There should be an obvious theme to these questions. As we suggested in the beginning of this recipe, we will need a lot of resources to perform this task. Why? Consider this diagram for a company with a somewhat large PostgreSQL cluster:
If every one of these nodes is in use while we are using the version 10 cluster, we must determine how many version 11 nodes we actually need.
The easiest way to proceed is to simply reproduce every node, as in the preceding diagram. In some cases, such as with an extremely busy online transactional processing (OLTP) cluster, it might be the only way. By migrating onto a cluster with the same ...
