Analysing and improving convergence of quantized congestion notification in Data Center Ethernet

Quantized Congestion Notification (QCN) has been approved as the standard congestion management mechanism for the Data Center Ethernet (DCE). However, lots of work pointed out that QCN suffers from the problem of unfairness among different flows. In this paper, we found that QCN could achieve fairness, merely the convergence time to fairness is quite long. Thus, we build a convergence time model to investigate the reasons of the slow convergence process of QCN. We validate the precision of our model by comparing with experimental data on the NetFPGA platform. The results show that the proposed model accurately well characterizes the convergence time to fairness of QCN. Based on the model, the impact of QCN parameters, network parameters, and QCN variants on the convergence time is analysed in detail. Results indicate that the convergence time of QCN can be decreased if sources have the same rate increase probability or the rate increase step becomes larger at steady state. Enlightened by the analysis, we proposed a mechanism called QCN-T, which replaces the original Byte Counter and Timer at sources with a single modified Timer to reduce the convergence time. Finally, evaluations show great improvements of QCN-T in both convergence and stability.