An Elephant Flows Scheduling Method Based on Feedforward Neural Network

摘要

In order to solve the problem of unbalanced network load caused by elephant flows in the data center network, a dynamic multi-path load balancing method based on feedforward neural network (FNN-LB) was proposed. In this method, topological perception and traffic information monitoring are carried out first, and the elephant flows are marked. The collected network traffic information is then used as input to estimate the load of each link through the feedforward neural network. Finally, the optimized ant colony algorithm is used to find the optimal paths for transmitting the elephant flows, so that the optimal paths are selected according to the real-time status of the links. Simulation results show that the proposed scheme can effectively reduce network transmission delay and packet loss rate, and improve link utilization.