Adapting to Fair Coexistence of Heterogeneous Transport-Protocol Traffic for Cloud Centers
The intelligent era is fast developing to promote complex services running in the cloud. In this context, Data traffic with heterogeneous transport protocols is intertwined in the cloud centers, especially the reactive protocols and proactive protocols, hurting the fairness of data transmission. In this paper, we propose FairHet, a credit-scheduled protocol that utilizes delay information as the congestion signal to achieve fair coexistence of heterogeneous traffic while maintaining high-efficiency data transmission. Compared to existing solutions, FairHet has a broader application scope through mere algorithm-level modifications. The experimental results demonstrate that FairHet handles heterogeneous traffic coexistence scenarios well while retaining high-performance characteristics, including low buffer occupancy and rapid convergence.