HPCC+:Fairness-Latency Tradeoff For Transport Control Protocols In Datacenter Networks

Many studies and researches on congestion control mechanisms in the Data Center Networks(DCNs)focus on how to achieve three main goals:high bandwidth,low latency,and stability in DCNs.HPCC(High Precision Congestion Control)achieves these goals simultaneously.This research mainly studies how to improve the fairness of HPCC.This research proposed to make effective adjustments to the HPCC,where we conducted an extensive study on one of the most important parameters in the HPCC WAI(window additive increase),which is responsible for controls the tradeoff between the number of concurrent flows and the speed of convergence to fairness.Where the focus was on finding the optimal value for this parameter by changing the formula that generates the value of this parameter.Relying mainly on the number of flows in the process of generating the value of WAI where every time the number of flows is changed,a new value of WAI is generated,and this process effects on the fairness.The key idea of the proposed optimal value for WAI is not to rely entirely on the number of flows but to find a constant value for WAI regardless of the number of flows or the size of its data to achieve fast convergence to fairness and zero queue length.In the simulation,the experimental results of the proposed method in the process of improving the HPCC performance through optimal fairness are very satisfactory.After many observations and evaluation processes,the value of 250MB was adopted as the optimal value of WAI,as it proved that this value provides the optimum value of fairness,high link utilization and zero queue length for HPCC.