Research On Cloud Workflow Scheduling Model And Multi-task Solving Strategy

With the continuous growth and development of data scale in various industries,a large number of complex cloud workflow tasks have been generated in the actual life and production process.However,most of the existing cloud workflow scheduling schemes optimize a single workflow task independently,and do not make reasonable use of the commonality or complementary knowledge between similar tasks.Moreover,the scheduling factors considered in the construction of the scheduling model are not comprehensive enough.In order to deal with multiple cloud workflow scheduling problems synchronously and efficiently,this paper builds a multi-objective and many-objective scheduling model according to the workflow structure characteristics and scheduling requirements.And use the correlation between similar tasks to design corresponding multi-task scheduling scenarios and efficient evolutionary multitask optimization algorithm to solve the model.The specific work content is as follows:(1)Most of the existing cloud workflow scheduling models focus on time or cost,and the considerations are not comprehensive enough,such as ignoring the needs of cloud service providers in the scheduling process.Therefore,this paper builds a multi-objective cloud workflow scheduling model(time,cost,energy consumption)by analyzing the structural characteristics of cloud workflow and its target requirements that are focused on during the scheduling process.Secondly,in order to simultaneously and effectively process multiple similar cloud workflow scheduling tasks,this paper regards different workflow scheduling problems as multiple different tasks,and constructs a multi-task scenario with the same goal and different problem scales.Finally,in the multi-task scenario designed above,an evolutionary multi-task optimization algorithm based on elite selection is designed to solve the model,and the experimental results show the superiority of the algorithm.(2)In the actual cloud workflow scheduling environment,tasks and cloud resources are constantly enriched,and the demands of users and cloud service providers are also increasing.Therefore,in order to fully describe the performance requirements of the entire system,this paper further comprehensively analyzes the scheduling factors that need to be considered in the workflow scheduling environment from the perspectives of both users and cloud service providers,and extends the multi-objective cloud workflow scheduling model to many-objective cloud workflow scheduling model(time,cost,Qos,energy consumption,resource utilization and throughput).Secondly,in order to improve the forward knowledge transfer between similar workflow scheduling tasks,so as to better use the associated knowledge existing between tasks to solve the model,we designed a many-objective evolutionary multi-task optimization algorithm combined with reference vector assistance and integrated index strategies.Finally,the algorithm is used to solve the model in the designed multi-task cloud workflow scheduling scenario,and the experimental results verify the rationality of the model and the effectiveness of the algorithm.(3)This paper uses the above-mentioned cloud workflow scheduling model and optimization algorithm as the back-end technical support,and designs a cloud workflow scheduling prototype system based on Matlab.First,by discussing and analyzing the needs of users,the functional modules including login and registration module,task submission module,algorithm parameter setting module and scheduling result display module are constructed.Secondly,the front-end interface corresponding to the relevant functional modules is designed by using the GUI control of Matlab.Finally,through the callback function,the front-end interface is correlated with the scheduling model and algorithm designed in this paper,so as to realize the display of the results of the cloud workflow scheduling scheme.