Evaluation of Parallel Multi-Dimensional Indexing System for Big Data Analysis

Abstract

Parallel multi-dimensional indexing system was proposed for big data analysis. The insertion and the retrieval performances were evaluated through randomly distributed data with three distribution schemes: area expansion, overlap, and proximity. This paper evaluates insertion and retrieval performances of this system through skewed data with the round robin method as well as three distribution schemes. It is experimentally clarified that the round robin method has good retrieval performance to randomly distributed data. It is also experimentally clarified that the proximity method has good retrieval performance especially for skewed data.