My specific use case is to deliver high availability across disparate data centers. Being able to run an application and load balance between physical locations provides immediate availability in the event of an issue with one data center. For example, if there is a telecom issue which takes down access to data center 1, through a global load balancing solution, traffic continues to route to the other data center. Having an active / active postgres instance provides this capability. If the database lives on one side and we perform log shipping, I have to be engaged to bring the “warm stdby” instance online.