Communication Scheduling Performance Evaluation of Industrial Data Center Based on Policy Gradient Algorithm
In order to improve the network scheduling capability of industrial data centers,a communication scheduling performance evaluation method based on policy gradient algorithm was designed.The resource data of coflow is transformed into the image to be executed,and the optimal communication scheduling is realized by the strategy gradient algorithm.The results show that increasing the coflow arrival rate will increase the average process processing time of coflow.When the coflow reach rate is very low,PGO and GCN have similar characteristics,and both are better than DNN.This research is helpful to improve the safety of industrial operation and save energy,and has good application value.
industrial benefitdata center networkcommunication schedulingstrategy gradient algorithmperformance evaluation
万方数据
维普
