1.\" 2.\" Copyright (c) 2014 Midori Kato <katoon@sfc.wide.ad.jp> 3.\" Copyright (c) 2014 The FreeBSD Foundation 4.\" All rights reserved. 5.\" 6.\" Portions of this documentation were written at Keio University, Japan. 7.\" 8.\" Redistribution and use in source and binary forms, with or without 9.\" modification, are permitted provided that the following conditions 10.\" are met: 11.\" 1. Redistributions of source code must retain the above copyright 12.\" notice, this list of conditions and the following disclaimer. 13.\" 2. Redistributions in binary form must reproduce the above copyright 14.\" notice, this list of conditions and the following disclaimer in the 15.\" documentation and/or other materials provided with the distribution. 16.\" 17.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 18.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20.\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR 21.\" ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27.\" SUCH DAMAGE. 28.\" 29.\" $FreeBSD$ 30.\" 31.Dd May 13, 2021 32.Dt CC_DCTCP 4 33.Os 34.Sh NAME 35.Nm cc_dctcp 36.Nd DCTCP Congestion Control Algorithm 37.Sh DESCRIPTION 38The DCTCP (data center TCP) congestion control algorithm aims to maximise 39throughput and minimise latency in data center networks by utilising the 40proportion of Explicit Congestion Notification (ECN) marks received from capable 41hardware as a congestion signal. 42.Pp 43DCTCP uses fraction of ECN marked packets to update congestion window. 44The window reduction ratio is always <= 1/2. 45Only when all of the packets are 46marked, congestion window is halved. 47.Pp 48In order to keep the accuracy of the ECN marked fraction, a DCTCP receiver 49mirrors back incoming (or missing) CE marks by setting (or clearing) ECE marks. 50This feedback methodology is also adopted when the receiver uses delayed ACK. 51.Pp 52The 53.Fx 54DCTCP implementation includes two minor modifications for the one-sided 55deployment. 56Considering the situation that DCTCP is used as sender and classic 57ECN is used as receiver, DCTCP sets the CWR flag as the reaction to the ECE 58flag. 59In addition, when classic ECN is used as sender and DCTCP is used as 60receiver, DCTCP avoids to mirror back ACKs only when the CWR flag is 61set in the incoming packet. 62.Pp 63The other specifications are based on the paper and the RFC referenced 64in the 65.Sx SEE ALSO 66section below. 67.Sh MIB Variables 68The algorithm exposes the following tunable variables in the 69.Va net.inet.tcp.cc.dctcp 70branch of the 71.Xr sysctl 3 72MIB: 73.Bl -tag -width ".Va slowstart" 74.It Va alpha 75The initial value to estimate the congestion on the link. 76The valid range is from 0 to 1024, where 1024 reduces the congestion 77window to half, if a CE is observed in the first window and 78.Va alpha 79could not yet adjust to the congestion level on that path. 80Default is 1024. 81.It Va shift_g 82An estimation gain in the 83.Va alpha 84calculation. 85This influences the responsiveness when adjusting alpha 86to the most recent observed window. 87Valid range from 0 to 10, the default is 4, resulting in an effective 88gain of 1 / ( 2 ^ 89.Va shift_g 90), or 1/16th. 91.It Va slowstart 92A flag if the congestion window should be reduced by one half after slow start. 93Valid settings 0 and 1, default 0. 94.El 95.Sh SEE ALSO 96.Xr cc_cdg 4 , 97.Xr cc_chd 4 , 98.Xr cc_cubic 4 , 99.Xr cc_hd 4 , 100.Xr cc_htcp 4 , 101.Xr cc_newreno 4 , 102.Xr cc_vegas 4 , 103.Xr mod_cc 4 , 104.Xr tcp 4 , 105.Xr mod_cc 9 106.Rs 107.%A "Mohammad Alizadeh" 108.%A "Albert Greenberg" 109.%A "David A. Maltz" 110.%A "Jitendra Padhye" 111.%A "Parveen Patel" 112.%A "Balaji Prabhakar" 113.%A "Sudipta Sengupta" 114.%A "Murari Sridharan" 115.%T "Data Center TCP (DCTCP)" 116.%U "http://research.microsoft.com/pubs/121386/dctcp-public.pdf" 117.%J "ACM SIGCOMM 2010" 118.%D "July 2010" 119.%P "63-74" 120.Re 121.Rs 122.%A "Stephen Bensley" 123.%A "Dave Thaler" 124.%A "Praveen Balasubramanian" 125.%A "Lars Eggert" 126.%A "Glenn Judd" 127.%T "Data Center TCP (DCTCP): TCP Congestion Control for Data Centers" 128.%U "https://tools.ietf.org/html/rfc8257" 129.Re 130.Sh HISTORY 131The 132.Nm 133congestion control module first appeared in 134.Fx 11.0 . 135.Pp 136The module was first released in 2014 by Midori Kato studying at Keio 137University, Japan. 138.Sh AUTHORS 139.An -nosplit 140The 141.Nm 142congestion control module and this manual page were written by 143.An Midori Kato Mt katoon@sfc.wide.ad.jp 144and 145.An Lars Eggert Mt lars@netapp.com 146with help and modifications from 147.An Hiren Panchasara Mt hiren@FreeBSD.org 148