Flexibility of block-length for magnetic files

Abstract

A constraint imposed on systems designers by over-restrictive software is that the unit of transfer between secondary and primary storage is the block (physical record) or, at best, that every read operation must start at the beginning of a block. This paper demonstrates that many current computers provide channel programming facilities which may be used to overcome this constraint, enabling a distinction to be made between ‘device block’ and ‘input block’. This distinction leads to greater efficiency of file handling (in terms of both transfer time and secondary storage utilisation) and to increased flexibility in file design. An algorithm is developed for selection of optimum input block length for a given file and a given input buffer length.