Distributed database systems have long been a topic of interest in the database research community. Existing designs focus on two principles; make the distribution transparent to users and provide a rich declarative query language with strict semantic guarantees. As a result, they have modest targets for network scalability with none of these systems being deployed on much more than a handful of distributed sites.

The Internet community has recently become interested in distributed query processing. Not surprisingly, they approach this problem from a very different angle than the traditional database literature. The fundamental goal of Internet systems is to operate at very large scale (thousands if not millions of nodes). To achieve this degrees of scale, these system sacrifice transparency and/or flexibility.

This thesis develops a system called PIER (which stands for "Peer-to-Peer Information Exchange and Retrieval") which provides a rich query language that provides location transparency and scalability with relaxed semantics. We explore the architecture of PIER, develop techniques for query processing (with specific focus on aggregation and join operations), and finally examine an optimization problem with multiple simultaneous aggregation queries.