Taming the elephants: New TCP slow start

@article{Ha2011TamingTE,
  title={Taming the elephants: New TCP slow start},
  author={Sangtae Ha and Injong Rhee},
  journal={Comput. Networks},
  year={2011},
  volume={55},
  pages={2092-2110}
}
  • S. Ha, I. Rhee
  • Published 1 June 2011
  • Computer Science
  • Comput. Networks
Gentle Slow Start to Alleviate TCP Incast in Data Center Networks
TLDR
The Gentle Slow Start algorithm is proposed, which adjusts the congestion window according to real-time congestion state in a gentle manner and smoothly switches from slow start to congestion avoidance phase and also integrate it into the state-of-the-art data center transport protocols.
Stateful-TCP—A New Approach to Accelerate TCP Slow-Start
TLDR
This work proposes a different approach called Stateful-TCP, where path bandwidth estimated in a previous flow is used to instantly ramp up the transmission rate of the subsequent flow to the same destination, enabling TCP to efficiently utilize the available path bandwidth right from the beginning.
TCP HyStart Performance over a Satellite Network
TCP slow start is designed to begin at a conservative bitrate, but quickly ramp up to the available bandwidth. To avoid overshooting, TCP slow start has a HyStart mode (on by default in Linux) that
TCP-FLASH - A Fast Reacting TCP for Modern Networks
TLDR
A novel TCP design called Fast Launch with Agile congeStion Handling (FLASH) that not only achieves improved performance for long TCP flows but also significantly raises the performance of short to medium TCP flows that are far more common in the Internet.
Making Linux TCP Fast
TLDR
This work describes several major changes that have recently been made and are still underway within the Linux TCP implementation, to decouple loss recovery from congestion control, and reworks loss detection to rest on a more general foundation of time-based analysis.
A client-side split-ACK tool for TCP Slow Start investigation
  • M. Welzl, Rolf Erik Normann
  • Computer Science
    2012 International Conference on Computing, Networking and Communications (ICNC)
  • 2012
TLDR
This work presents a tool that, by splitting ACKs into multiple pieces at the beginning of a connection, can be used to trick a host into sending a larger number of packets than normally intended within one RTT in Slow Start.
StopEG: Detecting when to stop exponential growth in TCP slow-start
TLDR
This paper introduces StopEG, an efficient mechanism to accurately and quickly detect when to stop this exponential growth ofTCP slow-start, which reacts to the changes on congestion window size rather than traditional congestion signals such as packet loss.
TCP Inigo : A mbidextrous Congestion Control
TLDR
TCP Inigo is presented, which uses independent delaybased algorithms on the sender and receiver (i.e. ambidextrously) to satisfy all five goals of an ideal network and in networks with single administrative domains, Inigo’s fairness, bandwidth, and latency indices are better than the best deployable solution.
Improving processing performance of Linux TCP SACK implementation
TLDR
It is shown that performance of the SACK scoreboard related operations do not scale without the modification whereas Linux kernel implementing the proposed modifications is able to perform at 1GbE and beyond speeds.
Agility based safety growth of Slow-Start Congestion Avoidance and Control Scheme in TCP
TLDR
The Enhanced Slow-Start algorithm, which can concentrate for avoiding heavy packet loss and improved network utilization for keeping a Congestion Window (CWND) increment/decrement manner, and perform very well while controlling packet loss with the standard TCP Reno algorithm.
...
...

References

SHOWING 1-10 OF 47 REFERENCES
TCP startup performance in large bandwidth networks
TLDR
Simulation experiments show that Astart can significantly improve the link utilization under various bandwidth, buffer size and round-trip propagation times, and Experiments also show thatAstart achieves good fairness and friendliness toward TCP New Reno.
Hybrid Slow Start for High-Bandwidth and Long-Distance Networks
TLDR
This work proposes a new algorithm, called Hybrid Slow Start, that maintains the existing Slow Start mechanism of TCPNewReno but provides trust-worthy signals to Slow Start for safely switching to Congestion Avoidance without incurring an extremely large number of packet losses.
Improving TCP startup performance using active measurements: algorithm and evaluation
  • N. Hu, P. Steenkiste
  • Business, Computer Science
    11th IEEE International Conference on Network Protocols, 2003. Proceedings.
  • 2003
TLDR
Paced Start transitions into congestion avoidance mode faster than Slow Start, has a significantly lower packet loss rate, and avoids the timeout that is often associated with slow start.
Quick-Start TCP : From Theory to Practice
TLDR
An overview of the experimental QuickStart TCP extension, an implementation in the Linux protocol stack, which has been extended to offer full QuickStart functionality, and measurements show how Quick-Start performs in practice and quantify the benefit compared to the existing TCP congestion control.
CUBIC: a new TCP-friendly high-speed TCP variant
TLDR
The CUBIC protocol modifies the linear window growth function of existing TCP standards to be a cubic function in order to improve the scalability of TCP over fast and long distance networks.
A simple refinement of slow-start of TCP congestion control
TLDR
A new variant of slow-start is presented, called smooth-start, which provides a smooth transition between the exponential and linear growth phases of the TCP congestion window, thus improving the performance of TCP congestion control at the start of a TCP connection or after a retransmission timeout.
Limited Slow-Start for TCP with Large Congestion Windows
TLDR
Limited Slow-Start is described as an optional mechanism for limiting the number of segments by which the congestion window is increased for one window of data during slow-start, in order to improve performance for TCP connections with large congestion windows.
RAPID: Shrinking the Congestion-Control Timescale
TLDR
A new approach for congestion-control is designed that allows TCP connections to boldly search for, and adapt to, the available bandwidth within a single RTT, and the rate-based design allows RAPID to be truly RTT-fair.
Scalable TCP: improving performance in highspeed wide area networks
TLDR
The preliminary results gathered suggest that the deployment of Scalable TCP would have negligible impact on existing network traffic at the same time as improving bulk transfer performance in highspeed wide area networks.
Improving processing performance of Linux TCP SACK implementation
TLDR
It is shown that performance of the SACK scoreboard related operations do not scale without the modification whereas Linux kernel implementing the proposed modifications is able to perform at 1GbE and beyond speeds.
...
...