Batched search of index sequential files

Abstract

The performance of batched search when applied to index sequential files is studied. Analysis provides exact formulae for the cost of searching when batching is applied as a function of the magnitude of the query and time. The analysis takes into account (a) accesses to the overflow area only, and (b) accesses to both primary and overflow area, as well as whether the records of the query are (a) distinct, or (b) nondistinct. Batched search is compared with the simple on-line search and a proposal is stated concerning the reorganization of index sequential files.