A Roadmap for Transmission Control Protocol (TCP) Specification Documents

A correct and efficient implementation of the Transmission Control Protocol (TCP) is a critical part of the software of most Internet hosts. As TCP has evolved over the years, many distinct documents have become part of the accepted standard for TCP. At the same time, a large number of experimental modifications to TCP have also been published in the RFC series, along with informational notes, case studies, and other advice. As an introduction to newcomers and an attempt to organize the plethora of information for old hands, this document contains a "roadmap" to the TCP-related RFCs. It provides a brief summary of the RFC documents that define TCP. This should provide guidance to implementers on the relevance and significance of the standards-track extensions, informational notes, and best current practices that relate to TCP. This document is not an update of RFC 1122 and is not a rigorous standard for what needs to be implemented in TCP. This document is merely an informational roadmap that captures, organizes, and summarizes most of the RFC documents that a TCP implementer, experimenter, or student should be aware of. Particular comments or broad categorizations that this document makes about individual mechanisms and behaviors are not to be taken as definitive, nor should the content of this document alone influence implementation decisions. This roadmap includes a brief description of the contents of each TCP-related RFC. In some cases, we simply supply the abstract or a key summary sentence from the text as a terse description. In addition, a letter code after an RFC number indicates its category in the RFC series (see BCP 9 for explanation of these categories): S - Standards Track (Proposed Standard, Draft Standard, or Internet Standard) E - Experimental I - Informational H - Historic B - Best Current Practice U - Unknown (not formally defined) Note that the category of an RFC does not necessarily reflect its current relevance. For instance, RFC 5681 is considered part of the required core functionality of TCP, although the RFC is only a Draft Standard. Similarly, some Informational RFCs contain significant technical proposals for changing TCP. Finally, if an error in the technical content has been found after publication of an RFC, this fact is indicated by the term "(Errata)" in the headline of the RFC's description. The contents of the errata can be found at the RFC editor home page . This roadmap is divided into three main sections. lists the RFCs that describe absolutely required TCP behaviors for proper functioning and interoperability. Further RFCs that describe strongly encouraged, but non-essential, behaviors are listed in . Experimental extensions that are not yet standard practices, but that potentially could be in the future, are described in . The reader will probably notice that these three sections are broadly equivalent to MUST/SHOULD/MAY specifications (per RFC 2119 ), and although the authors support this intuition, this document is merely descriptive; it does not represent a binding standards-track position. Individual implementers still need to examine the standards documents themselves to evaluate specific requirement levels. describes both the procedures that the Internet Assigned Numbers Authority (IANA) uses and an RFC author should follow when new TCP parameters are requested and finally assigned. A small number of older experimental extensions that have not been widely implemented, deployed, and used are noted in . Many other supporting documents that are relevant to the development, implementation, and deployment of TCP are described in . A small number of fairly ubiquitous important implementation practices that are not currently documented in the RFC series is listed in . Within each section, RFCs are listed in the chronological order of their publication dates.

A small number of documents compose the core specification of TCP. These define the required core functionalities of TCP's header parsing, state machine, congestion control, and retransmission timeout computation. These base specifications must be correctly followed for interoperability. This is the fundamental TCP specification document . Written by Jon Postel as part of the Internet protocol suite's core, it describes the TCP packet format, the TCP state machine and event processing, and TCP's semantics for data transmission, reliability, flow control, multiplexing, and acknowledgment. Section 3.6 of RFC 793, describing TCP's handling of the IP precedence and security compartment, is mostly irrelevant today. RFC 2873 (see ) changed the IP precedence handling, and the security compartment portion of the API is no longer implemented or used. In addition, RFC 793 did not describe any congestion control mechanism. Otherwise, however, the majority of this document still accurately describes modern TCPs. RFC 793 is the last of a series of developmental TCP specifications, starting in the Internet Experimental Notes (IENs) and continuing in the RFC series. This document updates and clarifies RFC 793 (see ), fixing some specification bugs and oversights. It also explains some features such as keep-alives and Karn's and Jacobson's RTO estimation algorithms . ICMP interactions are mentioned, and some tips are given for efficient implementation. RFC 1122 is an Applicability Statement, listing the various features that MUST, SHOULD, MAY, SHOULD NOT, and MUST NOT be present in standards-conforming TCP implementations. Unlike a purely informational "roadmap", this Applicability Statement is a standards document and gives formal rules for implementation. This document is of relevance to TCP because it defines how the pseudo-header for TCP's checksum computation is derived when 128-bit IPv6 addresses are used instead of 32-bit IPv4 addresses. Additionally, RFC 2675 (see ) describes TCP changes required to support IPv6 jumbograms. This document removes from the TCP specification all processing of the precedence bits of the TOS byte of the IP header. This resolves a conflict over the use of these bits between RFC 793 and Differentiated Services . Although RFC 793 (see ) did not contain any congestion control mechanisms, today congestion control is a required component of TCP implementations. This document defines the current versions of Van Jacobson's congestion avoidance and control mechanisms for TCP, based on his 1988 SIGCOMM paper . A number of behaviors that together constitute what the community refers to as "Reno TCP" are described in RFC 5681. The name "Reno" comes from the Net/2 release of the 4.3 BSD operating system. This is generally regarded as the least common denominator among TCP flavors currently found running on Internet hosts. Reno TCP includes the congestion control features of slow start, congestion avoidance, fast retransmit, and fast recovery. RFC 5681 details the currently accepted congestion control mechanism, while RFC 1122 mandates that such a congestion control mechanism must be implemented. RFC 5681 differs slightly from the other documents listed in this section, as it does not affect the ability of two TCP endpoints to communicate; however, congestion control remains a critical component of any widely deployed TCP implementation and is required for the avoidance of congestion collapse and to ensure fairness among competing flows. RFC 2001 and RFC 2581 are the conceptual precursors of RFC 5681. The most important changes relative to RFC 2581 are: The initial window requirements were changed to allow larger Initial Windows as standardized in (see ). During slow start and congestion avoidance, the usage of Appropriate Byte Counting (see ) is explicitly recommended. The use of Limited Transmit (see ) is now recommended. This document analyzes how current TCP stacks process TCP urgent indications, and how the behavior of widely deployed middleboxes affects the urgent indications processing. The document updates the relevant specifications such that it accommodates current practice in processing TCP urgent indications. Finally, the document raises awareness about the reliability of TCP urgent indications in the Internet, and recommends against the use of urgent mechanism. Abstract: "This document defines the standard algorithm that Transmission Control Protocol (TCP) senders are required to use to compute and manage their retransmission timer. It expands on the discussion in section 4.2.3.1 of RFC 1122 (see ) and upgrades the requirement of supporting the algorithm from a SHOULD to a MUST." . RFC 6298 updates RFC 2988 by changing the initial RTO from 3s to 1s This document clarifies what value to use with the TCP Maximum Segment Size (MSS) option when IP and TCP options are in use.

This section describes recommended TCP modifications that improve performance and security. represents fundamental changes to the protocol. and list improvements over the congestion control and loss recovery mechanisms as specified in RFC 5681 (see ). describes algorithms that allow a TCP sender to detect whether it has entered loss recovery spuriously. comprises Path MTU Discovery mechanisms. Schemes for TCP/IP header compression are listed in . Finally, deals with the problem of preventing preventing acceptance of forged segments and flooding attacks.

RFCs 2675 and XXXX represent fundamental changes to TCP by redefining how parts of the basic TCP header and options are interpreted. RFC XXXX defines the Window Scale Option, which re-interprets the advertised receive window. RFC 2675 specifies that MSS option and urgent pointer fields with a value of 65,535 are to be treated specially. IPv6 supports longer datagrams than were allowed in IPv4. These are known as jumbograms, and use with TCP has necessitated changes to the handling of TCP's MSS and Urgent fields (both 16 bits). This document explains those changes. Although it describes changes to basic header semantics, these changes should only affect the use of very large segments, such as IPv6 jumbograms, which are currently rarely used in the general Internet. Supporting the behavior described in this document does not affect interoperability with other TCP implementations when IPv4 or non-jumbogram IPv6 is used. This document states that jumbograms are to only be used when it can be guaranteed that all receiving nodes, including each router in the end-to-end path, will support jumbograms. If even a single node that does not support jumbograms is attached to a local network, then no host on that network may use jumbograms. This explains why jumbogram use has been rare, and why this document is considered a performance optimization and not part of TCP over IPv6's basic functionality. This document defines TCP extensions for window scaling, timestamps, and protection against wrapped sequence numbers, for efficient and safe operation over paths with large bandwidth-delay products. These extensions are commonly found in currently used systems. The predecessor of this document, RFC 1323, was published in 1992, and is deployed in most TCP implementations. This document includes fixes and clarifications based on the gained deployment experience. One specific issued addressed in this specification is a recommendation how to modify the algorithm for estimating the mean RTT when timestamps are used. RFC 1072, RFC 1185, and RFC RFC 1323 are the conceptual precursors of RFC XXXX.

Two of the most important aspects of TCP are its congestion control and loss recovery features. TCP treats lost packets as indicating congestion-related loss, and cannot distinguish between congestion-related loss and loss due to transmission errors. Even when ECN is in use, there is a rather intimate coupling between congestion control and loss recovery mechanisms. There are several extensions to both features, and more often than not, a particular extension applies to both. In this two sub-sections, we group enhancements to TCP's congestion control, while the next sub-section focus on TCP's loss recovery. This document defines a means for end hosts to detect congestion before congested routers are forced to discard packets. Although congestion notification takes place at the IP level, ECN requires support at the transport level (e.g., in TCP) to echo the bits and adapt the sending rate. This document updates RFC 793 (see ) to define two previously unused flag bits in the TCP header for ECN support. RFC 3540 (see ) provides a supplementary (experimental) means for more secure use of ECN, and RFC 2884 (see ) provides some sample results from using ECN. This document specifies an increase in the permitted initial window for TCP from one segment to three or four segments during the slow start phase, depending on the segment size. This document suggests that congestion control use the number of bytes acknowledged instead of the number of acknowledgments received. The ABC mechanism behaves differently than the standard method when there is not a one-to-one relationship between data segments and acknowledgments. ABC still operates within the accepted guidelines, but is more robust to delayed ACKs and ACK-division . ABC is recommended by RFC 5681 (see ). This document formally deprecates the use of ICMP Source Quench messages by transport protocols and recommends against the implementation of .

For the typical implementation of the TCP fast recovery algorithm described in RFC 5681 (see ), a TCP sender only retransmits a segment after a retransmit timeout has occurred, or after three duplicate ACKs have arrived triggering the fast retransmit. A single RTO might result in the retransmission of several segments, while the fast retransmit algorithm in RFC 5681 leads only to a single retransmission. Hence, multiple losses from a single window of data can lead to a performance degradation. Documents listed in this section aim to improve the overall performance of TCP's standard loss recovery algorithms. In particular, some of them allows TCP senders to recover more effectively when multiple segments are lost from a single flight of data. When more than one packet is lost during one round trip time TCP may experience poor performance since a TCP sender can only learn about a single lost packet per round trip time from cumulative acknowledgments. This document defines the basic selective acknowledgment (SACK) mechanism for TCP, which can help to overcome these limitations. The receiving TCP returns SACK blocks to inform the sender which data has been received. The sender can then retransmit only the missing data segments. Abstract: "This document proposes Limited Transmit, a new Transmission Control Protocol (TCP) mechanism that can be used to more effectively recover lost segments when a connection's congestion window is small, or when a large number of segments are lost in a single transmission window." Tests from 2004 showed that Limited Transmit was deployed in roughly one third of the web servers tested . Limited Transmit is recommended by RFC 5681 (see ). This document specifies a modification to the standard Reno fast recovery algorithm, whereby a TCP sender can use partial acknowledgments to make inferences determining the next segment to send in situations where SACK would be helpful but isn't available. Although it is only a slight modification, the NewReno behavior can make a significant difference in performance when multiple segments are lost from a single window of data. RFC 2582 and RFC 3782 are the conceptual precursors of RFC 6582. The main change in RFC 3782 relative to RFC 2582 was to specify the Careful variant of NewReno's Fast Retransmit and Fast Recovery algorithms and advance those two algorithms from Experimental to Standards Track status. The main change in RFC 6582 relative to RFC 3782 was to solve a performance degradation that could occur if FlightSize on Full ACK reception is zero. This document describes a conservative loss recovery algorithm for TCP that is based on the use of the selective acknowledgment (SACK) TCP option (see ). The algorithm conforms to the spirit of the congestion control specification in RFC 5681 (see ), but allows TCP senders to recover more effectively when multiple segments are lost from a single flight of data. RFC 6675 is a revision of RFC 3517 to address several situations that are not handled explicitly before. In particular it improves the loss detection in the event that the sender has outstanding segments that are smaller than SMSS. it modifies the definition of a "duplicate acknowledgment" to utilize the SACK information in detecting loss. it maintains the ACK clock under certain circumstances involving loss at the end of the window.

Spurious retransmission timeouts are harmful to TCP performance and multiple algorithms have been defined for detecting when spurious retransmissions have occurred, and then responding differently in order to recover performance. The IETF defined multiple algorithms because there are tradeoffs in whether or not certain TCP options need to be implemented, and concerns about IPR status. The Standards Track documents in this section are closely related to the Experimental documents in also addressing this topic. This document extends RFC 2018 (see ). It enables use of the SACK option to acknowledge duplicate packets. With this extension, called DSACK, the sender is able to infer the order of packets received at the receiver, and therefore to infer when it has unnecessarily retransmitted a packet. A TCP sender could then use this information to detect spurious retransmissions (see . This document describes the response portion of the Eifel algorithm, which can be used in conjunction with one of several methods of detecting when loss recovery has been spuriously entered, such as the Eifel detection algorithm in RFC 3522 (see ), the algorithm in RFC 3708 (see ), or F-RTO in RFC 5682 (see ). Abstract: "Based on an appropriate detection algorithm, the Eifel response algorithm provides a way for a TCP sender to respond to a detected spurious timeout. It adapts the retransmission timer to avoid further spurious timeouts, and can avoid - depending on the detection algorithm - the often unnecessary go-back-N retransmits that would otherwise be sent. In addition, the Eifel response algorithm restores the congestion control state in such a way that packet bursts are avoided." The F-RTO detection algorithm , originally described in RFC 4138, provides an option for inferring spurious retransmission timeouts. Unlike some similar detection methods (e.g. RFC 3522 in and RFC 3708 in ), F-RTO does not rely on the use of any TCP options. The basic idea is to send previously unsent data after the first retransmission after a RTO. If the ACKs advance the window, the RTO may be declared spurious.

The MTUs supported by different links and tunnels within the Internet can vary widely. Fragmentation of packets larger than the supported MTU on a hop is undesirable. As TCP is the segmentation layer for dividing an application's bytestream into IP packet payloads, TCP implementations generally include Path MTU Discovery (PMTUD) mechanisms in order to maximize the size of segments they send, without causing fragmentation within the network. Some algorithms may utilize signaling from routers on the path that the MTU has been exceeded. Abstract: "This memo describes a technique for dynamically discovering the MTU of an arbitrary Internet path. It specifies a small change to the way routers generate one type of ICMP message. For a path that passes through a router that has not been so changed, this technique might not discover the correct path MTU, but it will always choose a path MTU as accurate as, and in many cases more accurate than, the path MTU that would be chosen by current practice." Abstract: "This document describes Path MTU Discovery for IP version 6. It is largely derived from RFC 1191 (see ), which describes Path MTU Discovery for IP version 4." Abstract: "This document describes a robust method for Path MTU Discovery (PMTUD) that relies on TCP or some other Packetization Layer to probe an Internet path with progressively larger packets. This method is described as an extension to RFC 1191 (see ) and RFC 1981 (see ), which specify ICMP-based Path MTU Discovery for IP versions 4 and 6, respectively."

Especially in streaming applications, the overhead of TCP/IP headers could correspond to more then 50% of the total amount of data sent. Such large overheads may be tolerable in wired LANs where capacity is often not an issue, but are excessive for WANs and wireless systems where bandwidth is scarce. Header compression schemes for TCP/IP like "RObust Header Compression (ROHC) can significantly compress this overhead. It performs well over links with significant error rates and long round-trip times. This document describes a method for compressing the headers of TCP/IP datagrams to improve performance over low speed serial links. The method described in this document is limited in its handling of TCP options and cannot compress the headers of SYNs and FINs. From abstract: "This document specifies a RObust Header Compression (ROHC) profile for compression of TCP/IP packets. The profile, called ROHC-TCP, provides efficient and robust compression of TCP headers, including frequently used TCP options such as selective acknowledgments (SACKs) and Timestamps." RFC 6846 is the successor of RFC 4996. It fixes a technical issue with the SACK compression and clarifies other compression methods used.

By default, TCP lacks any cryptographic structures to differentiate legitimate segments from those spoofed from malicious hosts. Spoofing valid segments requires correctly guessing a number of fields. The documents in this sub-section describe ways to make that guessing harder, or to prevent it from being able to affect a connection negatively. This document discusses the recently increased vulnerability of long-lived TCP connections, such as BGP connections, to reset (send RST) spoofing attacks. The document analyzes the vulnerability, discussing proposed solutions at the transport level and their inherent challenges, as well as existing network level solutions and the feasibility of their deployment. This document describes a non-standard but widely implemented modification to TCP's handling of ICMP soft error messages that rejects pending connection-requests when such error messages are received. This behavior reduces the likelihood of long delays between connection-establishment attempts that may arise in some scenarios. This document describes the well-known TCP SYN flooding attack. It analyzes and discusses various countermeasures against these attacks, including their use and trade-offs. This document describes the TCP Authentication Option (TCP-AO), which is used to authenticate TCP segments. TCP-AO obsoletes the TCP MD5 Signature option of RFC 2385. It supports the use of stronger hash functions, protects against replays for long-lived TCP connections (as used, e.g., in BGP and LDP), coordinates key exchanges between endpoints, and provides a more explicit recommendation for external key management. Cryptographic algorithms for TCP-AO are defined in (see ). This document specifies the algorithms and attributes that can be used in TCP Authentication Option's (TCP-AO) (see ) current manual keying mechanism and provides the interface for future message authentication codes (MACs). Abstract: "This document discusses the use of the Internet Control Message Protocol (ICMP) to perform a variety of attacks against the Transmission Control Protocol (TCP). Additionally, this document describes a number of widely implemented modifications to TCP's handling of ICMP error messages that help to mitigate these issues." This document describes minor modifications to how TCP handles inbound segments. This renders TCP connections, especially long-lived connections such as H-323 or BGP, less vulnerable to spoofed packet injection attacks where the 4-tuple (the source and destination IP addresses and the source and destination ports) has been guessed. Abstract: "This document specifies an algorithm for the generation of TCP Initial Sequence Numbers (ISNs), such that the chances of an off-path attacker guessing the sequence numbers in use by a target connection are reduced. This document revises (and formally obsoletes) RFC 1948, and takes the ISN generation algorithm originally proposed in that document to Standards Track, formally updating RFC 793 (see ).

The RFCs in this section are still experimental, but they may become proposed standards in the future. At least part of the reason that they are still experimental is to gain more wide-scale experience with them before a standards track decision is made. At this point is worth mentioning that if the experimental RFC is a proposal for a new protocol capability or service, i.e., it requires a new TCP option code point, the implementation and experimentation should follows (see ), which describes how the experimental TCP option code points can concurrently support multiple TCP extensions. By their publication as experimental RFCs, it is hoped that the community of TCP researchers will analyze and test the contents of these RFCs. Although experimentation is encouraged, there is not yet formal consensus that these are fully logical and safe behaviors. Wide-scale deployment of implementations that use these features should be well thought-out in terms of consequences.

As multiple flows may share the same paths, sections of paths, or other resources, the TCP implementation may benefit from sharing information across TCP connections or other flows. Some Experimental proposals have been documented and some implementations have included the concepts. This document suggests how TCP connections between the same endpoints might share information, such as their congestion control state. To some degree, this is done in practice by a few operating systems; for example, Linux currently has a destination cache. Although this RFC is technically informational, the concepts it describes are in experimental use, so we include it in this section. This document , the Congestion Manager, is a related proposal to RFC 2140 (see ). The idea behind the Congestion Manager, moving congestion control outside of individual TCP connections, represents a modification to the core of TCP, which supports sharing information among TCP connections. Although a Proposed Standard, some pieces of the Congestion Manager support architecture have not been specified yet, and it has not achieved use or implementation beyond experimental stacks, so it is not listed among the standard TCP enhancements in this roadmap.

Like the standard documents listed in there newly exist also experimental RFCs that represent fundamental changes to TCP. One example is TCP Fast Open that deviates from the standard TCP semantics of . This document describes TCP Fast Open that allows data to be carried in the SYN and SYN-ACK packets and consumed by the receiver during the initial connection handshake. It saves up to one RTT compared to the standard TCP, which requires a three-way handshake to complete before data can be exchanged.

TCP congestion control has been an extremely active research area for many years (see RFC 5783, ), as it determines the performance of many applications that use TCP. A number of experimental RFCs address issues with flow start-up, overshoot, and steady-state behavior in the basic RFC 5681 (see ) algorithms. In this sub-sections, enhancements to TCP's congestion control are listed. The next sub-section focus on TCP's loss recovery. This document suggests reducing the congestion window over time when no packets are flowing. This behavior is more aggressive than that specified in RFC 5681 (see ), which says that a TCP sender SHOULD set its congestion window to the initial window after an idle period of an RTO or greater. This document describes an optional addition to ECN that protects against accidental or malicious concealment of marked packets from the TCP sender. This document proposes a modification to TCP's congestion control mechanism for use with TCP connections with large congestion windows, to allow TCP to achieve a higher throughput in high-bandwidth environments. This document describes a more conservative slow-start behavior to prevent massive packet losses when a connection uses a very large congestion window. This document specifies the optional Quick-Start mechanism for TCP. This mechanism allows connections to use higher sending rates at the beginning of the data transfer or after an idle period, provided that there is significant unused bandwidth along the path, and the sender and all of the routers along the path approve this higher rate. This document describes an experimental modification to ECN (see ) for the use of ECN in TCP SYN/ACK packets. This would allow to ECN-mark rather than drop the TCP SYN/ACK packet at an ECN-capable router, and to avoid the severe penalty of a retransmission timeout for a connection when the SYN/ACK packet is dropped. This document describes a congestion control mechanism for acknowledgment (ACKs) traffic in TCP. The mechanism is based on the acknowledgment congestion control of the Datagram Congestion Control Protocol's (DCCP's) Congestion Control Identifier (CCID) 2 . This document proposes to increase the TCP initial window from between 2 and 4 segments, as specified in RFC 3390 (see ), to 10 segments with a fallback to the existing recommendation when performance issues are detected.

This document proposes the "Early Retransmit" mechanism for TCP (and SCTP) that can be used to recover lost segments when a connection's congestion window is small. In certain special circumstances, Early Retransmit reduces the number of duplicate acknowledgments required to trigger fast retransmit to recover segment losses without waiting for a lengthy retransmission timeout. This document describes how standard ICMP messages can be used to disambiguate true congestion loss from non-congestion loss caused by connectivity disruptions. It proposes a reversion strategy of TCP's retransmission timer that enables a more prompt detection of whether or not the connectivity has been restored. This document describes an experimental Proportional Rate Reduction (PRR) algorithm as an alternative to the widely deployed Fast Recovery algorithm, to improve the accuracy of the amount of data sent by TCP during loss recovery.

In addition to the Standards Track extensions to deal with spurious retransmissions in , Experimental proposals have also been documented. The Eifel detection algorithm allows a TCP sender to detect a posteriori whether it has entered loss recovery unnecessarily by using the TCP timestamp option to solve the ACK ambiguity. Abstract: "TCP and Stream Control Transmission Protocol (SCTP) provide notification of duplicate segment receipt through Duplicate Selective Acknowledgement (DSACKs) and Duplicate Transmission Sequence Number (TSN) notification, respectively. This document presents conservative methods of using this information to identify unnecessary retransmissions for various applications." In the presence of non-congestion events, such as reordering an out-of-order segment does not necessarily indicates a lost segment and congestion. This document proposes to increase the threshold used to trigger a fast retransmission from the fixed value of three duplicate ACKs to about one congestion window of data in order to disambiguate true segment loss from segment reordering.

Besides the well known retransmission timeout the TCP standard defines two more timeouts: the user timeout and the time-wait timeout. This section lists documents that deals with TCP's various timouts. As a local per-connection parameter the TCP user timeout controls how long transmitted data may remain unacknowledged before a connection is forcefully closed. This document specifies the TCP User Timeout Option that allows one end of a TCP connection to advertise its current user timeout value. This information provides advice to the other end of the TCP connection to adapt its user timeout accordingly.

MultiPath TCP (MPTCP) is an ongoing effort within the IETF that allows a TCP connection to simultaneously use multiple IP-addresses/interfaces to spread their data across several subflows, while presenting a regular TCP interface to applications. Benefits of this include better resource utilization, better throughput and smoother reaction to failures. The documents listed in this section specify the Multipath TCP scheme, while the documents in Sections , , and provide some additional background information. This document presents a congestion control algorithm for multipath transport protocols such as Multipath TCP. It couples the congestion control algorithms running on different subflows by linking their increase functions, and dynamically controls the overall aggressiveness of the multipath flow. The result is an algorithm that is fair to TCP at bottlenecks while moving traffic away from congested links. This document presents protocol changes required to add multipath capability to TCP; specifically, those for signaling and setting up multiple paths ("subflows"), managing these subflows, reassembly of data, and termination of sessions.

RFCs listed here describes both the procedures that the Internet Assigned Numbers Authority (IANA) uses when handling assignments and the procedures an RFC author should follow when requesting new TCP option codepoints. Abstract: "This memo provides guidance for the IANA to use in assigning parameters for fields in the IPv4, IPv6, ICMP, UDP and TCP protocol headers." This document reserves both TCP options 253 and 254 for experimentation purposes. When such experiments are deployed in the Internet, they should follow the additional requirements in RFC 6994 (see ). From abstract: "This document defines the procedures that the Internet Assigned Numbers Authority (IANA) uses when handling assignment and other requests related to the Service Name and Transport Protocol Port Number registry." This document describes how the experimental TCP option code points can concurrently support multiple TCP extensions, even within the same connection. It creates an IANA registry for extensions to the experimental code points.

The RFCs listed here define extensions that have thus far failed to arouse substantial interest from implementers and have never seen widespread deployment, or were found to be defective for general use. Most of them are reclassified by to Historic status. RFC 721 addresses the problem of implementing a reliable out-of-band signal (interrupts) for use in a host-to-host protocol. The proposal was not included in the final TCP specification. This document proposes a protocol to contact multiple services on a single well-known TCP port using a service name instead of a well-known number. This RFC defined an alternative to the Window Scale option for using large windows and described the "negative acknowledgment" or NAK option. There is a comparison of NAK and SACK methods, and early discussion of TCP over satellite issues. RFC 1110 (see ) explains some problems with the approaches described in RFC 1106. The options described in this document have not been adopted by the larger community, although NAKs are used in the SCPS-TP adaptation of TCP for satellite and spacecraft use, developed by the Consultative Committee for Space Data Systems (CCSDS). Abstract: "The TCP Big Window option discussed in RFC 1106 (see ) will not work properly in an Internet environment which has both a high bandwidth * delay product and the possibility of disordering and duplicating packets. In such networks, the window size must not be increased without a similar increase in the sequence number space. Therefore, a different approach to big windows should be taken in the Internet." This document defined more robust TCP checksums than the 16-bit ones-complement in use today. A typographical error in RFC 1145 is fixed in RFC 1146; otherwise, the documents are the same. This document argues against "backwards compatible" TCP extensions. Specifically mentioned are several TCP enhancements that have been successful, including timestamps, window scaling, PAWS, and SACK. RFC 1263 presents an alternative approach called "protocol evolution", whereby several evolutionary versions of TCP would exist on hosts. These distinct TCP versions would represent upgrades to each other and could be header-incompatible. Interoperability would be provided by having a virtualization layer select the right TCP version for a particular connection. This idea did not catch on with the community, while the type of extensions RFC 1263 specifically targeted as harmful did become popular. See RFC 1644, . The inventors of TCP believed that cached connection state could have been used to eliminate TCP's 3-way handshake, to support two-packet request/response exchanges. RFC 1379 (see ) and RFC 1644 show that this is far from simple. Furthermore, T/TCP floundered on the ease of denial-of-service attacks that can result. One idea pioneered by T/TCP lives on in RFC 2140 (see ), in the sharing of state across connections. This document defines a TCP extension for applications that do not care about the order in which application-layer objects are received. Examples are multimedia and database applications. In practice, these applications either accept the possible performance loss because of TCP's strict ordering or they use specialized transport protocols other than TCP, such as PR-SCTP . To overcome the exhaustion of the IP class B address space, suggest this document that a new version of TCP (TCPng) needs to be developed and deployed. It proposes that a globally unique address be assigned to Transport layer to uniquely identify an Internet host without specifying any routing information. Later work on splitting locator and identifier values is summarized well in , but no resulting changes to TCP have occurred. This document describes a method to exchange a cookie (nonce) during the connection establishment to negotiate elimination of receiver state. These cookies are later used to inhibit premature closing of connections, and reduce retention of state after the connection has terminated. Since the cookie pair is too large to fit with the other TCP options in the 40 bytes of TCP option space, the document further describes a method to extent the option space after the connection establishment. Although RFC 6013 was published in 2011, the authors of this document places it in this section of the roadmap document due to two factors. The authors are not aware of any wide deployment and use of RFC 6013. RFC 6013 uses experimental TCP option codepoints, which prohibits a large scale deployment.

This section contains several classes of documents that do not necessarily define current protocol behaviors, but that are nevertheless of interest to TCP implementers. describes several foundational RFCs that give modern readers a better understanding of the principles underlying TCP's behaviors and development over the years. contains architectural guidelines and principles for TCP architects and designers. The documents listed in provide advice on using TCP in various types of network situations that pose challenges above those of typical wired links. Guidance for developing, analyzing, and evaluating TCP is given in . Some implementation notes and implementation advice can be found in . RFCs that describe tools for testing and debugging TCP implementations or that contain high-level tutorials on the protocol are listed . The TCP Management Information Bases are described in , and lists a number of case studies that have explored TCP performance.

The documents listed in this section contain information that is largely duplicated by the standards documents previously discussed. However, some of them contain a greater depth of problem statement explanation or other context. Particularly, RFCs 813 - 817 (known as the "Dave Clark Five") describe some early problems and solutions (RFC 815 only describes the reassembly of IP fragments and is not included in this TCP roadmap). This document is a very early precursor of the fundamental RFC 793 (see ), which already contained the three-way handshake in its final form and the concept of sliding windows for reliable data transmission. Apart from that the segment layout is totally different and the specified API differs from the latter RFC 793 (see ). This document is the immediate precursor of RFC 793 (see ). The header format, the connection establishment including the different connection states, and the overall API correspond mostly to the final Standard RFC 793 (see ). This document contains an early discussion of Silly Window Syndrome and its avoidance and motivates and describes the use of delayed acknowledgments. Suggestions and guidance for the design of tables and algorithms to keep track of various identifiers within a TCP/IP implementation are provided by this document . In this document , TCP's response to indications of network error conditions such as timeouts or received ICMP messages is discussed. This document contains implementation suggestions that are general and not TCP specific. However, they have been used to develop TCP implementations and describe some performance implications of the interactions between various layers in the Internet stack. Conclusion: "The sometimes-expressed fear that using TCP on a local net is a bad idea is unfounded." This document contains some early experiences with congestion collapse and some initial thoughts on how to avoid it using congestion control in TCP. Furthermore, it defined an algorithm for efficient transmission of small packets that is today known as the Nagle Algorithm. This document points out several specification bugs in the US Military's MIL-STD-1778 document, which was intended as a successor to RFC 793 (see ). This serves to remind us of the difficulty in specification writing (even when we work from existing documents!).

Some documents in this section contain architectural guidance and concerns, while others specify TCP- and congestion-control-related mechanisms that are broadly applicable and have impacts on TCP's congestion control techniques. Some of these documents are direct products of the Internet Architecture Board (IAB), giving their guidance on specific aspects of congestion control in the Internet. This document describes the underlying principles of the Internet architecture. It provides guidelines for network systems design that have proven useful in the evolution of the Internet. This document motivates the use of end-to-end congestion control for preventing congestion collapse and providing fairness to TCP. Later work on TCP has included several more aggressive mechanisms than Reno TCP includes, and RFC 5033 (see ) provides additional guidance on use of such algorithms. The fundamental architectural discussion in RFC 2914 remains valid, regarding the standards process role in defining protocol aspects that are critical to performance and avoiding congestion collapse scenarios. This document updates RFC 1958 (see ) by outlining some philosophical guidelines for architects and designers of Internet backbone networks. The document describes the Simplicity Principle, which states that complexity is the primary impediment to efficient scaling. This document discusses some of the issues in defining alternate semantics for the ECN field, and specifies requirements for a safe co-existence with routers that do not understand the defined alternate semantics. Abstract: "This document outlines architectural guidelines for the development of a Multipath Transport Protocol, with references to how these architectural components come together in the development of a Multipath TCP (MPTCP) (see ). This document lists certain high-level design decisions that provide foundations for the design of the MPTCP protocol, based upon these architectural requirements"

As the internetworking field has explored wireless, satellite, cellular telephone, and other kinds of link-layer technologies, a large body of work has built up on enhancing TCP performance for such links. The RFCs listed in this section describe some of these more challenging network environments and how TCP interacts with them. From abstract: "While TCP works over satellite channels there are several IETF standardized mechanisms that enable TCP to more effectively utilize the available capacity of the network path. This document outlines some of these TCP mitigations. At this time, all mitigations discussed in this document are IETF standards track mechanisms (or are compliant with IETF standards)." Several methods of improving TCP performance over long thin networks (i.e., networks with low bandwidth and high delay), such as geosynchronous satellite links, are discussed in this document . A particular set of TCP options is developed that should work well in such environments and be safe to use in the global Internet. The implications of such environments have been further discussed in RFC 3150 (see ) and RFC 3155 (see ), and these documents should be preferred where there is overlap between them and RFC 2757 (see ). This document discusses the advantages and disadvantages of several different experimental means of improving TCP performance over long-delay or error-prone paths. These include T/TCP, larger initial windows, byte counting, delayed acknowledgments, slow start thresholds, NewReno and SACK-based loss recovery, FACK , ECN, various corruption-detection mechanisms, congestion avoidance changes for fairness, use of multiple parallel flows, pacing, header compression, state sharing, and ACK congestion control, filtering, and reconstruction. Although RFC 2488 (see ) looks at standard extensions, this document focuses on more experimental means of performance enhancement. From abstract: "This document is a survey of Performance Enhancing Proxies (PEPs) often employed to improve degraded TCP performance caused by characteristics of specific link environments, for example, in satellite, wireless WAN, and wireless LAN environments. Different types of Performance Enhancing Proxies are described as well as the mechanisms used to improve performance." From abstract: "This document makes performance-related recommendations for users of network paths that traverse "very low bit-rate" links....This recommendation may be useful in any network where hosts can saturate available bandwidth, but the design space for this recommendation explicitly includes connections that traverse 56 Kb/second modem links or 4.8 Kb/second wireless access links - both of which are widely deployed." From abstract: "This document discusses the specific TCP mechanisms that are problematic in environments with high uncorrected error rates, and discusses what can be done to mitigate the problems without introducing intermediate devices into the connection." From abstract: "This document provides advice to the designers of digital communication equipment and link-layer protocols employing link-layer Automatic Repeat reQuest (ARQ) techniques. This document presumes that the designers wish to support Internet protocols, but may be unfamiliar with the architecture of the Internet and with the implications of their design choices for the performance and efficiency of Internet traffic carried over their links." From abstract: "This document describes TCP performance problems that arise because of asymmetric effects. These problems arise in several access networks, including bandwidth-asymmetric networks and packet radio subnetworks, for different underlying reasons. However, the end result on TCP performance is the same in both cases: performance often degrades significantly because of imperfection and variability in the ACK feedback from the receiver to the sender. The document details several mitigations to these effects, which have either been proposed or evaluated in the literature, or are currently deployed in networks." From abstract: "This document describes a profile for optimizing TCP to adapt so that it handles paths including second (2.5G) and third (3G) generation wireless networks." This document describes how TCP performance can be negatively affected by some particular lower-layer behaviors and provides guidance in designing lower-layer networks and protocols to be amicable to TCP. RFC 3366 (see ) specifically focuses on ARQ mechanisms, while RFC 3819 more widely covers additional aspects of the underlying layers

Documents in this section give general guidance for developing, analyzing, and evaluating TCP. Some of the documents discuss for example the properties of congestion control protocols that are "safe" for Internet deployment, as well as how to measure the properties of congestion control mechanisms and transport protocols. This document considers the evaluation of suggested congestion control algorithms that differ from the principles outlined in RFC 2914 (see ). It is useful for authors of such algorithms as well as for IETF members reviewing the associated documents. This document discusses metrics that needs to be considered when evaluating new or modified congestion control mechanisms for the Internet. Among others topics, the document discusses throughput, delay, loss rates, response times, fairness and robustness for challenging environments. This RFC summarizes the main open problems in the domain of Internet congestion control. As a good starting point for newcomers, the document describes several new challenges that are becoming important as the network grows, as well as some issues that have been known for many years. This document describes a threat analysis for Multipath TCP (MPTCP) (see ). The document discusses several types of attacks and provides recommendations for MPTCP designers how to create an MPTCP specification that is as secure as the current (single-path) TCP. From abstract: "This document describes a practical methodology for measuring end-to-end TCP throughput in a managed IP network. The goal is to provide a better indication in regard to user experience. In this framework, TCP and IP parameters are specified to optimize TCP throughput."

This document discusses on a high-level the realization of pre-emption in TCP. Abstract: "This memo discusses the TCP Maximum Segment Size Option and related topics. The purposes is to clarify some aspects of TCP and its interaction with IP. This memo is a clarification to the TCP specification, and contains information that may be considered as 'advice to implementers'." This document lists a number of implementation techniques for efficiently computing the Internet checksum (used by TCP). Incrementally updating the Internet checksum is useful to routers in updating IP checksums. Some middleboxes that alter TCP headers may also be able to update the TCP checksum incrementally. This document expands upon the explanation of the incremental update procedure in RFC 1071 (see ). This document describes the motivation for implementing the Internet checksum in hardware, rather than in software, and provides an implementation example. From abstract: "This memo catalogs a number of known TCP implementation problems. The goal is to improve conditions in the existing Internet by enhancing the quality of current TCP/IP implementations." From abstract: "This memo catalogs several known Transmission Control Protocol (TCP) implementation problems dealing with Path Maximum Transmission Unit Discovery (PMTUD), including the long-standing black hole problem, stretch acknowledgments (ACKs) due to confusion between Maximum Segment Size (MSS) and segment size, and MSS advertisement based on PMTU." This document is a plea that firewall vendors not send gratuitous TCP RST (Reset) packets when unassigned TCP header bits are used. This practice prevents desirable extension and evolution of the protocol and thus is potentially harmful to the future of the Internet. This document describes the de facto standard sockets API for programming with TCP. This API is implemented nearly ubiquitously in modern operating systems and programming languages. This document describes a number of simple and efficient methods for the selection of the client port number. It reduces the possibility of an attacker guessing the correct five-tuple (Protocol, Source/Destination Address, Source/Destination Port). This document describes the usage of the TCP Timestamps option (RFC XXXX, see ) to perform heuristics to determine whether or not to allow the creation of a new incarnation of a connection that is in the TIME-WAIT state. This document clarifies the actions that a TCP can be taken on connections that are experiencing the Zero Window Probe (ZWP) condition. This document characterizes the impact that Multipath TCP (MPTCP) (see ) may have on applications. It further discusses compatibility issues of MPTCP in combination with non-MPTCP-aware applications. Finally, it describes a basic API that is a simple extension of TCP's interface for MPTCP-aware applications.

This document is an extremely brief overview of the TCP/IP protocol suite as a whole. It gives some explanation as to how and where TCP fits in. A few of the tools that this document describes are still maintained and in use today; for example, ttcp and tcpdump. However, many of the tools described do not relate specifically to TCP and are no longer used or easily available. This document describes a number of TCP packet generation and analysis tools. Although some of these tools are no longer readily available or widely used, for the most part they are still relevant and usable. This document provides an overview of RFCs related to congestion control that have been published so far. The focus of the document is on end-host-based congestion control.

The first MIB module defined for use with Simple Network Management Protocol (SNMP) was a single monolithic MIB module, called MIB-I, defined in RFC 1156. This evolved over time to the MIB-II specification in RFC 1213, which obsoletes RFC 1156. It then became apparent that having a single monolithic MIB module was not scalable, given the number and breadth of MIB data definitions that needed to be included. Thus, additional MIB modules were defined, and those parts of MIB-II that needed to evolve were split off. Eventually, the remaining parts of MIB-II were also split off, the TCP-specific part being documented in RFC 2012. RFC 2012 was obsoleted by RFC 4022, which is the primary TCP MIB document today. For current TCP implementers, RFC 4022 should be supported. This document describes the required MIB fields for TCP implementations with minor corrections and no technical changes from RFC 1066, which it obsoletes. This is the standards track document for MIB-I. This document describes the second version of the MIB in a monolithic form. It is the immediate successor of RFC 1158, with minor modifications. It obsoletes the MIB-I, defined in RFC 1156 (see ). In an update to RFC 1213 (see ), this document defines the TCP MIB by splitting out the TCP-specific portions. It is now obsoleted by RFC 4022 (see ). This document augments RFC 2012 (see ) by adding an IPv6-specific connection table. The rest of RFC 2012 holds for any IP version. RFC 2452 is now obsoleted by RFC 4022 (see ). Although it is a standards track document, RFC 2452 is considered a historic mistake by the MIB community, as it is based on the idea of parallel IPv4 and IPv6 structures. Although IPv6 requires new structures, the community has decided to define a single generic structure for both IPv4 and IPv6. This will aid in definition, implementation, and transition between IPv4 and IPv6. This document obsoletes RFC 2012 (see ) and RFC 2452 (see ) and specifies the current standard for the TCP MIB that should be deployed. This document describes extended performance statistics for TCP. They are designed to use TCP's ideal vantage point to diagnose performance problems in both the network and the application.

This document presents a field report about the deployment of a very early version of TCP, the so-called INWN #39 protocol, which is originally described by Cerf and Kahn in INWG Note #39 to use a PDP-11 line printer via the ARPANET. This document is a status report about experiments concerning the TCP retransmission timeout calculation and also provides advices for implementers. This document points out a problem with acting on received reset segments while one is in the TIME-WAIT state. The main recommendation is that hosts in TIME-WAIT ignore resets. This recommendation might not currently be widely implemented. This document presents results of some simulations using TCP initial windows greater than 1 segment. The analysis indicates that user-perceived performance can be improved by increasing the initial window to 3 segments. This document uses simulation results to clear up some concerns about using an initial window of 4 segments when the network path has less provisioning. This document describes experimental results that show some improvements to the performance of both short- and long-lived connections due to ECN.

There are a few important implementation tactics for the TCP that have not yet been described in any RFC. Although this roadmap is primarily concerned with mapping the TCP RFCs, this section is included because an implementer needs to be aware of these important issues. Header prediction is a trick to speed up the processing of segments. Van Jacobson and Mike Karels developed the technique in the late 1980s. The basic idea is that some processing time can be saved when most of a segment's fields can be predicted from previous segments. A good description of this was sent to the TCP-IP mailing list by Van Jacobson on March 9, 1988: "Quite a bit of the speedup comes from an algorithm that we ('we' refers to collaborator Mike Karels and myself) are calling "header prediction". The idea is that if you're in the middle of a bulk data transfer and have just seen a packet, you know what the next packet is going to look like: It will look just like the current packet with either the sequence number or ack number updated (depending on whether you're the sender or receiver). Combining this with the "Use hints" epigram from Butler Lampson's classic "Epigrams for System Designers", you start to think of the tcp state (rcv.nxt, snd.una, etc.) as "hints" about what the next packet should look like. If you arrange those "hints" so they match the layout of a tcp packet header, it takes a single 14-byte compare to see if your prediction is correct (3 longword compares to pick up the send & ack sequence numbers, header length, flags and window, plus a short compare on the length). If the prediction is correct, there's a single test on the length to see if you're the sender or receiver followed by the appropriate processing. E.g., if the length is non-zero (you're the receiver), checksum and append the data to the socket buffer then wake any process that's sleeping on the buffer. Update rcv.nxt by the length of this packet (this updates your "prediction" of the next packet). Check if you can handle another packet the same size as the current one. If not, set one of the unused flag bits in your header prediction to guarantee that the prediction will fail on the next packet and force you to go through full protocol processing. Otherwise, you're done with this packet. So, the *total* tcp protocol processing, exclusive of checksumming, is on the order of 6 compares and an add." FACK includes an alternate algorithm for triggering fast retransmit , based on the extent of the SACK scoreboard. Its goal is to trigger fast retransmit as soon as the receiver's reassembly queue is larger than the duplicate ACK threshold, as indicated by the difference between the forward most SACK block edge and SND.UNA. This algorithm quickly and reliably triggers fast retransmit in the presence of burst losses -- often on the first SACK following such a loss. Such a threshold based algorithm also triggers fast retransmit immediately in the presence of any reordering with extent greater than the duplicate ACK threshold. FACK is implemented in Linux and turned on per default. In the last decade significant research effort has been put into experimental TCP congestion control modifications for obtaining high throughput with reduced startup and recovery times. Only few RFCs have been published on some of these modifications, including HighSpeed TCP (see ), Limited Slow-Start (see ), and Quick-Start (see ), but high-rate congestion control mechanisms are still considered an open issue in congestion control research. Some other schemes have been published as Internet-Drafts, e.g. CUBIC (the standard TCP congestion control algorithm in Linux), Compound TCP , and H-TCP or have been discussed a little by the IETF, but much of the work in this area has not been adopted within the IETF yet, so the majority of this work is outside the RFC series and may be discussed in other products of the IRTF Internet Congestion Control Research Group (ICCRG).

This document introduces no new security considerations. Each RFC listed in this document attempts to address the security considerations of the specification it contains.

This document contains no IANA considerations.

This document grew out of a discussion on the end2end-interest mailing list, the public list of the End-to-End Research Group of the IRTF, and continued development under the IETF's TCP Maintenance and Minor Extensions (TCPM) working group. We thank Mark Allman, Yuchung Cheng, Ted Faber, Fairhurst, Sally Floyd, Janardhan Iyengar, Reiner Ludwig, Pekka Savola, and Joe Touch for their contributions, in particular. Keith McCloghrie provided some useful notes and clarification on the various MIB-related RFCs.