A framework for designing update objects to improve server scalability in intermittently synchronized databases

We consider the class of mobile computing applications in which clients naturally operate on shared data without a connection to the server. When appropriate, a server connection is made in order to exchange updates. The update server computes and retransmits these updates on a clientby-client basis; consequently, the complexity of these operations is on the order of the number of clients, limiting scalability. We recently proposed exploiting overlap in client data subscriptions by organizing updates to these subscriptions into groups (with less overlap) instead of on a clientby-client basis. By grouping updates, update processing is performed only once per group, irrespective of the number of clients. Additionally, we may gain bandwidth scalability by employing broadcast delivery since, unlike in the case of the per-client approach, multiple clients may be interested in a group's updates. In this work, we model the operations of such database systems and introduce a framework for evaluation of group design. Since such fragmentation algorithms are computationally intractable, a heuristic graphbased group generation algorithm is speci ed. Performance results executed on a prototype developed using commercially available software are presented.