Thanks for the reply JD and Mike, Sorry I should have provided more information.
What we are looking at is more on the openstack architecture. Suppose our setup is such that we have a controller, having two compute nodes, let these be, Compute node A (with hypervisor like KVM) and Compute Node B (with hypervisor like ESXi). And when requests are placed to the controller, it always chooses A and only when A runs out of resources uses B. (this is possible to using filters and weights on nova-scheduler but we are still exploring on how to start using it.... ) Do advice if there is any other better way to achieving this. On Mon, Nov 10, 2014 at 8:13 PM, Mike Spreitzer <mspre...@us.ibm.com> wrote: > Venu Murthy <ve...@thoughtworks.com> wrote on 11/06/2014 08:34:37 PM: > > > In terms of the architecture for OpenStack setup, if we have a > > controller node, Compute Node A and Compute Node B. > > In the same availability zone. > > > > I do understand that this autoscaling is more for the instances. But > > how can we load balance or start spawning our instances on Compute > > Node B, if the Compute Node A's utilisation has reached a threshold.... > > OpenStack gives a HOT author no access to the relationship between host > and guest. You can autoscale based only on metrics about the guest or > things relatable to the guest in HOT. Would it work for you to autoscale > based on load or performance of your guests? > >
_______________________________________________ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
