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/
Home Products Support Community News


[Xen-users] scalability issue.

To: "xen-users@xxxxxxxxxxxxxxxxxxx" <xen-users@xxxxxxxxxxxxxxxxxxx>
Subject: [Xen-users] scalability issue.
From: "dkar4700@xxxxxxxxxxxxxxxxx" <dkar4700@xxxxxxxxxxxxxxxxx>
Date: Fri, 29 Oct 2010 00:56:18 +0000
Accept-language: en-AU, en-US
Delivery-date: Wed, 10 Nov 2010 10:23:34 -0800
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
List-help: <mailto:xen-users-request@lists.xensource.com?subject=help>
List-id: Xen user discussion <xen-users.lists.xensource.com>
List-post: <mailto:xen-users@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=unsubscribe>
Sender: xen-users-bounces@xxxxxxxxxxxxxxxxxxx
Thread-index: Act3BBhFpYtzYXRVS0mW/gDwRzd1nw==
Thread-topic: scalability issue.
Hi All,

I'm finding this results a bit surprising , could some please explain whether this is the expected behavior.
I'm running a single threaded application ( calculating pi for many decimal points ). Each VM has 1 vcpu with 128MB of ram. The physical box has quad-core processor ( HT disabled ). When I increase the number of VMs the time it takes to calculate the result taking longer ( Performance goes down ). I could clearly see that each VM will just use 1 CPU when it is running the tests.

Why there is a such a huge jump in time although there are enough physical cores available ( at least upto 3 Vms) ?
Please let me know if any more information is necessary.

VM1 – 31.85

Number of VMs.
VM1 – 35.98
VM2 – 36.11

Number of Vms
VM1 - 43.57
VM2 - 42.55
VM3  -42.48

Number of VMs
VM1 - 51.093
VM2 - 49.953
VM3 - 50.141
VM3 - 49.844

XEN version
dpkg -l | grep -i xen

ii  libc6-xen                         2.11.2-6                    Embedded GNU C Library: Shared libraries [Xen version]
ii  libxenstore3.0                    4.0.1-1                     Xenstore communications library for Xen
ii  linux-image-2.6.32-5-xen-686      2.6.32-23                   Linux 2.6.32 for modern PCs, Xen dom0 support
ii  xen-hypervisor-4.0-amd64          4.0.1-1                     The Xen Hypervisor on AMD64
ii  xen-tools                         4.2-1                       Tools to manage Xen virtual servers
ii  xen-utils-4.0                     4.0.1-1                     XEN administrative tools
ii  xen-utils-common                  4.0.0-1                     XEN administrative tools - common files
ii  xenstore-utils                    4.0.1-1                     Xenstore utilities for Xen

All guest machines 'uname -a'
Linux vm1 2.6.32-5-xen-686 #1 SMP Thu Aug 12 18:09:47 UTC 2010 i686 GNU/Linux

xm dmesg| grep -i credit
(XEN) Using scheduler: SMP Credit Scheduler (credit)

xm sched-credit
Name                                ID Weight  Cap
Domain-0                             0    256    0
vm1                                 57    256  100
vm2                                 58    256  100

Dom0 is pinned to CPU0.
Name                                ID  VCPU   CPU State   Time(s) CPU Affinity
Domain-0                             0     0     0   r--    6907.9 0
Domain-0                             0     1     -   --p      74.9 any cpu
Domain-0                             0     2     -   --p      92.4 any cpu
Domain-0                             0     3     -   --p      62.2 any cpu
vm1                                 57     0     2   -b-    6614.7 any cpu
vm2                                 58     0     1   -b-    5797.1 any cpu

Xen-users mailing list
<Prev in Thread] Current Thread [Next in Thread>
  • [Xen-users] scalability issue., dkar4700@xxxxxxxxxxxxxxxxx <=