一种多重约束下确保成功率的云工作流调度方法  

A CLOUD WORKFLOW SCHEDULING METHOD ENSURING SUCCESS RATE UNDER MULTIPLE CONSTRAINTS

作　　者：张艮山 刘旭宁 Zhang Genshan;Liu Xuning(Department of Information Technology,Shijiazhuang University,Shijiazhuang 050035,Hebei,China)

机构地区：[1]石家庄学院信息技术系,河北石家庄050035

出　　处：《计算机应用与软件》2021年第2期278-284,317,共8页Computer Applications and Software

基　　金：河北省教育科学研究“十二五”规划课题(13100063)。

摘　　要：为提高多重约束下的调度成功率,提出一种满足期限和预算双重约束的云工作流调度算法。将可行工作流调度方案求解分解为工作流结构分层、预算分配、期限分配、任务选择和实例选择。工作流结构分层将所有工作流任务划分层次形成包任务,以提高并行执行程度;预算分配对整体预算在层次间进行分割;期限分配将全局期限在不同层次间分割;任务选择基于任务最早开始时间确定优先级,得到任务调度次序;实例选择根据时间和代价均衡因子,获取任务执行最佳实例。仿真结果证明,该算法在调度成功率、同步优化工作流执行时间与执行代价上相较对比算法更好。In order to improve the scheduling success rate of cloud workflow under multiple constraints,a workflow scheduling algorithm meeting dual constraints of the deadline and budget is proposed.Our algorithm divides the feasible workflow scheduling scheme into:workflow structure leveling,budget distribution,deadline distribution,task selection and instance selection.Workflow structure leveling is to partition all workflow tasks into several levels that generates bag of tasks,which can improve the degree of parallel execution.Budget distribution is to allocate the user-defined budget to each defined level.Deadline distribution is to divide the user-defined deadline among different levels.Task selection is to set the priority for each task based on earliest start time and obtain the task scheduling sequence.Instance selection is to get the optimal objective instance for executing tasks according to the trade-off factor of time and cost.The experimental results show that our algorithm perform better than the baseline algorithms on the scheduling success rate and synchronous optimization of the execution time and execution cost.

关 键 词：云计算 工作流调度 期限分配 预算分配 实例选择 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]