Comparitive Study of Hadoop over Containers and Hadoop Over Virtual Machine

摘要

Purpose - The purpose of this paper is to elaborate the concept of Containerization which enables usage of Hadoop Infrastructure on a container which can allow clusters to run rapidly fast and accessible manner and it can be used for installing, processing and monitoring/analysis of Big Data under various configurations. Design/methodology/approach - This paper is a research-based comparative study of the VM services which deploy HADOOP framework upon them using a simple virtual machine and using container services from Docker. The analysis is done using a tool known as terasort, which has a program called Teragen which generates the random data on the HDFS cluster. Then it sorts the data using terasort services. The result is analysed upon the time taken by the Virtual Machine and Docker Container. Findings - The result of the research proves that the time taken during benchmark testing of the cluster on the container was much lesser than that of Virtual machine, and result also proves the better efficiency of the cluster on the container. The results shown in graphical analysis clearly indicates the difference in time taken by both the services. Docker Container implementation clearly takes lesser time when processing any task into the cluster. Originality/value - The originality in this research is the valuation and comparision of Docker Container with Virtual Machine to deploy HADOOP framework for the processing of data. The evaluation and time-based analysis is done solely by the authors using the tool known as Terasort. The value obtained during research can help in optimising the Data Centres which runs Virtual Machine to deploy their services.