MiFlO: A scheduling algorithm based on mice flows optimization in hybrid data center network

With the rapid development of cloud computing and emerging services, traffic in data center network is experiencing an explosive growth, which has brought a great challenge to the existing electrical data center networks. In order to take full advantage of the large bandwidth of optical interconnects, the optical and electrical hybrid data center network is proposed. However, there are still some defects in terms of flow scheduling, such as huge computing burden on controller and indeterminate mice trans-mission delay, which have a negative effect on network performance. In this paper, we propose MiFlO, a new scheduling algorithm which classifies the mice/elephant flows in servers and ensures the priority of mice flows. In this way, MiFlO can reduce the computing overhead of switches and improve the communication performance of data center network. The results show that MiFlO has better performance than traditional scheduling algorithm FIFO. The average latency of mice flows in the network can be reduced by about 12% with MiFlO deployed.