METHOD, MODELING AND EVALUATION OF ACCELERATION OF COMBINING ARRAYS WITH INTEGRATED VALUES INDEX

Abstract

The model of the array data is considered, such as raster indexes and inverted lists that can be used to encode the items of the elements in the data set, reference information about the format of floating numbers and compression, as well as the architecture of graphic processors. The array stores ordered multidimensional data. Unlike relational databases, where the cortas are stored disordered, the values in the array are ordered and organized by its size, which contributes to a much faster search speed.

This work developed a method, modeling and evaluating the acceleration of the association of arrays with an integrated index of values. The association of indexed arrays required the development of a new method of evaluating the acceleration of the association of arrays with an integrated index of values, as the dimension and submission of data in their elements differ significantly in various applied problems. In general, the structure of unification of indexed arrays corresponds to the general steps of combining simple arrays. The key differences are that the indexed arrays are not only organized by dimensional coordinates, but also organized with different bits that retain attributes. The grainness of the treatment, thus, is no longer segments, but subsets in each set. In order to effectively combine arrays with different fragmentation schemes, different types of association were implemented and modeling.

The areas of further research are to improve the architecture of the system that stores arrays with integrated index support and which will automatically combine them. Experiments, modeling and evaluation of acceleration of the association of arrays with an integrated index of values confirm the possibility of practical implementation of the developed method.