[PDF] Notes on burst mitigation for transport protocols | Semantic Scholar (2024)

After analyzing traffic from three networks, it is found that bursts are fairly rare and only large bursts cause loss in practice.

Inspired by the findings and insights from experimental observations, a Micro-burst-Aware Transport Control Protocol (MATCP), which leverages characteristic behaviors of micro-burst traffic derived from the time derivative of the queue occupancy, is proposed.

Simulation results show that the proposed new-CWV method provides faster convergence to the rate requested by a rate-limited application, demonstrated by a higher throughput, and better utilisation of unused capacity compared to Standard TCP or TCP with Congestion Window Validation.

Inspired by the findings and insights from experimental observations, S-ECN policy is proposed, which is an ECN marking policy leveraging the slope of queue length evolution, which can suppress the sharp queue length increment by over 2×, and reduce the average query completion time by ~12-27%.

To mitigate bursts of packets by the TCP proxy, this paper proposes a scheme that the proxy performs a pacing function, which places a gap between two consecutive pack- ets, which has an advantage that it requires a small amount of forwarding buffers for the pacing.

Inspired by the findings and insights from experimental observations, S-ECN policy is proposed, which is an ECN marking policy leveraging the slope of queue length increasing, which can suppress the sharp queue length increment by over 50%, and reduce the 99th percentile of query completion time by ~20%.

TC-Aix, a sender-side TCP algorithm designed to make TCP performance robust to packet reordering and delay, is described and evaluated.

The burst behavior of traffic emanating from a 10-Gbps end host across a variety of data center applications is studied, finding that at 10--100 microsecond timescales, the traffic exhibits large bursts (i.e., 10s of packets in length).

This work develops Valinor, a traffic measurement framework that consists of eBPF hooks and measurement modules in a programmable network that offers visibility into traffic burstiness over a wide span of timescales at multiple vantage points, and demonstrates that burstiness varies significantly across host stacks.

...

...

After analyzing traffic from three networks, it is found that bursts are fairly rare and only large bursts cause loss in practice.

It is shown that contrary to intuition, pacing often has significantly worse throughput than regular TCP because it is susceptible to synchronized losses and it delays congestion signals.

The Datagram Congestion Control Protocol is a transport protocol that provides bidirectional unicast connections of congestion-controlled unreliable datagrams that is suitable for applications that transfer fairly large amounts of data.

This document specifies an optional standard for TCP to increase the permitted initial window from one or two segment(s) to roughly 4K bytes, replacing RFC 2414, and provides guidance on implementation issues.

A new way for TCP to estimate available network capacity and swiftly scale its transmission rate at the start of a TCP connection is presented, which does estimate capacity, as predicted by prior simulation.

This document proposes a small modification to the way TCP increases its congestion window, basing the increase on the number of previously unacknowledged bytes each ACK covers, which improves the performance of TCP, as well as closing a security hole TCP receivers can use to induce the sender into increasing the sending rate too rapidly.

This paper uses simulation to study the dynamics of the congestion cent rol algorithm embedded in the BSD 4.3-Tahoe TCP implementation and finds that the one-way traffic clustering and loss-synchronization phenomena persist in this new situation, albeit in a slightly modified form.

This work identifies several causes of source-level bursts, including TCP's slow start, idle restart, window advancement after loss recovery, and segmentation of application messages into multiple UDP packets, and shows that the presence of packet bursts in individual flows can have a major impact on aggregate traffic.

This paper proposes a third alternative: pacing some packets at a certain rate until the ACK clock can be restarted, and describes the motivation and implementation and simulation results which show that it achieves the elapsed-time performance comparable to NSSR and loss behavior of SSR.

...

...