Mapresceduler: A Technique for Improving Scheduling in Hadoop

Abstract

In Hadoop all scheduling and Resource allocation decisions are made on a task and node slot level for both the map and reduce phases. I.e., not all tasks of a job may be scheduled at once Hadoop by default uses FCFS scheduling algorithm which shows that this leads to inefficient allocations and the need for social scheduling hence we present a scheduler on real multi-node complex server on realistic data sets .we enhancing the FCFS scheduling algorithms with sized based priority for efficient and effective allocation of resources. all task of job may be scheduled at a time due to file spliting. Our scheduler gives gaurentee of fairness of jobs.it has less execution time than the traditional FCFS scheduling algorithm due to spilting of dataset or cluster of fixed size. Sized based priority is given to the scheduler map reduce architecture is used for processing of data set. We also increase the performances of the maper.