A Comparative Analysis of MapReduce Scheduling Algorithms for Hadoop

Paper Topic :

Data Base Management System

Author Name :

Hiral M. Patel

Abstract :

Today’s Digital era causes escalation of datasets. These datasets are termed as “Big Data” due to its massive amount of volume, variety and velocity and is stored in distributed file system architecture. Hadoop is framework that supports Hadoop Distributed File System (HDFS)for storing and MapReduce for processing of large data sets in a distributed computing environment. Task assignment is possible by schedulers. Schedulers guarantee the fair allocation of resources among users. When a job is submitted by user, it will be placed into a job queue. A job will be then divided into tasks and distributed among different nodes. Proper assignment of tasks will reduce job completion time. This can guarantee improved performance of the job. In this paper we study Map Reduce model and evaluated task scheduling algorithms such as FIFO, Fair share, Capacity, Delay, IWRR and MTL of Hadoop platform.

Download Article