An availability zone is a physically separated set of resources (in our case in different data centers) but where they still exist within the same OpenStack region and can be managed that way.
In our primary region europe-se-1, there are two availability zones to use to spread data and workload across multiple sites. This is selectable in terms of volumes, instances (compute) and network services and is selected on a per resource basis when creating said resources.
Some things are important to keep in mind when working with several different accessibility zones:
- It is not possible to link a volume from one accessibility zone to an instance in another AZ (“cross attach”). The instance must be created in the same availability zone as the volume.
- It is possible to extend a network over both zones (ports can be created on the same network over both zones) but it does not provide a complete HA network, as router pairs still exist in the selected availability zone they are created in. To achieve complete separation, separate networks should be set up with separate public interfaces.
You can read more about this in OpenStack’s public documentation: