Cost and makespan aware workflow scheduling in IaaS clouds using hybrid spider monkey optimization

• Published in: Simulation modelling practice and theory 2021-07, Vol.110, p.102328, Article 102328
• Main Authors: Rizvi, Naela, Dharavath, Ramesh, Edla, Damodar Reddy
• Format: Article
• Language: English
• Subjects: Cloud computing; Deadline; Makespan; Penalty function; Workflow scheduling
• ISSN: 1569-190X; 1878-1462
• DOI: 10.1016/j.simpat.2021.102328

•A hybrid algorithm based SMO and BDSD for scheduling workflows has been proposed.•A novel approach is designed for the initialization and mapping of tasks to the VMs.•A penalty function is introduced when the cloud provider fails to meet the QoS constraints.•The formulation of a penalty based fitness function is introduced.•A statistical t-test is carried out to validate the performance of the proposed HSMO. The researcher's predilection towards the concerned infinite resources and the dynamic provisioning on rental premises encourages the scheduling of complex scientific applications in the cloud. The scheduling of workflows in the cloud is constrained to QoS parameters. Many heuristic and meta-heuristic algorithms are widely investigated for the QoS constrained workflow scheduling problem. However, it is still an open area of research, as most of the existing techniques concentrate on minimization of either cost or time and ignores the optimization of multiple QoS constraints simultaneously. To address this problem, in this paper, a Hybrid Spider Monkey Optimization (HSMO) algorithm has been proposed. The proposed algorithm optimizes the makespan and the cost while satisfying the budget and deadline constraints. The proposed algorithm is the hybridization of recently developed SMO and the other popular heuristic BDSD algorithm. BDSD is a budget and deadline constrained algorithm, which guides HSMO in generating a feasible schedule. Moreover, the proposed strategy involves the penalty function to restrict selecting those solutions that fail to satisfy the QoS constraints. Experimental results demonstrate the effectiveness of HSMO over existing ABC, Bi-Criteria PSO, and BDSD algorithms.