Database Block Management using Master Index

Abstract

A database is formed by a set of data files holding the data. These files are block oriented. Each row can be located by the ROWID address pointing to the data file, data block, and its position inside the block. For processing, block granularity is used for memory loading and evaluation. However, a block is fixed in size, thus, during the Update operations, block fragmentations can be present. Moreover, once the block is associated with the table, it is not commonly deallocated, whereas it is part of the extent, not allocated individually. All these facts have strong importance and impact on the performance of the data retrieval, mostly in the case of sequential block scanning. This paper deals with the Master index extension to locate fragmentations, manage shrinking and identify empty blocks. Thanks to that, database performance can be significantly improved. The study deals with the temporal environment.