Zoning Multidimensional Access Methods for Analytical Database Applications

In emerging analytical database applications that deal with large sets of multidimensional static data sets, the performance of the query system significantly depends on the performance of its access methods and the underlying disk system. In recent years, hard disks are manufactured with multiple physical zones, where seek times and data transfer rates vary significantly across the zones. However, there is a marked lack of investigation on how to optimize multidimensional access methods given a zoned disk model. Instead, conventional access methods have been developed based on a traditional disk model that comes with many simplifying assumptions such as an average seek-time and a single data transfer rate. The paper proposes a novel optimization technique that can be applied to a variety of multidimensional access method and that can fully utilize zoning characteristics of hard disks. In the proposed approaches, index structures are zoned in such a way that more frequently accessed index pages are stored in a faster disk zone. Then our localized query processing techniques significantly improve the query performance by reducing page retrieval times from the hard disk.