Predictive Technique Of Task Scheduling For BigData In Cloud

摘要

In the current era, Big data utilizes MapReduce strategies in task scheduling, most notably, Apache Hadoop, a programming library and framework that considers the distributed processing of enormous data across clusters of computers using simple programming models. Still, there persist longer wait times with MapReduce technique. Because, the scheduling in Cloud with huge data causes frequent obstructions to effective computing, prompting prolonged makespan, longer waiting time and expenses acquired by the client and the server end. Task Scheduling with a huge measure of information data can cause obstacles to proficient processing. The motive of this work is to accomplish the desired solution to overcome the issues obstructing the effective CPU/task Scheduling of VMs in the Cloud environment with massive data without the use of any complex algorithms. This paper presents a predictive task scheduling approach and introduces PCA (Principal component analysis) and utilizes nine different Machine classifiers and compares the results of the accuracy and time obtained by each ML classifiers with and without the use of PCA. Results are visualized and the percentage variation of comparison is discussed. Experiments are carried out in the Hadoop Environment, using MapReduce the dataset is generated and ML classifiers are executed in Python.