HISA-LB: A Novel Load Balancing Approach for Improvising Harmony Memory in Cloud Environment

Abstract

In a cloud context, task scheduling (TS) is an NP-hard combinational optimization issue since the number of tasks rises, and their durations vary rapidly. The mappings between resources and tasks are challenging to develop. As a result, we essential a capable task scheduling strategy that can better handle the work and address the NP-hard problem. Several researchers have concentrated on heuristic, meta-heuristic, and hybrid scheduling algorithms to address this problem. This work provides a unique load balancing strategy based on two optimization algorithms to address the problem of uneven load distribution at virtual machines (VMs) via proper task scheduling. This approach consists of music inspired harmony search algorithm (HSA) and simulated annealing (SA) algorithm, named HISA-LB (load balancing) approach. This approach works on HMCR and PAR values in which probability is checked. Based on the probability threshold, it calculates the best objective function by choosing either Harmony search or simulated annealing to distribute tasks across available cloud resources. This is accomplished by maintaining accurate information among the data center's tasks and resources. The CloudSim simulator is applied to implement the proposed algorithm for load balancing. The simulations have been done by considering two scenarios 1) 3 VMs and 10 to 50 cloudlets, 2) 5 VMs and 10 to 50 cloudlets. Both scenarios have been experimented with and considered only the five best performance results. The simulation outcomes indicate that the proposed HISA-LB strategy outperforms other existing LBMPSO approaches in decreasing Makespan and increasing resource usage. In addition, throughput is another parameter for comparison and achieved the highest throughput than the LBMPSO approach.