Query optimization in the ADDS multidatabase system

A description is given of the query optimization strategies that have been implemented in ADDS, a multidatabase system that allows new applications to retrieve data from preexisting heterogeneous databases without disrupting existing applications. ADDS' query optimizer uses only high-payoff but low-cost algorithms for reducing data retrieval and transmission costs. These techniques include composite schema replication, doing relational operations at the data sites, join and semijoin optimization, doing restrictions as early as possible, and common subquery elimination. Performance figures showing dramatic reductions in network traffic and substantial improvements in query execution speed indicate the soundness of the approach. The choice, organization, and sequencing of, the chosen query optimization techniques are the key contributions.<
>