Cloud Workflow Scheduling For Rental Cost Minimization Based On Trust And Deadline Constraints

Workflow applications such as earthquake detection and hurricane forecasting usually have deadline constraints and require trusted services.In order to save rental costs when executing these applications on heterogeneous Iaa S clouds,it is necessary to consider the rental cost minimization workflow scheduling problem based on trust and deadline constraints in heterogeneous Iaa S clouds,which has the following challenges:(1)In the cloud environment,the historical feedback data of workflow services is difficult to obtain,resulting in the low accuracy of the corresponding service credibility evaluation.In the absence of historical feedback data,it is a challenge to accurately assess the credibility of services;(2)For the workflow application such as earthquake detection and hurricane forecasting,there are partial order relationship constraints between its jobs,and it has deadline and credibility constraints,its demand for resources is unbalanced,and how to reasonably allocate resources for this workflow to meet deadline and credibility constraints and minimize rental cost is a challenge.The system framework and model of the problem are first proposed.For the cloud workflow service credibility evaluation problem,based on the existing service credibility evaluation methods,comprehensively considering the impact of job execution time and job execution duration on the feedback data weight,an indirect evaluation method is proposed.It first uses the historical cloud workflow service feedback data combined with the relationship between the job execution time and the feedback data weight to calculate the credibility of the virtual machine service,then use the obtained virtual machine service credibility data combined with the Gantt chart of the current cloud workflow execution to calculate the current cloud workflow service credibility.For the rental cost minimization cloud workflow scheduling problem based on trust and deadline constraints,a scheduling algorithm is proposed,whose framework is initial solution construction,local search and optimal solution update.The initial solution construction method of the algorithm is to randomly generate the initial solution.The algorithm adjusts the solution search granularity based on the size of the solution space,so that the performance of the algorithm can be optimal under different task scales and resource scales,and a dynamic search granularity algorithm component is proposed to dynamically adapt to different resource scales and task scales.The algorithm performs a local search based on a greedy selection method,so that the credibility and deadline constraints are satisfied and the rental cost is minimized.For the problem of constructing the constraint parameters of deadline and credibility threshold,the greedy strategy is used to obtain the deadline range and credibility scope of the cloud workflow service,and then different values in the constraint parameter range are taken to construct multiple sets of constraint parameters.For the cloud workflow service credibility evaluation problem and the rental cost minimization cloud workflow scheduling problem based on trust and deadline constraints,in order to verify the performance of the proposed algorithm components and algorithms,the performance differences between the algorithm components and the algorithms for the above two problems were analyzed by means of multi-factor variance analysis technique.First,the algorithm components are calibrated,and the optimal component combination is selected;then,the comparison algorithms are selected for comparison,and the performance differences between the algorithms are analyzed.The experimental results show that for the above two problems,the performance of the proposed algorithm is significantly better than other comparison algorithms.