Research On Cloud Task Scheduling Strategy Based On Min-Min And Max-Min Algorithm

In recent years,cloud computing as a new large-scale data computing and information storage method has been adopted by more and more scientific research institutions and commercial companies.With a large number of services being migrated to public cloud platforms,data centers are expanding and their tasks are increasing in geometric series with the number of resources.In view of this,how to design a reasonable and efficient cloud task scheduling algorithm has become an urgent problem in the current cloud computing field,which has important theoretical and engineering significance.With the cloud task scheduling algorithm,the data center assigns the tasks in the waiting queue that users upload to the virtual machine needed for execution and returns the end of processing data or services to users.The choice of a scheduling policy determines the time cost of the task and the overall resource utilization of the system.Minimizing overall task completion time and achieving overall load balancing of the system is a goal pursued by the Cloud Task Scheduling Center.Under this background,this paper takes virtual machine and cloud task as t he basic dispatching units,and combines the properties of the two,and makes a detailed study on the issues related to resource allocation such as virtual machine selection,cloud task allocation,etc.The main work of this paper is as follows:(1)For the lack of considering the similar number of simple tasks and complex tasks in existing task scheduling algorithms in the cloud environment,a two-stage greedy task scheduling algorithm(GTSA-TW)based on Min-Min algorithm and Max-Min algorithm is proposed.The algorithm first computes the expected completion time matrix from the task length of the cloud task and the virtual machine and then selects the simple and complex tasks in the current task queue to be executed by the low-load virtual machine node with the least expected completion time.Combining the two-stage idea,check whether the key nodes generated by the first-stage scheduling can schedule the most appropriate tasks to be assigned to the lightweight virtual machine to further reduce the overall completion time and achieve load balancing of the system.In order to better compare the effectiveness of the algorithm,avoid the current cloud task batch scheduling algorithm often only focus on single indicators such as task completion time or system load balancing degree,so that the corresponding algorithm can achieve better results under a single evaluation standard,but lack of comprehensive considerations on the overall task completion time and system load balancing degree,select task overall completion time and system load level is taken as comprehensive evaluation indicators to measure the performance of the algorithm.The experimental data based on Cloud Sim show that this algorithm can effectively improve the overall system load,shorten the overall task completion time,and achieve a win-win situation between users and cloud service providers compared with Max-Min and Min-Min algorithms.(2)Based on the user data of the actual cloud platform,the characteristics of cloud tasks in the cloud computing environment are further analyzed.The Min-Min algorithm and Max-Min algorithm are found to be suitable for situations where complex tasks are more simple than simple tasks and more complex tasks are less.However,the intermediate domain,where the number of simple tasks is close to that of complex tasks,is not considered.For this reason,this paper builds a cloud task scheduling model and its algorithm(CTSA-3WD)based on three-branch decision-making to achieve better results under different task distributions.First,the task queue is divided into light-loaded tasks and heavy-loaded tasks according to task length and virtual machine.Then,according to the proportion of light-loaded tasks and heavy-loaded tasks in the whole task queue,three situations are divided and corresponding scheduling strategies are adopted.Max-Min algorithm is used for light-loaded tasks with fewer tasks,Min-Min algorithm is used for heavy-loaded tasks with fewer tasks,and GTSA-TW algorithm is used for light-loaded tasks with a similar number of heavy-loaded tasks.Then,a two-stage processing mode is introduced to detect whether the pre-scheduling scheme can be further optimized to achieve better experimental results.Based on the experimental data,the algorithm fully considers the different possibilities of the cloud task queue and selects the corresponding task scheduling algorithm according to its characteristics,which greatly expands the application scope of the algorithm,and achieves the goal of shortening the overall task completion time and improving the system balance.