File organization and addressing

Abstract

The principal approaches to random-access file organization and addressing are reviewed in this paper. The review is general, in the sense that it is relatively independent of specific equipment. I n the case of a number of unsettled questions, the author's evaluations of alternatives are included. The relation between sorting and random-access file addressing is clarified by viewing both as belonging to a common class of ordering operations. Basic considerations of both sequential and randomaccess approaches, arithmetical key-to-address transformation methods with their overflow problems, and table lookup methods are discussed. Results of an experimental analysis of key transformation techniques are presented.

Applications of combinatorial designs in computer science

“…The essential feature of the bucket subsets is that each query subset be contained in at least one bucket subset. Now we are in a position to introduce combinatorial filing schemes as introduced in Bucholz [1963] and developed by Abraham et al [1968] and Ray-Chaudhuri [1968]. We need one further definition: A t-couering is a set system (X, 39) in which every t-subset appears in at least one block in H LetA = (al,.…”

A Practical Approach to Selecting Record Access Paths

“…If a block index is itself long, then the indexing technique may be reapplied to construct a multilevel or hierarchic block tndex. 4 A hierarchic block index is a preferred record-search technique for many practical design problems; it is reasonably fast, it is an inexpensive byproduct of sequential file processing, and it is well supported by computer vendors. Few secondary memory accesses are needed to retrieve a record via a hierarchic block index.…”

“…While the speed of block index retrieval is sufficient for most database applications, there are important information systems (such as airline reservations systems) whose high intensity of record retrieval (50 to 100 transactions per second [46]) demands even faster access times. For applications where direct addressing is impractical, addressing techniques known as randomizing or hashing algorithms have been developed [4,25]. In hashing, a given record identifier ("key") is transformed into the address of the memory cell ("bucket") used for the storage and retrieval of that record.…”

Identifier Search Mechanisms: A Survey and Generalized Model

“…Considerations dealing with the efficient construction of transformations are discussed by Buchholz [7], Knuth [20], Lum et al [22], Olson [26], and Roche [29]. Such considerations are not treated in detail here for two reasons:…”