SCALABILITY AND FAULT TOLERANCE OF MAPREDUCE FOR SPATIAL DATA

Abstract

This paper discusses the processing of spatial data on MapReduce – Hadoop platform. The Hadoop is known for its efficiency, ease of implementation and fault tolerance. The earlier existing technologies for parallel processing, such as, Grid Computing also provided fault tolerance, but its implementation using the replica management is not easy as compared to the internal replica management in Hadoop. This paper implements a Hadoop-GIS cluster for a spatial dataset and a spatial query is implemented in the MapReduce. The experimental run demonstrates the effectiveness of Hadoop cluster in terms of the scalability and fault tolerance.