A hybrid genetic-based task scheduling algorithm for cost-efficient workflow execution in heterogeneous cloud computing environment

DOI: https://doi.org/10.1007/s10586-024-04468-6

2024-05-12

Cluster Computing

Abstract:Many businesses utilize cost-efficient cloud services to execute their industrial and scientific workflow applications. Business continuity is a very important issue for both cloud users and providers. To have reliable workflow execution, the engagement of reliable resources is a challenging job that can supply business continuity. In addition, the lowest execution time and monetary cost are two cost features that adhere users to providers. In this regard, the task scheduling algorithm is very prominent in reducing costs in favor of users and providers. To address the issue, a system framework and different cost-type models are suggested. Then, the task scheduling issue is formulated into an optimization problem with an overall cost reduction viewpoint. To solve this NP-Hard problem, a hybrid genetic algorithm ( HGA ) is presented for reliable and cost-efficient task scheduling of workflow execution in a heterogeneous cloud computing environment. The proposed HGA has different phases chief amongst them is to apply new crossover and mutation operators for global search, and a Walking around procedure to enhance the quality of local search solutions. It makes a good balance between local and global searches in a huge search space that leads to efficient results. To verify the proposed hybrid algorithm, it has been tested in different twelve scenarios with variable communication to computation ratios datasets. The results of extensive simulations in twelve datasets scenarios prove that HGA significantly dominates other state-of-the-art in terms of three prominent cost metrics, namely, makespan , monetary cost, and failure cost in the amount of 14.10%, 18.70%, and 42.30% cost reduction respectively.

computer science, information systems, theory & methods