Fair budget constrained workflow scheduling approach for heterogeneous clouds

The phenomenal advancement of technology paved the way for the execution of complex scientific applications. The emergence of the cloud provides a distributed heterogeneous environment for the execution of large and complex workflows. Due to the dynamic and heterogeneous nature of the cloud, scheduling workflows become a challenging problem. Mapping and assignment of heterogeneous instances for each task while minimizing execution time and cost is a NP-complete problem. For efficient scheduling, it is required to consider various QoS parameters such as time, cost, security, and reliability. Among these, computation time and cost are the two notable parameters. In order to preserve the functionalities of these two parameters in heterogeneous cloud environments, in this paper, a fair budget-constrained workflow scheduling algorithm (FBCWS) is proposed. The novelty of the proposed algorithm is to minimize the makespan while satisfying budget constraints and a fair means of schedule for every task. FBCWS also provides a mechanism to save budget by adjusting the cost-time efficient factor of the minimization problem. The inclusion of a cost-time efficient factor in the algorithm provides flexibility to minimize the makespan or save budget. In order to validate the effectiveness of the proposed approach, several real scientific workflows are simulated, and experimental results are compared with other existing approaches, namely; Heterogeneous Budget Constrained Scheduling (HBCS), Minimizing Schedule Length using Budget Level (MSBL) and Pareto Optimal Scheduling Heuristic (POSH) algorithms. Experimental results prove that the proposed algorithm behaves outstandingly for compute-intensive workflows such as Epigenomic and Sipht. Also, FBCWS outperforms the existing HBCS in most of the cases. Moreover, FBCWS proves to be more time-efficient than POSH and more cost-efficient than MSBL. The effectiveness of the proposed algorithm is illustrated through the popular ANOVA test.