Research And Application Of Real-time Scheduling Method Based On Resource Granularity

With the popularity of the Internet,the increase of the number of users and the integration of a large number of Internet products into people's lives,the economic benefits brought by the Internet have attracted the attention of various countries.But the following problem is how to quickly calculate and process the growing mass data and a large number of data requests.Nowadays,experts and scholars also have many schemes in the research of this problem,such as increasing the processing time,improving the calculation efficiency,increasing the processing machine and so on,which are adopted by most people in increasing the processing machine.However,there are too many heterogeneous computers and the fine-grained hardware resources show obvious heterogeneity,such as different processor architecture,cache level and size,memory capacity,I / O read-write speed,etc.In the process of cluster management for the added machines,the existing task scheduling scheme is simple,mainly because the hardware resource granularity is not considered carefully enough to optimize the task scheduling and resource allocation.In this paper,a scheduling scheme based on fine-grained resources is proposed to classify the computer hardware resources,and four kinds of hardware resources are considered respectively,including CPU dominant frequency,cache cache capacity,memory dominant frequency and I / O read-write speed to simulate the impact of computer hardware on task processing.The data structure of tasks is the consumption demand of these four resources.Computers with high resource performance can process tasks with the same demand quickly.The scheduling scheme is evaluated in three aspects: task processing time,task response time and hardware resource performance change.Task processing time evaluation is to calculate the difference of processing time of each processing node under different scheduling schemes.Task response time is the difference of waiting time in the task processing queue under different scheduling schemes.The change of hardware resource performance is to calculate the consumption difference of each hardware resource of the same resource node under different scheduling schemes.Through the way of simulation computer processing tasks,we test and record the experimental data and evaluate the equal quantity and quality tasks of weighted round robin,weighted lead connection and resource fine-grained scheduling scheme.The evaluation results show that the fine-grained scheduling scheme based on resources has obvious effect on task processing time,the weighted round robin has an advantage in task response time,and the fine-grained scheduling scheme based on resources has better effect on hardware resource performance changes.