Big data analytics is the process of analyzing the huge volume of data and other useful information. While accessing large volume of data over the network, resource optimization and load balancing is the major concern to reduce the risk level. Therefore, an efficient technique is required to distribute the workload uniformly across all the nodes with minimum resource utilization for improving the data access. Lottery Scheduling based Softmax Regression Resource Optimization (LS-SRRO) technique is introduced for improving the load balancing with minimal resource utilization such as energy and bandwidth. The scheduler uses lottery scheduling technique for load balancing process. In lottery scheduling, each file is provided with number of lottery tickets. The number of lottery tickets is given based on file transfer rate and job deadline. Based on file transfer rate and job deadline, the scheduler identifies emergency and normal constrained files for providing the lottery tickets. The scheduler picks a random ticket to choose the subsequent process. The incoming files with more number of tickets have relatively higher priority of selection. Then the MapReduce system based softmax regression analysis is used to handle multiple classes (i.e., cardiologist, diabetologist and oncologist etc) for distributing the files with efficient resource optimization. After identifying the class, the scheduler distributes file to that particular class with minimal resources utilization. LS-SRRO technique balances the work load among multiple classes with minimal resource utilization. Experimental evaluation is carried out with different factors such as load balancing efficiency, scheduling time, and bandwidth utilization rate and energy consumption with respect to number of files. The results showed that the LS-SRRO technique is better in case of load balancing efficiency, resource utilization rate, scheduling time and memory consumption. Based on the results observations, LS-SRRO technique is more efficient than the other existing methods for improving the load balancing.
Volume 11 | 07-Special Issue
Pages: 1133-1149