Many mobile environments require optimistic replication for improved performance and reliability. Peer-to-peer replication
strategies provide advantages over traditional client-server models by enabling any-to-any communication. These advantages
are especially useful in mobile environments, when communicating with close peers can be cheaper than communicating with a
distant server. However, most peer solutions require that all replicas store the entire replication unit. Such strategies
are inefficient and expensive, forcing users to store unneeded data and to spend scarce resources maintaining consistency
on that data.
We have developed a set of algorithms and controls that implement selective replication, the ability to independently replicate individual portions of the large replication unit. We present a description of the
algorithms and their implementation, as well as a performance analysis. We argue that these methods permit the practical use
of peer optimistic replication.
This work was sponsored by the Advanced Research Projects Agency under contract DABT63-94-C-0080.