| The scientific workflow technology, which has the ability to promote the development ofscience and technology, can take full advantage of the cloud environment, for example, overcominggeographical restrictions and saving hardware and software resources. In order to improve usersatisfaction degree, researching scientific workflow scheduling in the cloud environment is of veryimportant significance. The article proposes a new approach, which uses the improved Algorithm：NSGAeda, take into account Clouds elasticity feature.Moreover, the approache considers theoverall completion time and the execution cost together, where the tasks allocation and schedulingproblem has not been converted neither to mono-criterion nor to constraints problem.NSGAeda algorithm combines the fast non-dominated sorting algorithm NSGA II anddistribution estimation algorithm EDA, utilizes the convergence properties of the population toproduce new individuals using genetic operation or probability model. The algorithm takes fulladvantage of the local information and global information of population. It makes the optimalsolution more evenly distributed, and shorten the convergence speed of the population. Theimproved algorithm is applied to the scheduling of scientific workflow in the cloud environment. Itevaluate the algorithm within a simulated CloudSim environment and compared with NSGA IIalgorithm at last. In order to assess the quality of the obtained solutions by the algorithms withexperiments, we define the lower bounds. The simulation results indicate that the algorithm hasexcellent optimization ability.The obtained Pareto solutions are less than2.5times as much as thelower bound.The improved algorithm makes full use of the advantage that the cloud allows users to allocateand release compute resources on-demand and pay only for what they use.The approach shortenthe time and cost to complete the scheduling task, improved resource utilization and meet the users’requirements of the service quality. |
PDF Full Text Request