So far, we have described how a primary object on a node can discover remote objects it is interested in and maintain up-to-date replicas of those objects. An important decision remaining is where to place primary objects in the system.
