Data Storage Hierarchy Systems for Data Base Computers

Abstract

The need for efficient storage and processing of very large databases to support decision-making coupled with advances in computer hardware and software technology have made research and development of specialized architectures for database management a very attractive and important area. The INFOPLEX data base computer proposed by Madnick applies the theory of hierarchical decomposition to obtain a specialized architecture for database management with substantial improvements in performance and reliability over conventional architectures. The storage subsystem of INFOPLEX is realized using a data storage hierarchy. A data storage hierarchy is a storage subsystem designed specifically for managing the storage and retrieval of very large databases using storage devices with different cost/performance characteristics arranged in a hierarchy. It makes use of locality of data references to realize a low cost storage subsystem with very large capacity and small access time. As part of the INFOPLEX research effort, this thesis is focused on the study of high performance, highly reliable data storage hierarchy systems. Concepts of the INFOPLEX data base computer are refined and new concepts of data storage hierarchy systems are developed. A preliminary design of a general structure for the INFOPLEX data storage hierarchy system is proposed. Theories of data storage hierarchy systems are developed. Madnick's model of a generalized storage hierarchy is extended and formalized for data storage hierarchy systems. The Least Recently Used (LRU) algorithm is extended to incorporate the read-through strategy and page overflow strategies to obtain four classes of data movement algorithms. These algorithms are formally defined. Important performance and reliability properties of data storage hierarchy systems that make use of these algorithms are identified and analyzed.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 01, 1979
Accession Number
ADA607279

Entities

People

  • Chat-yu Lam

Organizations

  • Massachusetts Institute of Technology

Tags

Communities of Interest

  • Energy and Power Technologies
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Access Time
  • Algorithms
  • Application Software
  • Computer Programming
  • Computer Programs
  • Computers
  • Content Addressable Memory
  • Data Analysis
  • Data Storage Systems
  • Database Management Systems
  • Databases
  • Hierarchies
  • Information Processing
  • Information Systems
  • Mass Storage
  • Operating Systems
  • Reliability

Fields of Study

  • Computer science

Readers

  • Parallel and Distributed Computing.