文献类型：期刊文献

英文题名：Cost-effective approaches for deadline-constrained workflow scheduling in clouds

作者：Li, Zengpeng[1];Yu, Huiqun[1,2];Fan, Guisheng[1]

机构：[1]East China Univ Sci & Technol, Sch Informat Sci & Engn, Shanghai 200237, Peoples R China;[2]Shanghai Key Lab Comp Software Evaluating & Testi, Shanghai 201112, Peoples R China

年份：2023

卷号：79

期号：7

起止页码：7484

外文期刊名：JOURNAL OF SUPERCOMPUTING

收录：;EI(收录号：20224913199908);WOS:【SCI-EXPANDED(收录号:WOS:000890149800001)】；

基金：This work was supported by the National Natural Science Foundation of China (No. 61772200), Natural Science Foundation of Shanghai (No. 21ZR1416300), and Capacity Building Project of Local Universities Science and Technology Commission of Shanghai Municipality (No. 22010504100).

语种：英文

外文关键词：Cloud computing; Workflow scheduling; Whale optimization; Deadline

摘要：Nowadays, heterogeneous cloud resources are charged by cloud providers according to the pay-as-you-go pricing model. To execute workflow applications in clouds under deadline constraints, cloud resources have to be utilized appropriately and judiciously, challenging traditional workflow scheduling algorithms, which are either inapplicable to the cloud environment or fail to fully exploit the features of scheduling problem for cost optimization. In this paper, we propose a heuristic algorithm CSDW and a meta-heuristic algorithm N-WOA to minimize the execution cost of the given workflow subject to the deadline constraint in clouds. CSDW first assigns the sub-deadline to each task based on the modified probabilistic upward rank, and then tasks are sorted and mapped to appropriate instances, finally instance-type upgrading and downgrading method is adopted to further accelerate workflow execution and reduce the total cost, respectively. N-WOA employs whale optimization algorithm for deadline-constrained cost optimization by refining the task ordering step in CSDW. By simulation experiments on scientific workflows with existing algorithms, the results demonstrate the capability of the proposed algorithms in meeting the deadlines and reducing the execution costs, CSDW is highly competitive and N-WOA achieves the best performance in all cases.