Heterogeneous Fair Resource Allocation and Scheduling for Big Data Streams in Cloud Environments

Abstract

In this paper, Heterogeneous Fair Resource Allocation and Scheduling (HFRAS) for cloud based Big Data Streams, is proposed. In this algorithm, a weight value is determined for the user for each of the requested resource, based on the resource priorities. Then each task is assigned a task priority index (TPI) based on this weight value, task arrival time and expected end time (EET). The requested tasks are divided into various priority queues based on the TPI of the tasks assigned. Then tasks are sorted in the ascending order of TPI and scheduled in which the Dominant Resource Share (DRS) is determined for each user. Experimental results have shown that HFRAS attains lesser execution time, minimum response delay and maximum CPU utilization, when compared to the existing algorithm.