MPCC: online learning multipath transport
MPCC: online learning multipath transport
复制标题
DOI：
10.1145/3386367.3433030
复制
发表时间：
2020-11
期刊：
影响因子：
--
通讯作者：
Tomer Gilad;Neta Rozen Schiff;Brighten Godfrey;C. Raiciu;Michael Schapira
中科院分区：
文献类型：
--
作者：
Tomer Gilad;Neta Rozen Schiff;Brighten Godfrey;C. Raiciu;Michael Schapira
MPTCP中包含的多路径传输用于提高移动和住宅接入网络的吞吐量和可靠性，其他用例包括在数据中心和广域网中分散负载。然而，MPTCP从根本上与TCP Reno的传统AIMD算法联系在一起，并且明显落后于现代单路径设计的性能。因此，MPTCP在许多实际环境中无法实现高性能。我们提出MPCC，一个高性能的多路径拥塞控制架构。为了在具有挑战性的环境中实现公平和高性能的综合目标，MPCC采用了在线凸优化（又名在线学习）。在模拟和实时网络上的内核实现实验中，MPCC明显优于MPTCP。
Multipath transport, as embodied in MPTCP, is deployed to improve throughput and reliability in mobile and residential access networks, with additional use-cases including spreading load in data centers and WANs. However, MPTCP is fundamentally tied to TCP Reno's legacy AIMD algorithm, and significantly lags behind the performance of modern single-path designs. Consequently, MPTCP fails to achieve high performance in many real-world environments. We present MPCC, a high-performance multipath congestion control architecture. To achieve our combined goals of fairness and high performance in challenging environments, MPCC employs online convex optimization (a.k.a. online learning). In experiments with a kernel implementation on emulated and live networks, MPCC significantly outperforms MPTCP.