Date of Award

1-1-2012

Document Type

Campus Access Thesis

Department

Computer Science and Engineering

First Advisor

Csilla Farkas

Abstract

Geographic Information Systems (GIS) were designed to model the world. However, with the growth of data and increasing sophistication of analy- sis and processing techniques, the traditional methods of performing GIS processing on desktops are insucient. This thesis addresses this issue by studying the applicability of Column-store databases for GIS. This is accomplished by implementing two sets of spatial database systems, one using row store and the other using column store. These implementations are then evaluated for eciency of data retrieval. Oracle is used as the row-store database system, while MonetDb is used as the column-store database system. Our hypothesis is that storing GIS in a Column-store format will increase the performance of data retrieval. To establish this hypothesis we 1. Calculate the cost of query processing for both column-store and row- store database systems. Analyze the result in context of Geographic Information System. 2. Design an GIS architecture using column-store database management system. 3. Implement the corresponding platform, including column-store database management system and Geographic Information Systems. 4. Perform experiments to evaluate the eciency of data retrieval using traditional row-store DBMS and column-stored DBMS on the context of GIS. 5. Analyze empirical results.

Share

COinS