云计算中效率与代价均衡的科学工作流调度

摘要

For solving the cost optimization problem of scientific workflow scheduling in cloud environment,a workflow scheduling algorithm ECT-CWS based on trade-off between efficiency and cost was presented.With an objective of meeting deadline constraint and minimizing the execution cost,the solving process of scheduling scheme was divided into four states including workflow levelling,proportional deadline distribution to partition the user deadline over the levels,task selection to prioritise ready tasks and selecting the best virtual machine,which obtained the trade-off scheduling scheme.Through simulation experiments,three different types of workflow structures were constructed as the testing data source,and the performance of the proposed algorithm was compared with that of workflow scheduling algorithms IC-PCP and GAIN.The results show that ECT-CWS can not only get higher tasks execution success rate,but also obtain less execution cost under meeting deadline constraint,which can achieve the trade-off between scheduling efficiency and execution cost.%为解决云环境中科学工作流调度的代价优化问题,提出一种基于效率与代价均衡的工作流调度算法ECT-CWS.以满足截止时间约束同时最小化执行代价为目标,将调度方案求解划分为4个阶段:工作流分级、截止时间在分级上正比例再分布、就绪任务调度选择及最优虚拟机选择,得到均衡调度方案.通过仿真实验,构建3种不同类型的工作流结构作为测试数据源,与工作流调度算法IC-PCP和GAIN进行性能比较,实验结果表明,该算法拥有更高的任务执行成功率,满足截止时间约束,能够得到更小的执行代价,实现调度效率与执行代价的均衡.