Resolving Result Set Contention in Heterogeneous Library Information Systems

Federated databases are one of the challenges for database system builders. In an e ort to create these systems, current implementations require every member of the federation to agree to a particular standard. We examine query optimization under the Z39.50 standard for heterogeneous library information systems, looking speci cally at the problem of managing temporary relations within a library system limiting the number of these relations held by a single user. We propose a strategy which uses knowledge of the current user query and an approximation of a set of future user queries in order to choose which temporary relation is the least likely to be useful in answer the current and future queries. Simulation results indicate that this strategy, called Look-Ahead, performs better than strategies based on operating system concepts. Work supported in part by a summer sabbatical from Baylor University