This document defines TCP's four congestion control algorithms: slow start, congestion avoidance, fast retransmit, and fast recovery. It also specifies how TCP should begin transmission after a long idle period and discusses acknowledgment generation methods. The document obsoletes RFC 2581. The document specifies an Internet standards track protocol and requests discussion and suggestions for improvements. It is subject to BCP 78 and the IETF Trust's Legal Provisions. It may contain material from IETF Documents or IETF Contributions published before November 10, 2008. The document outlines definitions, congestion control algorithms, additional considerations, security considerations, and changes relative to RFC 2581. It defines terms such as segment, sender maximum segment size (SMSS), receiver maximum segment size (RMSS), congestion window (cwnd), initial window (IW), loss window (LW), restart window (RW), flight size, duplicate acknowledgment, and others. The congestion control algorithms are described in detail, including slow start and congestion avoidance, fast retransmit, and fast recovery. The document explains how TCP uses these algorithms to manage network congestion, including how to handle loss and recovery. It also discusses how TCP should restart idle connections and generate acknowledgments. The document also addresses security considerations, including the potential for attackers to impair TCP performance by causing data packet loss or forging duplicate acknowledgments. It recommends using byte counting to counter such attacks. The document outlines changes between RFC 2001 and RFC 2581, and changes relative to RFC 2581, including updates to the initial congestion window, ssthresh, and other parameters. It also recommends advanced loss recovery algorithms and clarifies the treatment of ssthresh on retransmission timeout. The authors of the document are Mark Allman, Vern Paxson, and Ethan Blanton. The document is published by the IETF and is subject to the IETF Trust's Legal Provisions.This document defines TCP's four congestion control algorithms: slow start, congestion avoidance, fast retransmit, and fast recovery. It also specifies how TCP should begin transmission after a long idle period and discusses acknowledgment generation methods. The document obsoletes RFC 2581. The document specifies an Internet standards track protocol and requests discussion and suggestions for improvements. It is subject to BCP 78 and the IETF Trust's Legal Provisions. It may contain material from IETF Documents or IETF Contributions published before November 10, 2008. The document outlines definitions, congestion control algorithms, additional considerations, security considerations, and changes relative to RFC 2581. It defines terms such as segment, sender maximum segment size (SMSS), receiver maximum segment size (RMSS), congestion window (cwnd), initial window (IW), loss window (LW), restart window (RW), flight size, duplicate acknowledgment, and others. The congestion control algorithms are described in detail, including slow start and congestion avoidance, fast retransmit, and fast recovery. The document explains how TCP uses these algorithms to manage network congestion, including how to handle loss and recovery. It also discusses how TCP should restart idle connections and generate acknowledgments. The document also addresses security considerations, including the potential for attackers to impair TCP performance by causing data packet loss or forging duplicate acknowledgments. It recommends using byte counting to counter such attacks. The document outlines changes between RFC 2001 and RFC 2581, and changes relative to RFC 2581, including updates to the initial congestion window, ssthresh, and other parameters. It also recommends advanced loss recovery algorithms and clarifies the treatment of ssthresh on retransmission timeout. The authors of the document are Mark Allman, Vern Paxson, and Ethan Blanton. The document is published by the IETF and is subject to the IETF Trust's Legal Provisions.