Congestion control for high bandwidth-delay product networks

Abstract

Theory and experiments show that as the per-flow product of bandwidth and latency increases, TCP becomes inefficient and prone to instability, regardless of the queuing scheme. This failing becomes increasingly important as the Internet evolves to incorporate very high-bandwidth optical links and more large-delay satellite links.To address this problem, we develop a novel approach to Internet congestion control that outperforms TCP in conventional environments, and remains efficient, fair, scalable, and stable as the bandwidth-delay product increases. This new eXplicit Control Protocol, XCP, generalizes the Explicit Congestion Notification proposal (ECN). In addition, XCP introduces the new concept of decoupling utilization control from fairness control. This allows a more flexible and analytically tractable protocol design and opens new avenues for service differentiation.Using a control theory framework, we model XCP and demonstrate it is stable and efficient regardless of the link capacity, the round trip delay, and the number of sources. Extensive packet-level simulations show that XCP outperforms TCP in both conventional and high bandwidth-delay environments. Further, XCP achieves fair bandwidth allocation, high utilization, small standing queue size, and near-zero packet drops, with both steady and highly varying traffic. Additionally, the new protocol does not maintain any per-flow state in routers and requires few CPU cycles per packet, which makes it implementable in high-speed routers.

Emulating AQM from end hosts

“…Some of the algorithms used in AQM schemes need an estimation of the round trip time of the flow to determine the feedback [18,32] to the end-host. The routers use an average value for this purpose.…”

“…One school of thought has been that the routers at the bottleneck links know when the buffers are getting full and so are in the best position to judge the onset of congestion. Once congestion is detected, they inform the end-host implicitly by either a binary signal (e.g., [2,14,16,19,25]) or a non-binary signal (e.g., [18,32]) which informs the senders of what their sending rate should be. This is the area broadly known as Active Queue Management (AQM).…”

Active Window Management: Performance Assessment through an Extensive Comparison with XCP

“…There are two different classes of solutions that adopt this approach. The first class includes schemes that are implemented both in the hosts and in the routers [3], and need undesirable modifications to the TCP protocol, or even its replacement, in order to accept, understand and use feedback signals from gateways. On the contrary, the second class of solutions includes algorithms which are implemented in the gateways only.…”

“…The most authoritative solution in the first class is XCP [3], a new transport protocol that relies on the explicit cooperation among XCP routers and XCP senders or receivers. This protocol generalizes the Explicit Congestion Notification (ECN) [4] by allowing XCP routers to inform the XCP senders about the degree of congestion at the bottleneck.…”

Making a Case for Proactive Flow Control in Optical Circuit-Switched Networks

“…In order to improve TCP performance, significant research has been done to improve TCP congestion control [16,13,10,9]. Other complementary research has been done to improve flow control [11,14,5].…”