A Comparative Study of Very Large Data Bases

This monograph presents a comparison of methods for organizing very large amounts of stored data called a very large data base to facilitate fast retrieval of desired information on direct access storage devices. In a very large data base involving retrieval and updating, the major factor of immediate concern is the average number of accesses to the direct access storage device to complete a request. The average number of accesses to store and retrieve items on a direct access storage device for hashing methods using chaining with separate lists and linear probing is presented. A new algorithm and performance measures for chaining with coalescing lists is presented. New performance measures are presented for storing and retrieving with a. binary search tree and a trie stored on a direct access storage device. Algorithms are presented to perform retrieval, insertion, deletion and the Inverted file generation operations for an inverted file. New performance measures are presented for an inverted file. The methods are developed using a component concept. A hybrid method involving components is used for the linked files. All methods are analyzed, along with their data structures, to show their effect on the average number of accesses to the direct access storage device while processing a request. Finally, a comparison criterion is developed and each method is compared.

!!!No Mirrors below, please! Follow Rules!