Scale Congestion Control to Ultra-High Speed Ethernet

摘要

Currently, Ethernet is broadly used in LAN, datacenter and enterprisenetworks, storage networks, high performance computing networks and so on.Along with the popularity of Ethernet comes the requirement of enhancingEthernet with congestion control. On the other hand, Ethernet speed extends to40Gbps and 100Gbps recently, and even 400Gbps in the near future. Theultra-high speed requires congestion control algorithms to adapt to the broadchanges of bandwidth, and highlights the impacts of small delay by enlargingthe bandwidth delay product. The state-of-art standard QCN is heuristicallydesigned for the 1Gbps and 10Gbps Ethernet, and unaware of the challengesaccompanying the ultra-high speed. To scale congestion control to ultra-high speed Ethernet, we propose theAdaptive Sliding Mode (ASM) congestion control algorithm, which is simple,stable, has fast and smooth convergence process, can tolerate the impacts ofdelay and adapt to the wide changes of bandwidth. Real experiments andsimulations confirm these good properties and show that ASM outperforms QCN.Designing ASM, we find that the derivative of queue length is helpful to rateadjustment because it reflects the difference between bandwidth and aggregatedsending rate. We also argue for enforcing congestion control system staying atthe congestion boundary line, along which it automatically slides to stablepoint. These insights are also valuable to develop other congestion controlalgorithms in ultra-high speed networks.