Campus-wide distributed library search system

The paper discusses the mechanism of accessing multiple libraries geometrically distributed in an academic institute in a uniform manner. Searching of any library item in such a system may amount to blindly querying every single participating database in the system (broadcasting the query). If the information about these databases and their holdings is kept in the form of indices distributed over a group of brokers, the queries can be forwarded to a set of appropriate databases, thus avoiding a broadcast. The proposed distributed library search system is based on the client-server model and consists of a collection of database servers which maintain the multiple library databases in the campus, brokers which hold a variety of distributed indices, and clients. This approach helps minimize the network traffic which rises otherwise due to broadcasting, and reduces the average search time. The design approach here resembles that of a heterogeneous, bottom-up distributed database design.<>