If you have the appropriate permissions, you can select the specific host where the instance will be launched. This is done using the --availability_zone zone:host arguments to the nova boot command. For example: $ nova boot --image <uuid> --flavor m1.tiny --key_name test --availability-zone nova:server2 Starting with the Grizzly release, you can specify which roles are permitted to boot an instance to a specific host with the create:forced_host setting within policy.json on the desired roles. By default, only the admin role has this setting enabled. You can view the list of valid compute hosts by using the nova hypervisor-list command, for example:$ nova hypervisor-list +----+---------------------+ | ID | Hypervisor hostname | +----+---------------------+ | 1 | server2 | | 2 | server3 | | 3 | server4 | +----+---------------------+ The --availability_zone zone:host flag replaced the --force_hosts scheduler hint for specifying a specific host, starting with the Folsom release.