Data Block and Tuple Identification Using Master Index

Abstract

Relational databases are still very often used as a data storage, even for the sensor oriented data. Each data tuple is logically stored in the table referenced by relationships between individual tables. From the physical point of view, data are stored in the data files delimited by the tablespaces. Files are block-oriented. When retrieving data, particular blocks must be identified and transferred into the memory for the evaluation and processing. This paper deals with storage principles and proposes own methods for effective data block location and identification if no suitable index for the query is present in the system. Thanks to that, the performance of the whole system is optimized, and the processing time and costs are significantly lowered. The proposed solution is based on the master index, which points just to the blocks with relevant data. Thus, no sequential block scanning is necessary for consuming many system resources. The paper analyzes the impact of block size, which can have a significant impact on sensor oriented data, as well.