports higher utilizations but still not comparable to xentop (See attached summary of the data).

My question is: which tool is actually correctly reporting how much the physical CPUs are used?

My test system is:

OpenSuSE 10.3
Xen 3.1
default OpenSuSE kernels version 2.6.22

2 AMD dual-core 2212 HE CPUs at 2 GHz
10 GB RAM
hard drive is a IBM-ESXS ST936701SS 71GB

experiment:
commands executed: 
# flush cache (pagecache, dentries and inodes (see http://linux-mm.org/Drop_Caches))
echo 3 > /proc/sys/vm/drop_caches;
# start measurements, time copying of a file, stop measurements
# measurements were taken at 5 second intervals
time cp random_1000M.bin random2.bin;
# delete file
rm random2.bin

space-aligned in monospace font:
xen domain 0
with xen-kernel: runtime ~= 36-38 seconds
xentop:              0.0,    18.3,   13.8,   13.8,   13.5,   12.2,   12.1,
mpstat-sys:          0.15,   0.15,   0.20,   0.48,   0.16,   0.15,   0.90,
mpstat-user:         0,      0,      0,      0,      0,      0,      0,  
mpstat-iowait:       25.84,  24.84,  24.85,  23.75,  25.75,  24.88,  29.55
top-sys:             0.1,    0.1,    0.1,    0.8,    0.1,    0.4,    0.2,    0.4,
top-user:            0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,

kernel without xen: runtime ~= 33-38 seconds
mpstat-sys:              4.24,   2.5,    2.85,   3.05,   2.8,    3.15,   3.54,
mpstat-user:             0.05,   0,      0,      0,      0.05,   0.05,   0.0,
mpstat-iowait:           20.67,  31.87,  37.86,  42.06,  47.13,  33.05,  25.56
top-sys:        2.0,     4.4,    3.2,    1.7,    2.8,    3.1,    2.9,    2.2,    
top-user:       0.4,     0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0

thanks,
..Tim
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
<Prev in Thread] Current Thread [Next in Thread>
  • [Xen-devel] measuring cpu utilization under i/o workload, tim benke <=