Virtual Machine Task Classification Using Support Vector Machine and Improved MFO Based Task Scheduling

The processing of data in Big data computing necessitates a significant number of CPU cycles and network bandwidth. Dataflow is a huge data processing programming model that comprises of jobs structured in a graph structure. Scheduling these jobs is one of the most active study fields, with the primary goal of allocating tasks based on available resources. It is critical to efficiently schedule jobs in a way that minimizes task completion time and maximizes resource utilization. In recent years, many research works on task scheduling problems in cloud computing and various heuristic approaches have been evolved which the thesis focuses upon. Most of these efforts are focused on improving the performance such as minimizing the makespan and efficient utilization of cloud resources that benefits the cloud users and the providers. In this paper, Big Data analysis processing in cloud environment for efficient dynamic task scheduling by using different techniques as machine learning classifier and optimization approach. In machine learning classifier as Support Vector Machine (SVM) for classification of virtual machine task classification. In this classifier can classify the incoming request efficiently and reduce the makespan and execution time. Further, we used moth flame optimization technique for allocating the classified task from SVM classifier. In this proposed system have classification of virtual machine (VM) task and evaluate the resource allocation decision making procedures. In this experiment, we carried out by using cloud simulation environment to evaluate and analysis the proposed method. In this proposed scheme can efficiently reduce the makespan time and load balancing to improve the better VM Classification. And aslo we compare the proposed moth flame optimization (MFO) with min-max algorithm and particle swarm optimization to compare the performance respectively.