Queuing theory is used to classify the regularity of educational resource allocation by hierarchy in this paper. The Poisson distribution is used to label the resource queue interval parameters to obtain the expected value of individual queues. The average delay of each class of packets can be reasonably predicted by concatenating individual resources using the Lagrange multiplier algorithm. The load balancing algorithm in each queueing series calculates the average arrival rate of task requests to achieve an optimal model design. The results show that, under the guidance of queuing theory, the coefficient of variance in the participation of Chinese university students in global education governance is reduced by 0.25% on average, and the average of comprehensive efficiency reaches about 1.01%, which realizes the optimal and reasonable allocation of resources.