Buffer management for high performance database systems

This paper presents the development of a buffer algorithm named RESBAL, which exploits parallelism in order to offer high performance for query execution in relational database systems. The algorithm aims to provide both efficient and predictive data buffering by exploiting the use of prior knowledge of query reference behaviour. Designed to offer a high level of flexibility, RESBAL employs a multiple buffering strategy both on page fetch level and page replacement level. The evaluation of RESBAL has been carried out in a parallel database system environment based on a transputer architecture. The results of this evaluation allow comparisons to be made between different buffer algorithms, and demonstrate the feasibility and effectiveness of the RESBAL algorithm. It is hoped that this work will provide some useful input to research on developing high performance database systems. Their new uses such as data mining and data warehousing are presenting the research community with interesting and important challenges.