Harmonia: Explicit Congestion Notification and Credit-Reservation Transport Converged Congestion Control in Datacenters

Bursty traffic and thousands of concurrent flows incur inevitable network congestion in datacenter networks(DCNs)and then affect the overall performance.Various transport protocols are developed to mitigate the network congestion,including reactive and proactive protocols.Reactive schemes use different congestion signals,such as explicit congestion notification(ECN)and round trip time(RTT),to handle the network congestion after congestion arises.However,with the growth of scale and link speed in datacenters,reactive schemes encounter a significant problem of slow responding to congestion.On the contrary,proactive protocols(e.g.,credit-reservation protocols)are designed to avoid congestion before it occurs,and they have the advantages of zero data loss,fast convergence and low buffer occupancy.But credit-reservation protocols have not been widely deployed in current DCNs(e.g.,Microsoft,Amazon),which mainly deploy ECN-based protocols,such as data center transport control protocol(DCTCP)and data center quantized congestion notification(DCQCN).And in an actual deployment scenario,it is hard to guarantee one protocol to be deployed in every server at one time.When credit-reservation protocol is deployed to DCNs step by step,the network will be converted to multi-protocol state and will face the following fundamental challenges:1)unfairness,2)high buffer occupancy,and 3)heavy tail latency.Therefore,we propose Harmonia,aiming for converging ECN-based and credit-reservation protocols to fairness with minimal modification.To the best of our knowledge,Harmonia is the first to address the trouble of harmonizing proactive and reactive congestion control.Targeting the common ECN-based protocols-DCTCP and DCQCN,Harmonia leverages forward ECN and RTT to deliver real-time congestion information and redefines feedback control.After the evaluation,the results show that Harmonia effectively solves the unfair link allocation,eliminating the timeouts and addressing the buffer overflow.

An Optimal Flow Control Algorithm for Real-time Traffic over the Internet

According to the Wide Area Network model and queue dynamics in the router, the authors formulate the Internet flow control as a constrained convex programming problem, where the objective is to maximize the total utility and minimize transmission delay and delay jitter of all sources over their transmission rates. Based on this formulation, flow control can be solved by means of a gradient projection algorithm with properly rate iterations. The main difficulty facing the realization of the iteration algorithm is the distributed computation of the congestion measure. Fortunately, Explicit Congestion Notification (ECN) is likely to be used to improving the performance of TCP in the near future. By using ECN, it is possible to realize the iteration algorithm in IP networks. The algorithm is divided into two parts, algorithms in the router and the source. The main advantage of the scheme is its fast convergence ability and robustness, but small queue length fluctuation is unavoidable when the number of users increases.

An ECN-based Optimal Flow Control Algorithm for the Internet

According to the Wide Area Network model, we formulate Internet flow control as a constrained convex programming problem, where the objective is to maximize the total utility of all sources over their transmission rates. Based on this formulation, flow control can be converted to a normal unconstrained optimization problem through the barrier function method, so that it can be solved by means of a gradient projection algorithm with properly rate iterations. We prove that the algorithm converges to the global optimal point, which is also a stable proportional fair rate allocation point, provided that the step size is properly chosen. The main difficulty facing the realization of iteration algorithm is the distributed computation of congestion measure. Fortunately, Explicit Congestion Notification (ECN) is likely to be used to improve the performance of TCP in the near future. By using ECN, it is possible to realize the iteration algorithm in IP networks. Our algorithm is divided into two parts, algorithms in the router and in the source. The router marks the ECN bit with a probability that varies as its buffer occupancy varies, so that the congestion measure of links can be communicated to the source when the marked ECN bits are reflected back from its destination. Source rates are then updated by all sessions according to the received congestion measure. The main advantage of our scheme is its fast convergence ability and robustness; it can also provide the network with zero packet loss by properly choosing the queue threshold and provide differentiated service to users by applying different utility functions.