Research On The Execution Optimization Of Scientific Workflow In Cloud

Scientific Workflow is a new application paradigm, which can improve the automation of scientific experiments. To a large extent, it can save the resources and human costs. Tasks of Scientific Workflows are either data or computing sensitive. Therefore, traditional computing environment is hard to meet the resource needs of Scientific Workflows. Therefore, for Scientific Workflows, to find an ideal computational environment has become more and more important. Cloud Computing provides a good chance for solving this problem.Cloud Computing environments have the characteristics such as infinitely computing and storage capability providing, economy, scalability and Pay-as-you-go. Therefore, Cloud Computing environment is a good choice for Scientific Workflows. Obviously, improving the execution efficiency of a Scientific Workflow is also an effective way to its running cost.Focusing on the higher ratio of processor utilization and lower execution cost of a scientific workflow in Cloud, a policy of execution optimization is proposed. The policy will work well by bringing forward the completion time of the workflow and the reduction of the idle time of the processor. Firstly, to bring forward the Earliest Finish Time of a scientific workflow, its tasks are aggregated into several clusters based on task replications. So, the key tasks will be scheduled earlier. Moreover, the task clusters are aggregated once again when it is possible to facilitate the spare time of processors. Secondly, based on backward merging of task clusters, free time spans between workflow task executions are exploited to optimize the whole execution time of the workflow. Furthermore,in order to make full use of the slack time during task executions and improve the overall efficiency of the workflow, some workflow tasks are permitted to violate the local time constraints.Experiments and simulations show that the policy can improve the parallelism of workflow tasks, bring forward the Earliest Finish Time of the whole workflow, improve the utilization ratio of processors and lower the cost of workflow execution.