Improved swarm search algorithm for scheduling budget-constrained workflows in the cloud
Huifang Li,Danjing Wang,Guanghao Xu,Yan Yuan,Yuanqing Xia
DOI: https://doi.org/10.1007/s00500-022-06782-w
IF: 3.732
2022-02-02
Soft Computing
Abstract:Basic science is becoming more computing intensive with the incremental need for large-scale computing and storage resources. Cloud computing provides great potentials for hosting and executing scientific applications, which can be represented as workflows for automatic execution and run time provisioning. However, workflow scheduling is highly challenging under dynamic cloud environments since certain runtime QoS. In this paper, we propose an improved swarm search algorithm, i.e., an Owl Search Algorithm embedded with a newly designed Mutation strategy (OSAM) for scheduling workflows with makespan minimized under budget constraints. A population update mechanism, where each particle is updated in term of the impact of current best solution, is modified for an Owl Search Algorithm (OSA) to address discrete sequential optimization problems. We then adjust a step parameter βdocumentclass[12pt]{minimal}usepackage{amsmath}usepackage{wasysym}usepackage{amsfonts}usepackage{amssymb}usepackage{amsbsy}usepackage{mathrsfs}usepackage{upgreek}setlength{oddsidemargin}{-69pt}egin{document}$$eta $$end{document} such that it decreases adaptively with the number of iterations to enhance OSAM’s convergence speed. To further increase the diversity of population and enlarge OSAM’s global search ability, we embed a mutation strategy into OSA’s evolutionary process for better balancing between exploitation and exploration. A series of experiments are conducted to verify the proposed algorithm by comparing with benchmarking algorithms over well-known scientific workflows with different types and sizes through WorkflowSim. Experimental results show that in almost all the cases, the proposed OSAM outperforms the existing algorithms in the solution quality and constraint satisfiability, i.e., it can find near-optimal solutions that meet the tight and loose budget constraints within an acceptable time interval, especially in addressing large-scale applications, e.g., its RSR values are at least 0.10 greater than other algorithms in these cases. Besides, a nonparametric statistical hypothesis test, i.e., Wilcoxon signed-rank test, is applied to the resulting solutions, and the testing results demonstrate that our OSAM is significantly different from its peers.
computer science, artificial intelligence, interdisciplinary applications