Elephant Flow Detection Mechanism in SDN-Based Data Center Networks

Abstract

Aiming at the problem of network congestion and unbalanced load caused by a large amount of data capacity carried in elephant flow in the data center network, an elephant flow detection method based on SDN is proposed. This method adopts the autodetect upload (ADU) mechanism. ADU is divided into two parts: ADU-Client and ADU-Server, in which ADU-Client runs in the host computer and ADU-Server runs in the SDN controller. When the host sends the elephant flow, the ADU-Client generates a packet with forged source IP address and triggers the Packet_in message of the edge switch to report the information of the elephant flow to the SDN controller and the ADU-Server completes the elephant flow identification. Experimental results show that the ADU elephant flow detection mechanism can effectively detect elephant flow in the data center network, reduce detection time, and improve network performance.