R. Schaarschmidt and J. Lufter
Extremely large and ever growing databases lead to severe problems in storage, database administration, and performance. Archiving in database systems is introduced as a new approach to manage these problems. It is based on the observation that not all data is used permanently and with the same intensity. The objective of archiving is to relieve the database of some load by moving rarely used data from the database into an archive. An archive is thus characterized by infrequent, mostly reading access, time requirements are usually not critical. Data in database and archive is separated logically and physically, archived data can be placed on tertiary storage. In this paper we present an architecture for a seamless integration of archives and archiving functionality into database systems. This approach allows a homogeneous access to data in both, database and archive; furthermore, relationships between them can still be maintained. Time-related aspects play an important role in the development of functionality and a suitable infrastructure for archiving. We have formalized the functional integration of archiving into relational database systems by a compatible extension to SQL/92 named Archive SQL (ASQL). Following an introduction to database system integrated archiving and a discussion of some design issues, we focus on concepts for the definition of archives. We show that the impact of changes to a databases schema on associated archives can be handled by archive schema versioning. Our intentions of archiving are illustrated by examples using ASQL.
archive, database, schema versioning
@techreport{Sch98a:own,
author = {R. Schaarschmidt and J. Lufter},
title = {An Architecture for Archives in Database Systems},
type = {Forschungsergebnisse der Fakult{\"a}t f{\"u}r Mathematik
und Informatik},
number = {Math/Inf/98/18},
institution = {Institut f{\"u}r Informatik},
address = {Friedrich-Schiller-Universit{\"a}t Jena},
month = jun,
year = {1998}
}