To　address　potential　security　issues　in　OpenFlow-based　software　defined　network,　which　threat　OpenFlow　network　infrastructures　when　they　are　under　certain　attacks　and　significantly　affect　network　performance　when　abnormal　traffic　exists,　this　paper　investigated　security　issues　in　the　OpenFlow　network　and　then　proposed　a　traffic　monitor　architecture.　Combining　sFlow　sampling　technology　with　Floodlight　controller,　method　which　mitigated　network　attacks　by　changing　the　controller＇s　operational　mode　and　performing　rate-limiting　in　the　upper　layer　application　was　proposed.　Based　on　the　test　results,　it　is　proven　that　the　proposed　solution　reduces　controller　load　dramatically　under　certain　attacks,　as　it　filters　99.88%　Packet_In　packets　originated　from　attacks,　thus　effectively　reduces　effects　of　abnormal　traffic　on　hosts　and　network　itself.　The　proposed　solution　provides　real-time　attack　detection　and　mitigates　attack　effects　on　overall　network　performance.　©　2016,　Editorial　Department　of　Journal　of　Beijing　University　of　Technology.　All　right　reserved.