Before we go ahead with adding Trove to our production install, there are a couple of things we should do:
Decide where to install the Trove components
Take backups
In a production install of OpenStack, you would have a variety of nodes that would be functioning as:
Controller node
Compute node
Storage node
Network node
There might be a single node for each of these roles or multiples of these nodes depending on how large a footprint OpenStack is managing.
The best place you would install the services of Trove would be on the controller node, except of course the guest agent, which will be installed on the instance that Nova spins up.
If there are multiple controller nodes (in the case of an HA install of OpenStack), you will have to install the Trove components on both of them.