QoS Aware Task Scheduling Using Hybrid Genetic Algorithm in Cloud Computing

Abstract

Simultaneous improvement of Quality of Service (QoS) parameters is a challenge in the Cloud Computing (CC) environment if the overall QoS provided is not sufficient for end-users. In this paper, the Smart Message Passing Interface Approach (SMPIA) is combined with Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) algorithms for CC application. PSO-SMPIA and GA-SMPIA algorithms are presented for task scheduling and resource allocation with the aim of reducing makespan and total execution time, and increasing resource utilization in CC. The main contribution of this paper is to calculate the maximum cost for each transaction flow which has not been addressed in previous studies. This new multipurpose function includes flow load amount, load amount on makespan, capacity of Virtual Machines (VMs) and execution speed parameter. Telecommunication deals and tenders are categorized based on the type of flow. The transaction flow allocation matrix to the VMs is obtained with appropriate values. According to the selected matrix, the transaction flow is sent to the selected VMs. Also, we analyze the effect of makespan and total execution time on resource utilization. The results show that PSO-SMPIA performs better than Optimized-SMPIA (O-SMPIA), Fuzzy SMPIA (FSMPIA), SMPIA and GA-SMPIA in terms of average resource utilization. But FSMPIA and O-SMPIA perform more efficiently than other algorithms in terms of improving makespan and total execution time, respectively. The total execution time and makespan are reduced significantly compared to other algorithms which in turn increases the QoS delivered to end-users. Thus, GA-SMPIA outperforms other algorithms through simultaneous improvement of the total execution time and resource utilization besides further improvement (minimizing) of the makespan.