Massive Data Analysis Based MapReduce Structure on Hadoop System

Abstract

Massive Data analysis is becoming increasingly prominent in a variety of application fields ranging from scientific studies to business researches. In this paper, we demonstrate the necessity and possibility of using MapReduce [1] module on Hadoop System [2]. Furthermore, we conducted MapReduce module to implement Clustering Algorithms [3] on our Hadoop System [4] and improved the efficiency of the Clustering Algorithms sharply. We showed how to design parallel clustering algorithms based on Hadoop System. Experiments by different size of data demonstrate that our purposed clustering algorithms have good performance on speed-up, scale-up and size-up. So, it is suitable for big data mining and analysis.