[Xen-users] Re: [Xen-devel] SLA Monitoring of XEN DomU's

[Tim Post <echo@xxxxxxxxxxxx>]

On Mon, 2010-01-04 at 14:19 +0000, Ahmad Hassan wrote:
> Dear All,
> 
> We have set up a XEN testbed on AMD64 based cluster nodes. I would
> like to know the possible ways to monitor the activities of DomU's
> from Dom0. Is there any transparent way to monitor cpu usage and
> network traffic of DomU's at Dom0 level in order to ensure that SLA
> gurantees to each DomU.
> 
> You help is much appreciated.
> 
> Thanks.
> 
> Regards, Hassan

Network traffic is easily monitored by making sure that you assign
static vifnames to guests. I.e. if your guest is named "vps-123", then
your network config might look like this:

vif = ['bridge = br0, mac = xx:xx:xx:..., vifname = vps-123.0', 
       'bridge = br1, mac = xx:xx:xx:..., vifname = vps-123.1' ]

... where vps-123.0 is the device to monitor (on dom-0) for eth0 in that
guest, vps-123.1 the one to monitor for eth1, etc .. etc.

After that, snmp takes care of logging if using something like cacti.

It becomes a little complicated if you tend to migrate guests from node
to node, but not unmanageable.

I'm not quite sure what kind of SLA you are backing with this. I can
produce graphs that tell people they have 100% of all 16 cores in a node
rather easily. Since its impossible for the customer to independently
verify, its hardly usable in a SLA claim. Unless I missed something?

Cheers,
--Tim

> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-devel


_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users