一种云环境下的科学工作流均衡调度算法  

A Scientific Workflow Trade-off Scheduling Algorithm in Cloud Environment

作　　者：吴文兵[1] WU Wen-bing(College of Information,Fuzhou University of International Studies and Trade,Fuzhou 350202,China)

机构地区：[1]福州外语外贸学院信息系,福建福州350202

出　　处：《微电子学与计算机》2018年第10期121-126,共6页Microelectronics & Computer

基　　金：国家自然科学基金项目(61562063)

摘　　要：为了优化云环境中预算约束下的科学工作流调度问题,提出一种工作流均衡调度算法BDWTS.算法以满足工作流预算约束并同步优化执行代价和执行时间为目标,将工作流调度划分为四个阶段:工作流分级、预算分割、任务选择和实例选择.工作流分级通过自顶向下的方式对所有工作流任务进行分级,预算分割中设计了六种用户预算在不同工作流分级上的子划分方法,任务选择通过最早开始时间原则选择优先的调度任务,实例选择则综合考虑时间因子和代价因子选择最优执行资源.通过仿真实验,证明了算法可以在预算约束下得到降低工作流执行时间和代价的均衡调度方案.For optimizing scientific workflow schedulimg with budget constraint m ciuuu, t, Workflow Trade-off Scheduling （BDWTS） algorithm is presented. With synchronous optimization of makespan and cost meeting budget constraint as the objective, the algorithm is divided into four stages during scheduling workflow： workflow leveling, budget division, task selection and instance selection. Workflow leveling partitions the workfiow tasks into different levels by the top-down method, budget division allocates the user-defined budget to each defined levels by six strategies, task selection selects the scheduled-priority task by the earliest start time principle, instance selection selects the optimal execution resource by overall considering the time factor and cost factor. It is proved that our algorithm can obtain the trade-of~ scheduling scheme of the execution time and execution cost under meeting budget constraint by simulation experimental results.

关 键 词：云环境 工作流调度 均衡调度 预算分配 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]