浅缓冲区网络中BBR与CUBIC的传输公平性研究

A STUDY ON TRANSPORT FAIRNESS OF BBR AND CUBIC IN A SHALLOW BUFFER NETWORK

  • 摘要: 当前常用的CUBIC拥塞控制算法已无法满足多数高质量网络应用的需求,具有高吞吐量和低时延的BBR(BottleneckBandwidthandRound-trippropagationtime)拥塞控制算法被提出并广泛应用,二者将在网络中长期共存。当BBR与CUBIC共享同一条浅缓冲区瓶颈链路时,BBR流会抢占大量带宽,严重影响CUBIC流的正常传输。通过研究BBR算法侵略性的产生原因,提出BBR-Compatible(BBR-C)算法,该算法通过主动发现队列堆积,及时更新往返传输时延,并在检测到丢包重传时调整发送行为,以提升瓶颈链路的传输公平性。实验结果表明:与BBR相比,BBR-C与CUBIC在浅缓冲区网络中共存时,BBR-C在保证传输质量的同时侵略性降低,链路传输的公平性显著提升,Jain公平指数均值提升了1.5至1.58倍。

     

    Abstract: Compared with the currently commonly used CUBIC congestion control algorithm, the BBR congestion control algorithm has the advantages of high throughput and low delay, and the two will coexist for a long time in the future. When BBR and CUBIC share the shallow buffer bottleneck link, the BBR flow will occupy a large amount of bandwidth, seriously affecting the normal transmission of the CUBIC flow. By studying the reasons for the aggressiveness of the BBR algorithm, the BBR-Compatible (BBR-C) algorithm is proposed. This algorithm actively discovered queue accumulation, updated the round-trip transmission delay in time, and adjusted the sending behavior when packet loss and retransmission were detected, so as to improve the transmission fairness of the bottleneck link. The experimental results show that, compared with BBR, BBR-C reduces aggression while guaranteeing transmission quality, improves the fairness of link transmission significantly, and increases the mean Jain Fairness Index by 1.5 to 1.58 times.

     

/

返回文章
返回