Analysis and Experimental Study of HDFS Performance

Abstract

In the age of big data, the amount of data that people generate and use on a daily basis has far exceeded the storage and processing capabilities of a single computer system. That motivates the use of distributed big data storage and processing system such as Hadoop. It provides a reliable, horizontallyscalable, fault-tolerant and efficient service, based on the Hadoop Distributed File System (HDFS) and MapReduce. The purpose of this research is to experimentally determine whether (and to what extent) the network communication speed, the file replication factor, the files’ sizes and their number, and the location of the HDFS client influence the performance of the HDFS read/write operations.