Scientific workflow on distributed computing environment has huge data transfer and storage, Aiming at this problem, on the environment of an executing machine which have limited storage resource, this paper disassembles the tasks of workflow into data jobs and computing jobs. The model of new workflow which include data jobs and computing jobs is proposed. It inserts data clear job when those data are no longer needed and transfer jobs when data need to transfer from one executing node to others, the method of creating the new workflow is provided. Schedule algorithm for scientific workflow based on the consider of available storage is proposed. Experimental results indicate that the schedule algorithm consumes less response time and raises the efficiency.%针对数据密集型科学工作流需要大量的数据传送和数据存储的问题,在执行节点可用存储资源受限的情况下,构造计算作业与数据作业分离的工作流模型,没计数据与计算分离后的工作流牛成算法,增加数据转送作业、数据清除作业、数据作业及其依赖关系.给出资源受限情况下数据密集工作流的预估存储调度算法,并对其进行系统评价,取得了较好的效果.
展开▼
