Abstract: The present invention provides a “data centric” approach to updating databases on computer systems of an intermittently connected database system. In this approach, the storage and processing complexity of the database server are de-coupled from the number of clients to be supported, thereby, improving the scalability of the server. Instead of focusing on the data required by individual clients, this method tracks changes to data subsets pertaining to groups of clients. Thus, the server need only track and record changes to these subsets of data, instead of tracking changes for individual clients. Clients download the subsets, which contain the data relevant to the group, and merges data or deletes superfluous data from the subsets. Since the server is tracking a limited set of data subsets related to groups of clients instead of the actual number of the clients, the overall scalability of the system is increased.