Quantized Congestion Notification for multicast in data center networks
Yuki Tanisawa, Yuki Hayashi, Miki Yamamoto · 2012
Quantized Congestion Notification with Bottleneck Selection (QCN/BS) is proposed as a new Layer 2 congestion control algorithm in multicast environments. In data center networks, group communication is currently playing an important role and multicast communications is an effective way to support group communication for large number of virtual machines. Layer-2 congestion control named QCN (Quantized Congestion Notification) has been proposed for supporting high reliability required by LAN/SAN integration in data center networking. Currently, when conventional QCN is applied to multicast flows, a sending device might receive congestion feedback from multiple bottleneck locations on a multicast tree and multicast flows may suffer lower throughput than unicast flows going through the same bottlenecks. In QCN/BS, the switch in the worst congestion level is selected and the transmission rate of the sending device is calculated exclusively according to feedback from the selected switch. Simulation results show that when conventional QCN is used a multicast flow is subject to lower and severely unfair throughput compared to a unicast flow. The proposed QCN/BS resolves this problem.