(PAMRS H-PSFP) Performance Analysis of Map Reduce Scheduler’s with H-PFSP

Abstract

Map Reduce offers a data parallel programming, functional model and computing framework that is being used for processing of big data.
Data processing are developed using open-source Hadoop implementation of map reduce. Scheduler is main module of Map Reduce which
runs on master server by performing centralized control of Map Reduce cluster. We analyse FIFO Scheduling enhancing it with Hybrid
Parallel pessimistic Fair Schedule Protocol (H-PFSP) by using a statistical technique called EWMA (exponential weighted moving average)
and conclude that H-PFSP scheduler can integrate it in fair scheduler and hence every job can be scheduled by H-PFSP in every pool. During
comparison of H-PFSP and Fair scheduler, function to analyze that Fair scheduler has poor performance than H-PFSP in most cases The most
prominent feature of H-PFSP is that it can close each job before Fair scheduler does. Contrasting Fair scheduler, with moderately precise job
progress estimation H-PFSP can progress the per-job performance of Map Reduce systems.