using live/ volume migration

Link: http://docs.openstack.org/grizzly/ope...

OpenStack does have live migration (if implemented correctly) from the command line, and it moves the data with it to a new hypervisor host, also as mentioned earlier, evacuate is also an option.

$ nova live-migration --block-migrate <instance_id> <hypervisor_host_name>
$ nova live-migration --block-migrate c9c5fde1-3b9b-4695-8b35-75317128100e mhv3.demo1.mc.metacloud.in

Many vendor do have HA within the Management or control tier as an option so your OpenStack services (nova, glance, cinder ...) all are HA, but not your VM's.

OpenStack's design approach is to mirror EC2 experience, so in that world the HA is build into the Application tier and not the platform. That is not to say you could not have HA agents within OpenStack between to VM's, you can, its just not built in to OpenStack natively.

Hi..

I was just about to access the URL http://docs.openstack.org/grizzly/openstack-compute/admin/content/live-migration-usage.html (http://docs.openstack.org/grizzly/ope...), but seems it is no longer available.

About this Live Migration, is it automated as well just like vMotion? I am currently working with one cloud service provider that is using openstack / kvm as the cloud platform. And seeking an availability features that provide automated evacute whenever a node / host are experiencing failure. So the instances inside the fail node able to evacute to the available nodes. This should be only happen within the single region (within the stack or server pool).

Appreciate your helpful effort in advance.

I would answer in the following way: vmotion is supported (called live migration). You can find some references in the internet on how to achieve that, main requirement - shared storage (although, non-shared also promosed to be supported, but I didn't try it personally). For example, this article - http://blog.zhaw.ch/icclab/setting-up... - describes live migration.

HA - as far as I know, there is no option in GUI for that, but there's a command "evacuate", which can be ecexuted via API or directly from CLI.

Unfortunately, don't have any info for storage vmotion - but, my opinion - as soon as you have distributed storage (like gluster or ceph) - you might not need this feature.