<?xmlversion="1.0" encoding="US-ASCII"?> <!-- Convert to HTML and Text with xml2rfc: http://xml2rfc.ietf.org. -->version='1.0' encoding='utf-8'?> <!DOCTYPE rfc SYSTEM"rfc2629.dtd" [ <!ENTITY RFC5533 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.5533.xml"> <!ENTITY RFC5062 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.5062.xml"> <!ENTITY RFC5061 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.5061.xml"> <!ENTITY RFC4960 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.4960.xml"> <!ENTITY RFC4987 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.4987.xml"> <!ENTITY RFC6234 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6234.xml"> <!ENTITY RFC4086 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.4086.xml"> <!ENTITY RFC5681 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.5681.xml"> <!ENTITY RFC2119 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2119.xml"> <!ENTITY RFC2992 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2992.xml"> <!ENTITY RFC2979 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2979.xml"> <!ENTITY RFC2104 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2104.xml"> <!ENTITY RFC2018 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2018.xml"> <!ENTITY RFC1918 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.1918.xml"> <!ENTITY RFC0793 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.0793.xml"> <!ENTITY RFC7323 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.7323.xml"> <!ENTITY RFC1122 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.1122.xml"> <!ENTITY RFC3135 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.3135.xml"> <!ENTITY RFC3022 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.3022.xml"> <!ENTITY RFC6181 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6181.xml"> <!ENTITY RFC6182 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6182.xml"> <!ENTITY RFC6356 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6356.xml"> <!ENTITY RFC6555 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6555.xml"> <!ENTITY RFC8126 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.8126.xml"> <!ENTITY RFC6897 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6897.xml"> <!ENTITY RFC6528 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6528.xml"> <!ENTITY RFC5961 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.5961.xml"> <!ENTITY RFC7413 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.7413.xml"> <!ENTITY RFC7430 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.7430.xml"> <!ENTITY RFC8174 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.8174.xml"> <!ENTITY RFC8041 SYSTEM "https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.8041.xml"> ]> <?xml-stylesheet type='text/xsl' href='rfc2629.xslt' ?> <?rfc strict="no" ?> <?rfc toc="yes"?> <?rfc tocdepth="4"?> <?rfc symrefs="yes"?> <?rfc sortrefs="yes" ?> <?rfc compact="yes" ?> <?rfc subcompact="no" ?> <?rfc rfcedstyle="yes"?>"rfc2629-xhtml.ent"> <rfc xmlns:xi="http://www.w3.org/2001/XInclude" submissionType="IETF" category="std" consensus="true" docName="draft-ietf-mptcp-rfc6824bis-18" number="8684" ipr="trust200902"obsoletes="6824">obsoletes="6824" updates="" xml:lang="en" tocInclude="true" symRefs="true" sortRefs="true" version="3"> <!-- xml2rfc v2v3 conversion 2.27.0 --> <front> <title abbrev="Multipath TCP">TCP Extensions for Multipath Operation with Multiple Addresses</title> <seriesInfo name="RFC" value="8684"/> <author fullname="Alan Ford" initials="A." surname="Ford"> <organization>Pexip</organization> <address><!-- <postal> <street>Beech Court</street> <city>Hurst</city> <region>Berkshire</region> <code>RG10 0RQ</code> <country>UK</country> </postal> --><email>alan.ford@gmail.com</email> </address> </author> <author fullname="Costin Raiciu" initials="C." surname="Raiciu"> <organization abbrev="U.PolitechnicaPolitehnica of Bucharest">University Politehnica of Bucharest</organization> <address> <postal> <street>Splaiul Independentei 313</street> <city>Bucharest</city> <country>Romania</country> </postal> <email>costin.raiciu@cs.pub.ro</email> </address> </author> <author fullname="Mark Handley" initials="M." surname="Handley"> <organization abbrev="U. College London">University College London</organization> <address> <postal> <street>Gower Street</street> <city>London</city> <code>WC1E 6BT</code><country>UK</country><country>United Kingdom</country> </postal> <email>m.handley@cs.ucl.ac.uk</email> </address> </author> <author fullname="Olivier Bonaventure" initials="O." surname="Bonaventure"> <organization abbrev="U. catholique deLouvain">UniversitéLouvain" ascii="Universite catholique de Louvain">Université catholique de Louvain</organization> <address> <postal> <street>Pl. Ste Barbe, 2</street> <code>1348</code> <city>Louvain-la-Neuve</city> <country>Belgium</country> </postal> <email>olivier.bonaventure@uclouvain.be</email> </address> </author> <author fullname="Christoph Paasch" initials="C." surname="Paasch"> <organization abbrev="Apple, Inc.">Apple, Inc.</organization> <address> <postal><street></street><street/> <city>Cupertino</city><country>US</country><region>CA</region> <country>United States of America</country> </postal> <email>cpaasch@apple.com</email> </address> </author> <dateyear="2019" /> <area>General</area> <workgroup>Internet Engineering Task Force</workgroup> <keyword>tcp extensions multipath multihomed subflow</keyword>year="2020" month="March"/> <keyword>tcp</keyword> <keyword>extensions</keyword> <keyword>multipath</keyword> <keyword>multihomed</keyword> <keyword>subflow</keyword> <abstract> <t>TCP/IP communication is currently restricted to a single path per connection, yet multiple paths often exist between peers. The simultaneous use of these multiple paths for a TCP/IP session would improve resource usage within the networkand, thus,and thus improve user experience through higher throughput and improved resilience to network failure.</t> <t>Multipath TCP provides the ability to simultaneously use multiple paths between peers. This document presents a set of extensions to traditional TCP to support multipath operation. The protocol offers the same type of service to applications as TCP (i.e., a reliable bytestream), and it provides the components necessary to establish and use multiple TCP flows across potentially disjoint paths.</t> <t>This document specifies v1 of Multipath TCP, obsoleting v0 as specified inRFC6824,RFC 6824, through clarifications and modifications primarily driven by deployment experience.</t> </abstract> </front> <middle> <sectiontitle="Introduction" anchor="sec_intro">anchor="sec_intro" numbered="true" toc="default"> <name>Introduction</name> <t>Multipath TCP (MPTCP) is a set of extensions to regular TCP <xreftarget="RFC0793"/>target="RFC0793" format="default"/> to provide a Multipath TCP service <xreftarget="RFC6182"/> service,target="RFC6182" format="default"/>, which enables a transport connection to operate across multiple paths simultaneously. This document presents the protocol changes required to add multipath capability toTCP;TCP -- specifically, those for signaling and setting up multiple paths ("subflows"), managing these subflows, reassembly of data, and termination of sessions. This is not the only information required to create a Multipath TCP implementation, however. This document is complemented by three others:<list style="symbols"> <t>Architecture <xref target="RFC6182"/>,</t> <ul spacing="normal"> <li><xref target="RFC6182" format="default"/> (MPTCP architecture), which explains the motivations behind Multipath TCP, contains a discussion of high-level design decisions on which this design is based, and provides an explanation of a functional separation through which an extensible MPTCP implementation can bedeveloped.</t> <t>Congestion control <xref target="RFC6356"/>developed.</li> <li><xref target="RFC6356" format="default"/> (congestion control), which presents a safe congestion control algorithm for coupling the behavior of the multiple paths in order to "do no harm" to other networkusers.</t> <t>Application considerations <xref target="RFC6897"/>users.</li> <li><xref target="RFC6897" format="default"/> (application considerations), which discusses what impact MPTCP will have on applications, what applications will want to do with MPTCP, and as a consequence of these factors, what API extensions an MPTCP implementation shouldpresent.</t> </list>present.</li> </ul> <t> This documentis an update to, and obsoletes,obsoletes the v0 specification of Multipath TCP(RFC6824).<xref target="RFC6824"/>. This document specifies MPTCP v1, which is not backward compatible with MPTCP v0. This document additionally defines version negotiation procedures for implementations that support both versions. </t> <sectiontitle="Design Assumptions" anchor="sec_assum">anchor="sec_assum" numbered="true" toc="default"> <name>Design Assumptions</name> <t>In order to limit the potentially huge design space, themptcp working groupMPTCP Working Group imposed two key constraints on the Multipath TCP design presented in this document:<list style="symbols"> <t>It</t> <ul spacing="normal"> <li>It must bebackwards-compatiblebackward compatible with current, regular TCP, to increase its chances ofdeployment.</t> <t>Itdeployment.</li> <li>It can be assumed that one or both hosts are multihomed andmultiaddressed.</t> </list> </t>multiaddressed.</li> </ul> <t>To simplify the design, we assume that the presence of multiple addresses at a host is sufficient to indicate the existence of multiple paths. These paths need not be entirely disjoint: they may share one or many routers between them. Even in such a situation, making use of multiple paths is beneficial, improving resource utilization and resilience to a subset of node failures. The congestion controlalgorithmsalgorithm defined in <xreftarget="RFC6356"/> ensure thistarget="RFC6356" format="default"/> ensures that the use of multiple paths does not act detrimentally. Furthermore, there may be some scenarios where different TCP ports on a single host can provide disjoint paths (such as through certain Equal-Cost Multipath (ECMP) implementations <xreftarget="RFC2992"/>),target="RFC2992" format="default"/>), and so the MPTCP design also supports the use of ports in path identifiers.</t> <t>There are three aspects to thebackwards-compatibilitybackward compatibility listed above (discussed in more detail in <xreftarget="RFC6182"/>): <list style="hanging"> <t hangText="External Constraints:">target="RFC6182" format="default"/>): </t> <dl newline="false" spacing="normal" indent="3"> <dt>External Constraints:</dt> <dd> The protocol must function through the vast majority of existing middleboxes such as NATs, firewalls, and proxies, and as such must resemble existing TCP as far as possible on the wire. Furthermore, the protocol must not assume that the segments it sends on the wire arrive unmodified at the destination: they may be split or coalesced; TCP options may be removed or duplicated.</t> <t hangText="Application Constraints:"></dd> <dt>Application Constraints:</dt> <dd> The protocol must be usable with no change to existing applications that use the common TCP API (although it is reasonable that not all features would be available to such legacy applications). Furthermore, the protocol must provide the same service model as regular TCP to theapplication.</t> <t hangText="Fallback:">application.</dd> <dt>Fallback:</dt> <dd> The protocol should be able to fall back to standard TCP with no interference from the user, to be able to communicate with legacyhosts.</t> </list> </t>hosts.</dd> </dl> <t>The complementary application considerations document <xreftarget="RFC6897"/>target="RFC6897" format="default"/> discusses the necessary features of an API to providebackwards-compatibility,backward compatibility, as well as API extensions to convey the behavior of MPTCP at a level of control and information equivalent to that available with regular, single-path TCP.</t> <t>Further discussion of the design constraints and associated design decisionsareis given in the MPTCPArchitecturearchitecture document <xreftarget="RFC6182"/>target="RFC6182" format="default"/> and in <xreftarget="howhard"/>.</t>target="howhard" format="default"/>.</t> </section> <sectiontitle="Multipathanchor="sec_layers" numbered="true" toc="default"> <name>Multipath TCP in the NetworkingStack" anchor="sec_layers">Stack</name> <t>MPTCP operates at the transport layer and aims to be transparent to both higher and lower layers. It is a set of additional features on top of standard TCP; <xref target="fig_arch"/>format="default"/> illustrates this layering. MPTCP is designed to be usable by legacy applications with no changes; detailed discussion of its interactions with applications is given in <xreftarget="RFC6897"/>.</t>target="RFC6897" format="default"/>.</t> <figurealign="center" anchor="fig_arch" title="Comparisonanchor="fig_arch"> <name>Comparison of Standard TCP and MPTCP ProtocolStacks">Stacks</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ +-------------------------------+ | Application | +---------------+ +-------------------------------+ | Application | | MPTCP | +---------------+ + - - - - - - - + - - - - - - - + | TCP | | Subflow (TCP) | Subflow (TCP) | +---------------+ +-------------------------------+ | IP | | IP | IP | +---------------+ +-------------------------------+ ]]></artwork> </figure> </section> <sectiontitle="Terminology">numbered="true" toc="default"> <name>Terminology</name> <t>This document makes use of a number of terms that are eitherMPTCP-specificMPTCP specific or have defined meaning in the context of MPTCP, as follows:<list style="hanging"> <t hangText="Path:"></t> <dl newline="false" spacing="normal" indent="3"> <dt>Path:</dt> <dd> A sequence of links between a sender and a receiver, defined in this context by a 4-tuple of source and destinationaddress/port pairs.</t> <t hangText="Subflow:">address&wj;/port pairs.</dd> <dt>Subflow:</dt> <dd> A flow of TCP segments operating over an individual path, which forms part of a larger MPTCP connection. A subflow is started and terminatedsimilarsimilarly to a regular TCPconnection.</t> <t hangText="(MPTCP) Connection:">connection.</dd> <dt>(MPTCP) Connection:</dt> <dd> A set of one or more subflows, over which an application can communicate between two hosts. There is aone-to-oneone‑to‑one mapping between a connection and an applicationsocket.</t> <t hangText="Data-level:">socket.</dd> <dt>Data-level:</dt> <dd> The payload data is nominally transferred over a connection, which in turn is transported over subflows. Thus, the term "data-level" is synonymous with"connection level","connection-level", in contrast to "subflow-level", which refers to properties of an individualsubflow.</t> <t hangText="Token:">subflow.</dd> <dt>Token:</dt> <dd> A locally unique identifier given to a multipath connection by a host. May also be referred to as a "ConnectionID".</t> <t hangText="Host:">ID".</dd> <dt>Host:</dt> <dd> An end host operating an MPTCP implementation, and either initiating or accepting an MPTCPconnection.</t> </list>connection.</dd> </dl> <t> In addition to these terms, note that MPTCP's interpretation of, and effect on, regular single-path TCP semantics are discussed in <xreftarget="sec_semantics"/>.</t>target="sec_semantics" format="default"/>.</t> </section> <sectiontitle="MPTCP Concept" anchor="sec_operation">anchor="sec_operation" numbered="true" toc="default"> <name>MPTCP Concept</name> <t>This section provides a high-level summary of normal operation ofMPTCP, andMPTCP; this type of scenario is illustratedby the scenario shownin <xreftarget="fig_scenario"/>.target="fig_scenario" format="default"/>. A detailed description ofoperationhow MPTCP operates is given in <xreftarget="sec_protocol"/>. <list style="symbols"> <t>Totarget="sec_protocol" format="default"/>. </t> <figure anchor="fig_scenario"> <name>Example MPTCP Usage Scenario</name> <artwork align="left" name="" type="" alt=""><![CDATA[ Host A Host B ------------------------ ------------------------ Address A1 Address A2 Address B1 Address B2 ---------- ---------- ---------- ---------- | | | | | (initial connection setup) | | |----------------------------------->| | |<-----------------------------------| | | | | | | (additional subflow setup) | | |--------------------->| | | |<---------------------| | | | | | | | | | ]]></artwork> </figure> <ul spacing="normal"> <li>To a non-MPTCP-aware application, MPTCP will behave the same as normal TCP. Extended APIs could provide additional control to MPTCP-aware applications <xreftarget="RFC6897"/>.target="RFC6897" format="default"/>. An application begins by opening a TCP socket in the normal way. MPTCP signaling and operation are handled by the MPTCP implementation.</t> <t>An</li> <li>An MPTCP connection begins similarly to a regular TCP connection. This is illustrated in <xreftarget="fig_scenario"/>target="fig_scenario" format="default"/>, where an MPTCP connection is established between addresses A1 and B1 on Hosts A and B,respectively.</t> <t>Ifrespectively.</li> <li>If extra paths are available, additional TCP sessions (termed MPTCP "subflows") are created on thesepaths,paths and are combined with the existing session, which continues to appear as a single connection to the applications at both ends. The creation of the additional TCP session is illustrated between Address A2 on Host A and Address B1 on HostB.</t> <t>MPTCPB.</li> <li>MPTCP identifies multiple paths by the presence of multiple addresses at hosts. Combinations of these multiple addresses equate to the additional paths. In the example, other potential paths that could be set up are A1<->B2 and A2<->B2. Although this additional session is shown as being initiated from A2, it could equally have been initiated from B1 orB2.</t> <t>TheB2.</li> <li>The discovery and setup of additional subflows will be achieved through a path management method; this document describes a mechanism by which a host can initiate new subflows by using its own additionaladdresses,addresses or by signaling its available addresses to the otherhost.</t> <t>MPTCPhost.</li> <li>MPTCP adds connection-level sequence numbers to allow the reassembly of segments arriving on multiple subflows with differing network delays.</t> <t>Subflows</li> <li>Subflows are terminated as regular TCP connections, with afour-wayfour‑way FIN handshake. The MPTCP connection is terminated by a connection-levelFIN.</t> </list>FIN.</li> </ul> </section> <section numbered="true" toc="default"> <name>Requirements Language</name> <t> The key words "<bcp14>MUST</bcp14>", "<bcp14>MUST NOT</bcp14>", "<bcp14>REQUIRED</bcp14>", "<bcp14>SHALL</bcp14>", "<bcp14>SHALL NOT</bcp14>", "<bcp14>SHOULD</bcp14>", "<bcp14>SHOULD NOT</bcp14>", "<bcp14>RECOMMENDED</bcp14>", "<bcp14>NOT RECOMMENDED</bcp14>", "<bcp14>MAY</bcp14>", and "<bcp14>OPTIONAL</bcp14>" in this document are to be interpreted as described in BCP 14 <xref target="RFC2119"/> <xref target="RFC8174"/> when, and only when, they appear in all capitals, as shown here. </t><?rfc needLines='17'?> <figure align="center" anchor="fig_scenario" title="Example MPTCP Usage Scenario"> <artwork align="left"><![CDATA[ Host A Host B ------------------------ ------------------------ Address A1 Address A2 Address B1 Address B2 ---------- ---------- ---------- ---------- | | | | | (initial connection setup) | | |----------------------------------->| | |<-----------------------------------| | | | | | | (additional subflow setup) | | |--------------------->| | | |<---------------------| | | | | | | | | | ]]></artwork> </figure> </section> <section title="Requirements Language"> <t>The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 <xref target="RFC2119"/> <xref target="RFC8174"/> when, and only when, they appear in all capitals, as shown here.</t> </section> </section> <section title="Operation Overview" anchor="sec_overview"> <t>This section presents a single description of common</section> </section> <section anchor="sec_overview" numbered="true" toc="default"> <name>Operation Overview</name> <t>This section presents a single description of common MPTCP operation, with reference to the protocol operation. This is a high-level overview of the key functions; the full specification follows in <xreftarget="sec_protocol"/>.target="sec_protocol" format="default"/>. Extensibility and negotiated features are not discussed here. Considerable reference is made to symbolic names of MPTCP options throughout this section -- these are subtypes of theIANA-assignedIANA‑assigned MPTCP option (see <xreftarget="IANA"/>),target="IANA" format="default"/>), and their formats are defined in the detailed protocol specificationthat followsprovided in <xreftarget="sec_protocol"/>.</t>target="sec_protocol" format="default"/>.</t> <t>A Multipath TCP connection provides a bidirectional bytestream between two hosts communicating like normal TCPand, thus,and thus does not require any change to the applications. However, Multipath TCP enables the hosts to use different paths with different IP addresses to exchange packets belonging to the MPTCP connection. A Multipath TCP connection appears like a normal TCP connection to an application. However, to the network layer, each MPTCP subflow looks like a regular TCP flow whose segments carry a new TCP option type. Multipath TCP manages the creation, removal, and utilization of these subflows to send data. The number of subflows that are managed within a Multipath TCP connection is notfixedfixed, and it can fluctuate during the lifetime of the Multipath TCP connection.</t> <t>All MPTCP operations are signaled with a TCP option -- a single numerical type for MPTCP, with"sub-types""subtypes" for each MPTCP message. What follows is a summary of the purpose and rationale of these messages.</t> <sectiontitle="Initiatingnumbered="true" toc="default"> <name>Initiating an MPTCPConnection">Connection</name> <t>This is the same signaling as for initiating a normal TCP connection, but the SYN, SYN/ACK, and initial ACK (and data) packets also carry the MP_CAPABLE option. This option has a variable length and serves multiple purposes. Firstly, it verifies whether the remote host supports Multipath TCP; secondly, this option allows the hosts to exchange some information to authenticate the establishment of additional subflows. Further details are given in <xreftarget="sec_init"/>.</t> <figure><artwork align="left"><![CDATA[target="sec_init" format="default"/>.</t> <artwork align="left" name="" type="" alt=""><![CDATA[ Host A Host B ------ ------ MP_CAPABLE -> [flags] <- MP_CAPABLE [B's key, flags] ACK + MP_CAPABLE (+ data) -> [A's key, B's key, flags, (data-level details)]]]></artwork></figure>]]></artwork> <t>Retransmission of the ACK + MP_CAPABLE can occur if it is not known if it has been received. The following diagrams show all possible exchanges for the initial subflow setup to ensure this reliability.</t><figure><artwork align="left"><![CDATA[<artwork align="left" name="" type="" alt=""><![CDATA[ Host A (with data to send immediately) Host B ------ ------ MP_CAPABLE -> [flags] <- MP_CAPABLE [B's key, flags] ACK + MP_CAPABLE + data -> [A's key, B's key, flags, data-level details] Host A (with data to send later) Host B ------ ------ MP_CAPABLE -> [flags] <- MP_CAPABLE [B's key, flags] ACK + MP_CAPABLE -> [A's key, B's key, flags] ACK + MP_CAPABLE + data -> [A's key, B's key, flags, data-level details] Host A Host B (sending first) ------ ------ MP_CAPABLE -> [flags] <- MP_CAPABLE [B's key, flags] ACK + MP_CAPABLE -> [A's key, B's key, flags] <- ACK + DSS + data [data-level details]]]></artwork></figure>]]></artwork> </section> <sectiontitle="Associatingnumbered="true" toc="default"> <name>Associating a New Subflow with an Existing MPTCPConnection">Connection</name> <t>The exchange of keys in the MP_CAPABLE handshake provides material that can be used to authenticate the endpoints when new subflows will be set up. Additional subflows begin in the same way as initiating a normal TCP connection, but the SYN, SYN/ACK, and ACK packets also carry the MP_JOIN option. </t> <t>Host A initiates a new subflow between one of its addresses and one of Host B's addresses. The token -- generated from the key -- is used to identify which MPTCP connection it is joining, and theHMACHash‑based Message Authentication Code (HMAC) is used for authentication. TheHash-based Message Authentication Code (HMAC)HMAC uses the keys exchanged in the MP_CAPABLEhandshake,handshake and the random numbers (nonces) exchanged in these MP_JOIN options. MP_JOIN also contains flags and an Address ID that can be used to refer to the source address without the sender needing to know if it has been changed by a NAT. Further details are given in <xreftarget="sec_join"/>.</t> <figure><artwork align="left"><![CDATA[target="sec_join" format="default"/>.</t> <artwork align="left" name="" type="" alt=""><![CDATA[ Host A Host B ------ ------ MP_JOIN -> [B's token, A's nonce, A's Address ID, flags] <- MP_JOIN [B's HMAC, B's nonce, B's Address ID, flags] ACK + MP_JOIN -> [A's HMAC] <- ACK]]></artwork></figure>]]></artwork> </section> <sectiontitle="Informingnumbered="true" toc="default"> <name>Informing the Other Host about Another PotentialAddress">Address</name> <t>The set of IP addresses associated to a multihomed host may change during the lifetime of an MPTCP connection. MPTCP supports the addition and removal of addresses on a host both implicitly and explicitly. If Host A has established a subflow starting ataddress/portaddress&wj;/port pair IP#-A1 and wants to open a second subflow starting ataddress/portaddress&wj;/port pair IP#-A2, it simply initiates the establishment of the subflow as explained above. The remote host will then be implicitly informed about the new address.</t> <t>In some circumstances, a host may want to advertise to the remote host the availability of an address without establishing a newsubflow,subflow -- for example, when a NAT prevents setup in one direction.In theIn the example below, Host A informs Host B about its alternativeIP address/portIP address&wj;/port pair (IP#-A2). Host B may later send an MP_JOIN to this new address. The ADD_ADDR option containsaan HMAC to authenticate the address as having been sent from the originator of the connection. The receiver of this option echoes it back to the client to indicate successful receipt. Further details are given in <xreftarget="sec_add_address"/>.</t> <figure><artwork align="left"><![CDATA[target="sec_add_address" format="default"/>.</t> <artwork align="left" name="" type="" alt=""><![CDATA[ Host A Host B ------ ------ ADD_ADDR -> [Echo-flag=0, IP#-A2, IP#-A2's Address ID, HMAC of IP#-A2] <- ADD_ADDR [Echo-flag=1, IP#-A2, IP#-A2's Address ID, HMAC of IP#-A2]]]></artwork></figure>]]></artwork> <t>There is a corresponding signal for address removal, making use of the Address ID that is signaled in theadd addressADD_ADDR handshake. Further details are given in <xreftarget="sec_remove_addr"/>.</t> <figure><artwork align="left"><![CDATA[target="sec_remove_addr" format="default"/>.</t> <artwork align="left" name="" type="" alt=""><![CDATA[ Host A Host B ------ ------ REMOVE_ADDR -> [IP#-A2's Address ID]]]></artwork></figure>]]></artwork> </section> <sectiontitle="Datanumbered="true" toc="default"> <name>Data Transfer UsingMPTCP">MPTCP</name> <t>To ensure reliable, in-order delivery of data over subflows that may appear and disappear at any time, MPTCP uses a 64-bitdata sequence numberData Sequence Number (DSN) to number all data sent over the MPTCP connection. Each subflow has its own 32-bit sequence number space,utilisingutilizing the regular TCP sequence number header, and an MPTCP option maps the subflow sequence space to the data sequence space. In this way, data can be retransmitted on different subflows (mapped to the same DSN) in the event of failure.</t> <t>The Data Sequence Signal (DSS) carries the Data Sequence Mapping. The Data Sequence Mapping consists of the subflow sequence number, data sequence number, and length for which this mapping is valid. This option can also carry a connection-level acknowledgment (the "Data ACK") for the received DSN.</t> <t>With MPTCP, all subflows share the same receive buffer and advertise the same receive window. There are two levels of acknowledgment in MPTCP. Regular TCP acknowledgments are used on each subflow to acknowledge the reception of the segments sent over the subflow independently of their DSN. In addition, there are connection-level acknowledgments for the data sequence space. These acknowledgments track the advancement of the bytestream and slide thereceivingreceive window.</t> <t>Further details are given in <xreftarget="sec_generalop"/>.</t> <figure><artwork align="left"><![CDATA[target="sec_generalop" format="default"/>.</t> <artwork align="left" name="" type="" alt=""><![CDATA[ Host A Host B ------ ------ DSS -> [Data Sequence Mapping] [Data ACK] [Checksum]]]></artwork></figure>]]></artwork> </section> <sectiontitle="Requestingnumbered="true" toc="default"> <name>Requesting a Change in a Path'sPriority">Priority</name> <t>Hosts can indicate at initial subflow setup whether they wish the subflow to be used as a regular or backup path -- a backup path only being used if there are no regular paths available. During a connection, Host A can request a change in the priority of a subflow through the MP_PRIO signal to Host B. Further details are given in <xreftarget="sec_policy"/>.</t> <figure><artwork align="left"><![CDATA[target="sec_policy" format="default"/>.</t> <artwork align="left" name="" type="" alt=""><![CDATA[ Host A Host B ------ ------ MP_PRIO ->]]></artwork></figure>]]></artwork> </section> <sectiontitle="Closingnumbered="true" toc="default"> <name>Closing an MPTCPConnection">Connection</name> <t>When a host wants to close an existingsubflow,subflow but not the whole connection, it can initiate a regular TCP FIN/ACK exchange.</t> <t>When Host A wants to inform Host B that it has no more data to send, it signals this "Data FIN" as part of theData Sequence SignalDSS (see above). It has the same semantics and behavior as a regular TCP FIN, but at the connection level. Once all the data on the MPTCP connection has been successfully received,thenthis message is acknowledged at the connection level with a Data ACK. Further details are given in <xreftarget="sec_close"/>.</t> <figure><artwork align="left"><![CDATA[target="sec_close" format="default"/>.</t> <artwork align="left" name="" type="" alt=""><![CDATA[ Host A Host B ------ ------ DSS -> [Data FIN] <- DSS [Data ACK]]]></artwork></figure>]]></artwork> <t>There is an additional method of connection closure, referred to as "Fast Close", which is analogous to closing a single-path TCP connection with a RST signal. The MP_FASTCLOSE signal is used to indicate to the peer that the connection will be abruptly closed and no data will be accepted anymore. This can be used on an ACK(ensuring(which ensures reliability of thesignal),signal) or a RST (whichisdoes not). Both examples are shown in the following diagrams. Further details are given in <xreftarget="sec_fastclose"/>.</t> <figure><artwork align="left"><![CDATA[target="sec_fastclose" format="default"/>.</t> <artwork align="left" name="" type="" alt=""><![CDATA[ Host A Host B ------ ------ ACK + MP_FASTCLOSE -> [B's key] [RST on all other subflows] -> <- [RST on all subflows] Host A Host B ------ ------ RST + MP_FASTCLOSE -> [B's key] [on all subflows] <- [RST on all subflows]]]></artwork></figure>]]></artwork> </section> <sectiontitle="Notable Features">numbered="true" toc="default"> <name>Notable Features</name> <t>It is worth highlighting that MPTCP's signaling has been designed with several key requirements in mind:<list style="symbols"> <t>To</t> <ul spacing="normal"> <li>To cope with NATs on the path, addresses are referred to by Address IDs, in case the IP packet's source address gets changed by a NAT. Setting up a new TCP flow is not possible if the receiver of the SYN is behind a NAT; to allow subflows to be created when either end is behind a NAT, MPTCP uses the ADD_ADDR message.</t> <t>MPTCP</li> <li>MPTCP falls back to ordinary TCP if MPTCP operation is notpossible,possible -- for example, if one host is not MPTCP capable or if a middlebox alters the payload. This is discussed in <xreftarget="sec_fallback"/>.</t> <t>Totarget="sec_fallback" format="default"/>.</li> <li>To address the threats identified in <xreftarget="RFC6181"/>,target="RFC6181" format="default"/>, the following steps are taken: keys are sent in the clear in the MP_CAPABLE messages; MP_JOIN messages are secured with HMAC-SHA256 (<xreftarget="RFC2104"/>,target="RFC2104" format="default"/> using the algorithm in <xreftarget="RFC6234"/>)target="RFC6234" format="default"/>) using those keys; and standard TCP validity checks are made on the other messages (ensuring that sequence numbers arein-windowin‑window <xreftarget="RFC5961"/>).target="RFC5961" format="default"/>). Residual threats to MPTCP v0 were identified in <xreftarget="RFC7430"/>,target="RFC7430" format="default"/>, and those affecting the protocol(i.e. modification(i.e., modifications to ADD_ADDR) have been incorporated in this document. Further discussion of security can be found in <xreftarget="sec_security"/>.</t> </list></t>target="sec_security" format="default"/>.</li> </ul> </section> </section> <sectiontitle="MPTCP Protocol" anchor="sec_protocol">anchor="sec_protocol" numbered="true" toc="default"> <name>MPTCP Operations: An Overview</name> <t>This section describes the operation ofthe MPTCP protocol, and is subdivided into sections forMPTCP. The subsections below discuss each key part of the protocol operation.</t> <t>All MPTCP operations are signaled using optional TCP header fields. A single TCP option number ("Kind") has been assigned by IANA for MPTCP (see <xreftarget="IANA"/>),target="IANA" format="default"/>), and then individual messages will be determined by a "subtype", the values of which are also stored in an IANA registry (and are also listed in <xreftarget="IANA"/>).target="IANA" format="default"/>). As with all TCP options, the Length field is specified inbytes,bytes and includes the2 bytes2 bytes of Kind and Length.</t> <t>Throughout this document, when reference is made to an MPTCP option by symbolic name, such as "MP_CAPABLE", this refers to a TCP option with the single MPTCP option type, and with the subtype value of the symbolic name as defined in <xreftarget="IANA"/>.target="IANA" format="default"/>. This subtype is a 4-bit field -- the first 4 bits of the option payload, as shown in <xreftarget="fig_option"/>.target="fig_option" format="default"/>. The MPTCP messages are defined in the following sections.</t><?rfc needLines='8'?><figurealign="center" anchor="fig_option" title="MPTCPanchor="fig_option"> <name>MPTCP OptionFormat">Format</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-----------------------+ | Kind | Length |Subtype| | +---------------+---------------+-------+ | | Subtype-specific data | | (variable length) | +---------------------------------------------------------------+ ]]></artwork> </figure> <t>Those MPTCP options associated with subflow initiation are used on packets with the SYN flag set. Additionally, there is one MPTCP option for signaling metadata to ensure that segmented data can be recombined for delivery to the application.</t> <t>The remaining options, however, are signals that do not need to be on a specific packet, such as those for signaling additional addresses.WhilstWhile an implementation may desire to send MPTCP options as soon as possible, it may not be possible to combine all desired options (both those for MPTCP and for regular TCP, such as SACK (selective acknowledgment) <xreftarget="RFC2018"/>)target="RFC2018" format="default"/>) on a single packet. Therefore, an implementation may choose to send duplicate ACKs containing the additional signaling information. This changes the semantics of a duplicate ACK; these are usually only sent as a signal of a lost segment <xreftarget="RFC5681"/>target="RFC5681" format="default"/> in regular TCP. Therefore, an MPTCP implementation receiving a duplicate ACK that contains an MPTCP optionMUST NOT<bcp14>MUST NOT</bcp14> treat it as a signal of congestion. Additionally, an MPTCP implementationSHOULD NOT<bcp14>SHOULD NOT</bcp14> send more than two duplicate ACKs in a row for the purposes of sending MPTCP options alone, in order to ensure that no middleboxes misinterpret this as a sign of congestion.</t> <t>Furthermore, standard TCP validity checks (such as ensuring that the sequence number and acknowledgment number are within the window)MUST<bcp14>MUST</bcp14> be undertaken before processing any MPTCP signals, as described in <xreftarget="RFC5961"/>,target="RFC5961" format="default"/>, and initial subflow sequence numbersSHOULD<bcp14>SHOULD</bcp14> be generated according to the recommendations in <xreftarget="RFC6528"/>.</t>target="RFC6528" format="default"/>.</t> <sectiontitle="Connection Initiation" anchor="sec_init">anchor="sec_init" numbered="true" toc="default"> <name>Connection Initiation</name> <t>Connection initiation begins with a SYN, SYN/ACK, ACK exchange on a single path. Each packet contains the Multipath Capable (MP_CAPABLE) MPTCP option (<xreftarget="tcpm_capable"/>).target="tcpm_capable" format="default"/>). This option declares its senderiscapable of performing Multipath TCP and wishes to do so on this particular connection.</t><t>The MP_CAPABLE exchange in this specification (v1) is different to that specified in v0. If a host supports multiple versions of MPTCP, the sender of the MP_CAPABLE option SHOULD signal the highest version number it supports. In return, in its MP_CAPABLE option, the<figure anchor="tcpm_capable"> <name>Multipath Capable (MP_CAPABLE) Option</name> <artwork align="left" name="" type="" alt=""><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-------+---------------+ | Kind | Length |Subtype|Version|A|B|C|D|E|F|G|H| +---------------+---------------+-------+-------+---------------+ | Option Sender's Key (64 bits) | | (if option Length > 4) | | | +---------------------------------------------------------------+ | Option Receiver's Key (64 bits) | | (if option Length > 12) | | | +-------------------------------+-------------------------------+ | Data-Level Length (16 bits) | Checksum (16 bits, optional) | +-------------------------------+-------------------------------+ ]]></artwork> </figure> <t>The MP_CAPABLE exchange in this specification (v1) is different than that specified in v0. If a host supports multiple versions of MPTCP, the sender of the MP_CAPABLE option <bcp14>SHOULD</bcp14> signal the highest version number it supports. In return, in its MP_CAPABLE option, the receiver will signal the version number it wishes to use, whichMUST<bcp14>MUST</bcp14> be equal to or lower than the version number indicated in the initial MP_CAPABLE. There is acaveat thoughcaveat, though, with respect to this version negotiation with old listeners that only support v0. A listener that supports v0 expects that the MP_CAPABLE option in theSYN-segment includesSYN segment will include the initiator's key.IfIf, however, the initiatorhoweveralready upgraded to v1, it won't include the key in theSYN-segment.SYN segment. Thus, the listener will ignore the MP_CAPABLE of thisSYN-segmentSYN segment and reply with a SYN/ACK that does not include an MP_CAPABLE. The initiatorMAY<bcp14>MAY</bcp14> choose to immediately fall back to TCP orMAY<bcp14>MAY</bcp14> choose to attempt a connection using MPTCP v0 (if the initiator supports v0), in order to discover whether the listener supports the earlier version of MPTCP. Ingeneral ageneral, an MPTCP v0 connectioniswill likelytobe preferredtoover a TCPone, howeverconnection; however, in a particular deploymentscenarioscenario, it may be known that the listener is unlikely to supportMPTCPv0MPTCP v0 and so the initiator may prefer not to attempt a v0 connection. An initiatorMAY<bcp14>MAY</bcp14> cache information for a peer about what version of MPTCP itsupportssupports, if any, and use this information for future connection attempts.</t> <t>The MP_CAPABLE option isvariable-length,of variable length, with different fieldsincludedincluded, depending on which packet the option is used on. The full MP_CAPABLE option is shown in <xreftarget="tcpm_capable"/>.</t> <?rfc needLines='10'?> <figure align="center" anchor="tcpm_capable" title="Multipath Capable (MP_CAPABLE) Option"> <artwork align="left"><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-------+---------------+ | Kind | Length |Subtype|Version|A|B|C|D|E|F|G|H| +---------------+---------------+-------+-------+---------------+ | Option Sender's Key (64 bits) | | (if option Length > 4) | | | +---------------------------------------------------------------+ | Option Receiver's Key (64 bits) | | (if option Length > 12) | | | +-------------------------------+-------------------------------+ | Data-Level Length (16 bits) | Checksum (16 bits, optional) | +-------------------------------+-------------------------------+ ]]></artwork> </figure>target="tcpm_capable" format="default"/>.</t> <t>The MP_CAPABLE option is carried on the SYN, SYN/ACK, and ACK packets that start the first subflow of an MPTCP connection, as well as the first packet that carries data, if the initiator wishes to send first. The data carried by each option is as follows, whereA = initiatorA = initiator and B = listener.<list style="symbols"> <t>SYN</t> <ul spacing="normal"> <li>SYN (A->B): only the firstfour4 octets (Length =4).</t> <t>SYN/ACK4).</li> <li>SYN/ACK (B->A): B'sKeykey for this connection (Length =12).</t> <t>ACK12).</li> <li>ACK (no data) (A->B): A'sKeykey followed by B'sKeykey (Length =20).</t> <t>ACK20).</li> <li>ACK (with first data) (A->B): A'sKeykey followed by B'sKeykey followed by Data-Level Length, and optional Checksum (Length = 22 or24).</t> </list>24).</li> </ul> <t> The contents of the optionisare determined by the SYN and ACK flags of the packet, along with the option'slengthLength field.For the diagram shown inIn <xreftarget="tcpm_capable"/>, "sender"target="tcpm_capable" format="default"/>, "Sender" and"receiver""Receiver" refer to the sender or receiver of the TCP packet (which can be either host).</t> <t>The initial SYN, containing just the MP_CAPABLE header, is used to define the version of MPTCP beingrequested, as well as exchangingrequested and also to exchange flags to negotiate connection features, as described later.</t> <t>This option is used to declare the 64-bit keys that the end hosts have generated for this MPTCP connection. These keys are used to authenticate the addition of future subflows to this connection. This is the only time the key will be sent in the clear on the wire (unless"fast close", <xref target="sec_fastclose"/>,"Fast Close" (<xref target="sec_fastclose" format="default"/>) is used); all future subflows will identify the connection using a 32-bit "token". This token is a cryptographic hash of this key. The algorithm for this process is dependent on the authentication algorithm selected; the method of selection is defined later in this section.</t> <t>Upon reception of the initialSYN-segment,SYN segment, a stateful server generates a random key and replies with a SYN/ACK. The key's method of generation is implementation specific. The keyMUST<bcp14>MUST</bcp14> be hard to guess, and itMUST<bcp14>MUST</bcp14> be unique for the sending host across all its current MPTCP connections. Recommendations for generating random numbers for use in keys are given in <xreftarget="RFC4086"/>.target="RFC4086" format="default"/>. Connections will be indexed at each host by the token (a one-way hash of the key). Therefore, an implementation will require a mapping from each token to the corresponding connection, and in turn to the keys for the connection.</t> <t>There is a risk that two different keys will hash to the same token. The risk of hash collisions is usually small, unless the host is handling many tens of thousands of connections. Therefore, an implementationSHOULD<bcp14>SHOULD</bcp14> check its list of connection tokens to ensure that there is no collision before sending its key, and if there is, then it should generate a new key. This would, however, be costly for a server with thousands of connections. The subflow handshake mechanism (<xreftarget="sec_join"/>)target="sec_join" format="default"/>) will ensure that new subflows only join the correct connection, however, through the cryptographic handshake, as well as checking the connection tokens in both directions, and ensuring that sequence numbers are in-window.SoSo, in the worstcasecase, if there was a token collision, the new subflow would not succeed, but the MPTCP connection would continue to provide a regular TCP service.</t> <t>Since key generation isimplementation-specific,implementation specific, there is no requirement that theybesimply be random numbers. An implementation is free to exchange cryptographic materialout-of-bandout of band and generate these keys fromthis,this material, in order to provide additional mechanisms by which to verify the identity of the communicating entities. For example, an implementation could choose to link its MPTCP keys to those used in higher-layer TLS or SSH connections.</t> <t>If the server behaves in a stateless manner, it has to generate its own key in a verifiable fashion. This verifiable way of generating the key can be done by using a hash of the 4-tuple, sequencenumbernumber, and a local secret (similar to what is done for theTCP-sequenceTCP sequence number <xreftarget="RFC4987"/>).target="RFC4987" format="default"/>). It will thus be able to verify whether it is indeed the originator of the key echoed back in thelatersubsequent MP_CAPABLE option. As for a stateful server, the tokensSHOULD<bcp14>SHOULD</bcp14> be checked foruniqueness, howeveruniqueness; however, if uniqueness is notmet,met and there is no way to generate an alternative verifiable key, then the connectionMUST<bcp14>MUST</bcp14> fall back to using regular TCP by not sendingaan MP_CAPABLE in theSYN/ACK.</t>SYN&wj;/ACK.</t> <t>The ACK carries both A's key and B's key. This is the first time that A's key is seen on the wire, although it is expected that A will have generated a key locally before the initial SYN. The echoing of B's key allows B to operate statelessly, as described above. Therefore, A's key must be delivered reliably to B, and in order to do this, the transmission of this packet must be made reliable.</t> <t>If B has data to send first, then the reliable delivery of theACK+MP_CAPABLE can be inferredACK + MP_CAPABLE is ensured by the receipt of this data withaan MPTCP Data Sequence Signal (DSS) option (<xreftarget="sec_generalop"/>).target="sec_generalop" format="default"/>) containing a DATA_ACK for the MP_CAPABLE (which is the first octet of the data sequence space). If, however, A wishes to send data first, it has two options to ensure the reliable delivery of theACK+MP_CAPABLE.ACK + MP_CAPABLE. If it immediately has data to send, then thethirdfirst ACK (with data) would also contain an MP_CAPABLE option with additional data parameters (the Data-Level Length and optional Checksum as shown in <xreftarget="tcpm_capable"/>).target="tcpm_capable" format="default"/>). If A does not immediately have data to send, itMUST<bcp14>MUST</bcp14> include the MP_CAPABLE on thethirdfirst ACK, but without the additional data parameters. When A does have data to send, it must repeat the sending of the MP_CAPABLE option from thethirdfirst ACK, with additional data parameters. This MP_CAPABLE option is used in place of theDSS,DSS and simply specifiesthe data-level length(1) the Data-Level Length of thepayload,payload andthe(2) the checksum (if the use of checksums is negotiated). This is the minimal data required to establishaan MPTCP connection--- it allows validation of the payload, and given that it is the first data, the Initial Data Sequence Number (IDSN) is also known (as it is generated from the key, as described below). Conveying the keys on the first data packet allows the TCP reliability mechanisms to ensure that the packet is successfully delivered. The receiver will acknowledge this data at the connection level with a Data ACK, as if a DSS option has been received.</t> <t>There could be situations where both A and B attempt to transmit initial data at the same time. For example, if A did not initially have data tosend,send but then needed to transmit data before it had received anything from B, it would useaan MP_CAPABLE option with data parameters (since it would not know if the MP_CAPABLE on the ACK was received). In such a situation, B may also have transmitted data with a DSS option, but it had not yet been received at A. Therefore, B has received data withaan MP_CAPABLE mapping after it has sent data with a DSS option. To ensure that these situations can be handled, it follows that the data parameters inaan MP_CAPABLE are semantically equivalent to those in a DSS option and can be used interchangeably. Similar situations could occur when the MP_CAPABLE with data is lost and retransmitted. Furthermore, in the case of TCPSegmentation Offloading,segmentation offloading, the MP_CAPABLE with data parameters may be duplicated across multiple packets, and implementations must also be able to cope with duplicate MP_CAPABLE mappings as well as duplicate DSS mappings.</t> <t>Additionally, the MP_CAPABLE exchange allows the safe passage of MPTCP options on SYN packets to be determined. If any of these options are dropped, MPTCP will gracefully fall back to regular single-path TCP, as documented in <xreftarget="sec_fallback"/>.target="sec_fallback" format="default"/>. If at any point in the handshake either party thinks the MPTCP negotiation iscompromised,compromised -- forexampleexample, by a middlebox corrupting the TCPoptions,options or by unexpected ACK numbers beingpresent,present -- the hostMUST<bcp14>MUST</bcp14> stop using MPTCP and no longer include MPTCP options in future TCP packets. The other host will then also fall back to regular TCP using thefall backfallback mechanism. Note that new subflowsMUST NOT<bcp14>MUST NOT</bcp14> be established (using the process documented in <xreftarget="sec_join"/>)target="sec_join" format="default"/>) until aData Sequence Signal (DSS)DSS option has been successfully received across the path (as documented in <xreftarget="sec_generalop"/>).</t>target="sec_generalop" format="default"/>).</t> <t>Like all MPTCP options, the MP_CAPABLE option starts with the Kind and Length to specify theTCP-optionTCP option's kind anditslength.Followed by thatThis information is followed by the MP_CAPABLE option. The first 4 bits of the first octet in the MP_CAPABLE option (<xreftarget="tcpm_capable"/>)target="tcpm_capable" format="default"/>) define the MPTCPoption subtypeOption Subtype (see <xreftarget="IANA"/>;target="IANA" format="default"/>; for MP_CAPABLE, this value is 0x0), and the remaining4 bits4 bits of this octet specify the MPTCP version in use (for this specification, thisis 1).</t>value is 1).</t> <t>The second octet is reserved for flags, allocated as follows:<list style="hanging"> <t hangText="A:"></t> <dl newline="false" spacing="normal" indent="14"> <dt>A:</dt> <dd> The leftmost bit, labeled "A",SHOULD<bcp14>SHOULD</bcp14> be set to 1 to indicate "ChecksumRequired",required", unless the system administrator has decided that checksums are not required (for example, if the environment is controlled and no middleboxes exist that might adjust thepayload).</t> <t hangText="B:">payload).</dd> <dt>B:</dt> <dd> The second bit, labeled "B", is an extensibilityflag, and MUSTflag. It <bcp14>MUST</bcp14> be set to 0 for current implementations. This flag will be used for an extensibility mechanism in a future specification, and the impact of this flag will be defined at a later date. It is expected, but not mandated, that this flag would be used as part of an alternative security mechanism that does not require a full version upgrade of theprotocol,protocol but does require redefining some elements of the handshake. If receiving a message with the'B'"B" flag set to1,1 and this is not understood, then the MP_CAPABLE in this SYNMUST<bcp14>MUST</bcp14> be silently ignored, which triggers a fallback to regular TCP; the sender is expected to retry with a format compatible with this legacy specification. Note that the length of the MP_CAPABLE option, and the meanings of bits "D" through "H", may be altered by settingB=1.</t> <t hangText="C:">B=1.</dd> <dt>C:</dt> <dd> The third bit, labeled "C", is set to"1"1 to indicate that the sender of this option will not accept additional MPTCP subflows to the source address and port, and therefore the receiverMUST NOT<bcp14>MUST NOT</bcp14> try to open any additional subflowstowardstoward this address and port. Thisis animproves efficiencyimprovement forin situations where the sender knows a restriction is inplace,place -- forexampleexample, if the sender is behind a strictNAT,NAT or operating behind a legacy Layer 4 loadbalancer.</t> <t hangText="Dbalancer.</dd> <dt>D throughH:">H:</dt> <dd> The remaining bits, labeled "D" through "H", are used for crypto algorithm negotiation. In thisspecificationspecification, only the rightmost bit, labeled "H", is assigned. Bit "H" indicates the use of HMAC-SHA256 (as defined in <xreftarget="sec_join"/>).target="sec_join" format="default"/>). An implementation that only supports this methodMUST<bcp14>MUST</bcp14> set bit "H" to1,1 and bits "D" through "G" to0.</t> </list> A0.</dd> </dl> <t>A crypto algorithmMUST<bcp14>MUST</bcp14> be specified. If flag bitsD"D" throughH"H" are all 0, the MP_CAPABLE optionMUST<bcp14>MUST</bcp14> be treated as invalid and ignored (that is, it must be treated as a regular TCP handshake).</t> <t>The selection of the authentication algorithm also impacts the algorithm used to generate the token and theInitial Data Sequence Number (IDSN).IDSN. In this specification, with only the SHA-256 algorithm (bit "H") specified and selected, the tokenMUST<bcp14>MUST</bcp14> be a truncated (most significant32 bits)32 bits) SHA-256 hash(<xref target="RFC6234"/>)<xref target="RFC6234" format="default"/> of the key. A different, 64-bit truncation (the least significant 64 bits) of the SHA-256 hash of the keyMUST<bcp14>MUST</bcp14> be used as the IDSN. Note that the keyMUST<bcp14>MUST</bcp14> be hashed in network byte order. Also note that the "least significant" bitsMUST<bcp14>MUST</bcp14> be the rightmost bits of the SHA-256 digest, as per <xreftarget="RFC6234"/>.target="RFC6234" format="default"/>. Future specifications of the use of the crypto bits may choose to specify different algorithms for token and IDSN generation.</t> <t>Both the crypto and checksum bits negotiate capabilities in similar ways. For theChecksum Required"Checksum required" bit (labeled "A"), if either host requires the use of checksums, checksumsMUST<bcp14>MUST</bcp14> be used. In other words, the only way for checksums not to be used is if both hosts in their SYNs set A=0. This decision is confirmed by the setting of the "A" bit in the third packet (the ACK) of the handshake. For example, if the initiator sets A=0 in theSYN,SYN but the responder sets A=1 in the SYN/ACK, checksumsMUST<bcp14>MUST</bcp14> be used in both directions, and the initiator will set A=1 in the ACK. The decision regarding whether to use checksums will be stored by an implementation in a per-connection binary state variable. If A=1 is received by a host that does not want to use checksums, itMUST<bcp14>MUST</bcp14> fall back to regular TCP by ignoring the MP_CAPABLE option as if it was invalid.</t> <t>For crypto negotiation, the responder has the choice. The initiator creates a proposal setting a bit for each algorithm it supports to 1 (in this version of the specification, there is only one proposal, so bit "H" willbealways be set to 1). The responder responds with only1 bit1 bit set -- this is the chosen algorithm. The rationale for this behavior is that the responder will typically be a server with potentially many thousands of connections, so it may wish to choose an algorithm with minimal computational complexity, depending on the load. If a responder does not support (or does not want to support) any of the initiator's proposals, itMUST<bcp14>MUST</bcp14> respond without an MP_CAPABLE option, thus forcing a fallback to regular TCP.</t> <t>The MP_CAPABLE option is only used in the first subflow of a connection, in order to identify the connection; allfollowingsubsequent subflows will use the"Join"MP_JOIN option (see <xreftarget="sec_join"/>)target="sec_join" format="default"/>) to join the existing connection.</t> <t>If a SYN contains an MP_CAPABLE option but the SYN/ACK does not, it is assumed that the sender of the SYN/ACK is not multipath capable; thus, the MPTCP sessionMUST<bcp14>MUST</bcp14> operate as a regular, single-pathTCP.TCP session. If a SYN does not containaan MP_CAPABLE option, the SYN/ACKMUST NOT<bcp14>MUST NOT</bcp14> contain one in response. If the third packet (the ACK) does not contain the MP_CAPABLE option, then the sessionMUST<bcp14>MUST</bcp14> fall back to operating as a regular, single-pathTCP.TCP session. This is done to maintain compatibility with middleboxes on the path that drop some or all TCP options. Note that an implementationMAY<bcp14>MAY</bcp14> choose to attempt sending MPTCP options more than one time before making this decision to operate as regular TCP (see <xreftarget="heuristics"/>).</t>target="heuristics" format="default"/>).</t> <t>If the SYN packets are unacknowledged, it is up to local policy to decide how to respond. It is expected that a sender will eventually fall back to single-path TCP (i.e., without the MP_CAPABLE option) in order to work around middleboxes that may drop packets with unknown options; however, the number of multipath-capable attempts that are made first will be up to local policy. It is possible that MPTCP and non-MPTCP SYNs could get reordered in the network. Therefore, the final state is inferred from the presence or absence of the MP_CAPABLE option in the third packet of the TCP handshake. If this option is not present, the connectionSHOULD<bcp14>SHOULD</bcp14> fall back to regular TCP, as documented in <xreftarget="sec_fallback"/>.</t>target="sec_fallback" format="default"/>.</t> <t>Theinitial data sequence numberIDSN on an MPTCP connection is generated from the key. The algorithm for IDSN generation is also determined from the negotiated authentication algorithm. In this specification, with only the SHA-256 algorithm specified and selected, the IDSN of a hostMUST<bcp14>MUST</bcp14> be the least significant64 bits64 bits of the SHA-256 hash of its key, i.e., IDSN-A = Hash(Key-A) and IDSN-B = Hash(Key-B). This deterministic generation of the IDSN allows a receiver to ensure that there are no gaps in sequence space at the start of the connection. The SYN with MP_CAPABLE occupies the first octet of data sequence space, although this does not need to be acknowledged at the connection level until the first data is sent (see <xreftarget="sec_generalop"/>).</t>target="sec_generalop" format="default"/>).</t> </section> <sectiontitle="Startinganchor="sec_join" numbered="true" toc="default"> <name>Starting a NewSubflow" anchor="sec_join">Subflow</name> <t>Once an MPTCP connection has begun with the MP_CAPABLE exchange, further subflows can be added to the connection. Hosts have knowledge of their ownaddress(es),address(es) and can become aware of the other host's addresses through signaling exchanges as described in <xreftarget="sec_pm"/>.target="sec_pm" format="default"/>. Using this knowledge, a host can initiate a new subflow over a currently unused pair of addresses. It ispermittedpermissible for either host in a connection to initiate the creation of a new subflow, but it is expected that this will normally be the original connection initiator (see <xreftarget="heuristics"/>target="heuristics" format="default"/> for heuristics).</t> <t>A new subflow is started as a normal TCP SYN/ACK exchange. The Join Connection (MP_JOIN) MPTCP option is used to identify the connection to be joined by the new subflow. It uses keying material that was exchanged in the initial MP_CAPABLE handshake (<xreftarget="sec_init"/>),target="sec_init" format="default"/>), and that handshake also negotiates the crypto algorithm in use for the MP_JOIN handshake.</t> <t>This section specifies the behavior of MP_JOIN using the HMAC-SHA256 algorithm. An MP_JOIN option is present in the SYN, SYN/ACK, and ACK of the three-way handshake, although in each case with a different format.</t> <t>In the first MP_JOIN on the SYN packet, illustrated in <xreftarget="tcpm_join"/>,target="tcpm_join" format="default"/>, the initiator sends a token, random number, andaddressAddress ID.</t> <figure anchor="tcpm_join"> <name>Join Connection (MP_JOIN) Option (for Initial SYN)</name> <artwork align="left" name="" type="" alt=""><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-----+-+---------------+ | Kind | Length = 12 |Subtype|(rsv)|B| Address ID | +---------------+---------------+-------+-----+-+---------------+ | Receiver's Token (32 bits) | +---------------------------------------------------------------+ | Sender's Random Number (32 bits) | +---------------------------------------------------------------+ ]]></artwork> </figure> <t>The token is used to identify the MPTCP connection and is a cryptographic hash of the receiver's key, as exchanged in the initial MP_CAPABLE handshake (<xreftarget="sec_init"/>).target="sec_init" format="default"/>). In this specification, the tokens presented in this option are generated by the SHA-256 algorithm <xreftarget="RFC6234"/> algorithm,target="RFC6234" format="default"/>, truncated to the most significant 32 bits. The token included in the MP_JOIN option is the token that the receiver of the packet uses to identify this connection; i.e., Host A will send Token-B (which is generated from Key-B). Note that the hash generation algorithm can be overridden by the choice of cryptographic handshake algorithm, as defined in <xreftarget="sec_init"/>.</t>target="sec_init" format="default"/>.</t> <t>The MP_JOIN SYN sends not only the token (which is static for a connection) but also random numbers (nonces) that are used to prevent replay attacks on the authentication method. Recommendations for the generation of random numbers for this purpose are given in <xreftarget="RFC4086"/>.</t>target="RFC4086" format="default"/>.</t> <t>The MP_JOIN option includes an "Address ID". This is an identifier generated by the sender of the option, used to identify the source address of this packet, even if the IP header has been changed in transit by a middlebox. The numeric value of this field is generated by the sender and must map uniquely to a source IP address for the sending host. The Address ID allows address removal (<xreftarget="sec_remove_addr"/>)target="sec_remove_addr" format="default"/>) without needing to know what the source address at the receiver is, thus allowing address removal through NATs. The Address ID also allows correlation between new subflow setup attempts and address signaling (<xreftarget="sec_add_address"/>),target="sec_add_address" format="default"/>), to prevent setting up duplicate subflows on the same path, if an MP_JOIN and ADD_ADDR are sent at the same time.</t> <t>The Address IDs of the subflow used in the initial SYN exchange of the first subflow in the connection areimplicit,implicit and have the value zero. A hostMUST<bcp14>MUST</bcp14> store the mappings between Address IDs and addresses both for itself and the remote host. An implementation will also need to know which local and remote Address IDs are associated with which established subflows, for when addresses are removed from a local or remote host.</t> <t>The MP_JOIN option on packets with the SYN flag set also includes4 bits4 bits of flags, 3 of which are currently reserved andMUST<bcp14>MUST</bcp14> be set tozero0 by the sender. The final bit, labeled "B", indicates whether the sender of this optionwishes(1) wishes this subflow to be used as a backup path (B=1) in the event of failure of otherpaths,paths orwhether it wants it(2) wants the subflow to be used as part of the connection immediately. By setting B=1, the sender of the option is requesting that the other hosttoonly send data on this subflow if there are no available subflows where B=0. Subflow policy is discussed in more detail in <xreftarget="sec_policy"/>.</t> <?rfc needLines='10'?> <figure align="center" anchor="tcpm_join" title="Join Connection (MP_JOIN) Option (for Initial SYN)"> <artwork align="left"><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-----+-+---------------+ | Kind | Length = 12 |Subtype|(rsv)|B| Address ID | +---------------+---------------+-------+-----+-+---------------+ | Receiver's Token (32 bits) | +---------------------------------------------------------------+ | Sender's Random Number (32 bits) | +---------------------------------------------------------------+ ]]></artwork> </figure>target="sec_policy" format="default"/>.</t> <t>When receiving a SYN with an MP_JOIN option that contains a valid token for an existing MPTCP connection, the recipientSHOULD<bcp14>SHOULD</bcp14> respond with a SYN/ACK also containing an MP_JOIN option containing a random number and a truncated (leftmost64 bits) Hash-based Message Authentication Code (HMAC).64 bits) HMAC. This version of the option is shown in <xreftarget="tcpm_join2"/>.target="tcpm_join2" format="default"/>. If the token isunknown,unknown or the host wants to refuse subflow establishment (for example, due to a limit on the number of subflows it will permit), the receiver will send back a reset (RST) signal, analogous to an unknown port in TCP, containingaan MP_TCPRST option (<xreftarget="sec_reset"/>)target="sec_reset" format="default"/>) withaan "MPTCP specific error" reason code. Although calculating an HMACrequires cryptographic operations, it is believed that the 32-bit token in the MP_JOIN SYN gives sufficient protection against blind state exhaustion attacks; therefore, there is no need to provide mechanisms to allow a responder to operate statelessly at the MP_JOIN stage.</t>requires cryptographic operations, it is believed that the 32-bit token in the MP_JOIN SYN gives sufficient protection against blind state exhaustion attacks; therefore, there is no need to provide mechanisms to allow a responder to operate statelessly at the MP_JOIN stage.</t> <figure anchor="tcpm_join2"> <name>Join Connection (MP_JOIN) Option (for Responding SYN/ACK)</name> <artwork align="left" name="" type="" alt=""><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-----+-+---------------+ | Kind | Length = 16 |Subtype|(rsv)|B| Address ID | +---------------+---------------+-------+-----+-+---------------+ | | | Sender's Truncated HMAC (64 bits) | | | +---------------------------------------------------------------+ | Sender's Random Number (32 bits) | +---------------------------------------------------------------+ ]]></artwork> </figure> <t>An HMAC is sent by both hosts -- by the initiator (Host A) in the third packet (the ACK) and by the responder (Host B) in the second packet (the SYN/ACK). Doing the HMAC exchange at this stage allows both hosts to have first exchanged random data (in the first two SYN packets) that is used as the "message". This specification defines that HMAC as defined in <xreftarget="RFC2104"/>target="RFC2104" format="default"/> is used, along with the SHA-256 hash algorithm <xreftarget="RFC6234"/>,target="RFC6234" format="default"/>, and that the output is truncated to the leftmost 160 bits (20 octets). Due to option space limitations, the HMAC included in the SYN/ACK is truncated to the leftmost 64 bits, but this isacceptableacceptable, since random numbers are used; thus, an attacker only has one chance to correctly guess the HMAC that matches the random number previously sent by the peer (if the HMAC is incorrect, the TCP connection is closed, so a new MP_JOIN negotiation with a new random number is required).</t> <t>The initiator's authentication information is sent in its first ACK (the third packet of the handshake), as shown in <xreftarget="tcpm_join3"/>.target="tcpm_join3" format="default"/>. This data needs to be sent reliably, since it is the only time this HMAC is sent; therefore, receipt of this packetMUST<bcp14>MUST</bcp14> trigger a regular TCP ACK in response, and the packetMUST<bcp14>MUST</bcp14> be retransmitted if this ACK is not received. In other words, sending the ACK/MP_JOIN packet places the subflow in the PRE_ESTABLISHED state, and it moves to the ESTABLISHED state only on receipt of an ACK from the receiver. It is notpermittedpermissible to send data while in the PRE_ESTABLISHED state. The reserved bits in this optionMUST<bcp14>MUST</bcp14> be set tozero0 by the sender.</t><t>The key for the HMAC algorithm, in the case of the message transmitted by Host A, will be Key-A followed by Key-B, and in the case of Host B, Key-B followed by Key-A. These are the keys that were exchanged in the original MP_CAPABLE handshake. The "message" for the HMAC algorithm in each case is the concatenations of random number for each host (denoted by R): for Host A, R-A followed by R-B; and for Host B, R-B followed by R-A.</t> <?rfc needLines='10'?><figurealign="center" anchor="tcpm_join2" title="Joinanchor="tcpm_join3"> <name>Join Connection (MP_JOIN) Option(for Responding SYN/ACK)">(for Initiator's First ACK)</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1+---------------+---------------+-------+-----+-+---------------++---------------+---------------+-------+-----------------------+ | Kind | Length =16 |Subtype|(rsv)|B| Address ID24 |Subtype| (reserved) | +---------------+---------------+-------+-----------------------+ | |+---------------+---------------+-------+-----+-+---------------+| | | Sender's Truncated HMAC(64(160 bits) | | |+---------------------------------------------------------------+|Sender's Random Number (32 bits)| +---------------------------------------------------------------+ ]]></artwork> </figure><?rfc needLines='12'?> <figure align="center" anchor="tcpm_join3" title="Join Connection (MP_JOIN) Option (for Third ACK)"> <artwork align="left"><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-----------------------+ | Kind | Length = 24 |Subtype| (reserved) | +---------------+---------------+-------+-----------------------+ | | | | | Sender's Truncated<t>The key for the HMAC algorithm, in the case of the message transmitted by Host A, will be Key-A followed by Key-B; and in the case of Host B, Key-B followed by Key-A. These are the keys that were exchanged in the original MP_CAPABLE handshake. The "message" for the HMAC(160 bits) | | | | | +---------------------------------------------------------------+ ]]></artwork> </figure>algorithm in each case is the concatenations of random numbers for each host (denoted by R): for Host A, R-A followed by R-B; and for Host B, R-B followed by R-A.</t> <t>These various MPTCP options fit together to enable authenticated subflow setup as illustrated in <xreftarget="fig_tokens"/>.</t> <?rfc needLines='24'?>target="fig_tokens" format="default"/>.</t> <figurealign="center" anchor="fig_tokens" title="Exampleanchor="fig_tokens"> <name>Example Use of MPTCPAuthentication">Authentication</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ Host A Host B ------------------------ ---------- Address A1 Address A2 Address B1 ---------- ---------- ---------- | | | | | SYN + MP_CAPABLE | |--------------------------------------------->| |<---------------------------------------------| | SYN/ACK + MP_CAPABLE(Key-B) | | | | | ACK + MP_CAPABLE(Key-A, Key-B) | |--------------------------------------------->| | | | | | SYN + MP_JOIN(Token-B, R-A) | | |------------------------------->| | |<-------------------------------| | | SYN/ACK + MP_JOIN(HMAC-B, R-B) | | | | | | ACK + MP_JOIN(HMAC-A) | | |------------------------------->| | |<-------------------------------| | | ACK | HMAC-A =HMAC(Key=(Key-A+Key-B), Msg=(R-A+R-B))HMAC(Key=(Key-A + Key-B), Msg=(R-A + R-B)) HMAC-B =HMAC(Key=(Key-B+Key-A), Msg=(R-B+R-A))HMAC(Key=(Key-B + Key-A), Msg=(R-B + R-A)) ]]></artwork> </figure> <t>If the token received at Host B is unknown or local policy prohibits the acceptance of the new subflow, the recipientMUST<bcp14>MUST</bcp14> respond with a TCP RST for the subflow. If appropriate,aan MP_TCPRST option withaan "Administratively prohibited" reason code (<xreftarget="sec_reset"/>)target="sec_reset" format="default"/>) should be included.</t> <t>If the token is accepted at HostB,B but the HMAC returned to Host A does not match the one expected, Host AMUST<bcp14>MUST</bcp14> close the subflow with a TCP RST. Inthis,this and allfollowingsubsequent cases of sending a RST as described in this section, the senderSHOULD<bcp14>SHOULD</bcp14> sendaan MP_TCPRST option (<xreftarget="sec_reset"/>)target="sec_reset" format="default"/>) on this RST packet with the reason code fora "MPTCP specifican "MPTCP-specific error".</t> <t>If Host B does not receive the expectedHMAC,HMAC or the MP_JOIN option is missing from the ACK, itMUST<bcp14>MUST</bcp14> close the subflow with a TCP RST.</t> <t>If the HMACs are verified as correct, then both hosts have verified each other as being the same peers as those that existed at the start of the connection, and they have agreed of which connection this subflow will become a part.</t> <t>If the SYN/ACK as received at Host A does not have an MP_JOIN option, Host AMUST<bcp14>MUST</bcp14> close the subflow with a TCP RST.</t> <t>This covers all cases of the loss of an MP_JOIN. In more detail, if an MP_JOIN is stripped from the SYN on the path from A toB, and HostB and Host B does not have a listener on the relevant port, it will respond with a RST in the normal way. If in response to a SYN with an MP_JOINoption,option a SYN/ACK is received without the MP_JOIN option(either since(because it was either stripped on the return path, orit wasstripped on the outgoing pathbutleading to Host Brespondedresponding as if itwerewas a new regular TCP session), then the subflow is unusable and Host AMUST<bcp14>MUST</bcp14> close it with a RST.</t> <t>Note that additional subflows can be created between any pair of ports (but see <xreftarget="heuristics"/>target="heuristics" format="default"/> for heuristics); no explicit application-level accept calls or bind calls are required to open additional subflows. To associate a new subflow with an existing connection, the token supplied in the subflow's SYN exchange is used for demultiplexing. This then binds the 5-tuple of the TCP subflow to the local token of the connection.AOne consequence is that it is possible to allow any port pairs to be used for a connection. </t> <t>Demultiplexing subflow SYNsMUST<bcp14>MUST</bcp14> be done using the token; this is unlike traditional TCP, where the destination port is used for demultiplexing SYN packets. Once a subflow is set up, demultiplexing packets is done using the 5-tuple, as in traditional TCP. The 5-tuples will be mapped to the local connection identifier (token). Note that Host A will know its local token for the subflow even though it is not sent on the wire -- only the responder's token is sent.</t> </section> <sectiontitle="General MPTCP Operation" anchor="sec_generalop">anchor="sec_generalop" numbered="true" toc="default"> <name>MPTCP Operation and Data Transfer</name> <t>This section discusses the operation of MPTCP for data transfer. At a high level, an MPTCP implementation will take one input data stream from anapplication,application and split it into one or more subflows, with sufficient control information to allow it to be reassembled and delivered reliably and in order to the recipient application. The following subsections define this behavior in detail.</t> <t>Thedata sequence mappingData Sequence Mapping and the Data ACK are signaled in theData Sequence Signal (DSS)DSS option (<xreftarget="tcpm_dsn"/>).target="tcpm_dsn" format="default"/>). Either or both can be signaled in one DSS, depending on the flags set. Thedata sequence mappingData Sequence Mapping defines how the sequence space on the subflow maps to the connection level, and the Data ACK acknowledges receipt of data at the connection level. These functions are described in more detail in the following two subsections.</t><?rfc needLines='18'?><figurealign="center" anchor="tcpm_dsn" title="Dataanchor="tcpm_dsn"> <name>Data Sequence Signal (DSS)Option">Option</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+----------------------+ | Kind | Length |Subtype| (reserved) |F|m|M|a|A| +---------------+---------------+-------+----------------------+ | Data ACK (4 or 8 octets, depending on flags) | +--------------------------------------------------------------+ | Datasequence numberSequence Number (4 or 8 octets, depending on flags) | +--------------------------------------------------------------+ | Subflow Sequence Number (4 octets) | +-------------------------------+------------------------------+ | Data-Level Length (2 octets) | Checksum (2 octets) | +-------------------------------+------------------------------+ ]]></artwork> </figure> <t>The flags, when set, define the contents of this option, as follows:<list style="symbols"> <t>A</t> <ul spacing="normal"> <li>A = Data ACKpresent</t> <t>apresent</li> <li>a = Data ACK is 8 octets (if not set, Data ACK is 4octets)</t> <t>Moctets)</li> <li>M = Data Sequence Number (DSN), Subflow Sequence Number (SSN), Data-Level Length, and Checksum (if negotiated)present</t> <t>mpresent</li> <li>m = Datasequence numberSequence Number is 8 octets (if not set, DSN is 4octets)</t> </list>octets)</li> </ul> <t> The flags'a'"a" and'm'"m" only have meaning if the corresponding'A'"A" or'M'"M" flags are set; otherwise, they will be ignored. The maximum length of this option, with all flags set, is 28 octets.</t> <t>The'F'"F" flag indicates "Data FIN". If present, this means that this mapping covers the final data from the sender. This is the connection-level equivalenttoof the FIN flag in single-path TCP. A connection is not closed unless there has been a Data FIN exchange,aan MP_FASTCLOSE (<xreftarget="sec_fastclose"/>)target="sec_fastclose" format="default"/>) message, or animplementation-specific,implementation-specific connection-level send timeout. The purpose of the Data FIN and the interactions between this flag, the subflow-level FIN flag, and thedata sequence mappingData Sequence Mapping are described in <xreftarget="sec_close"/>.target="sec_close" format="default"/>. The remaining reserved bitsMUST<bcp14>MUST</bcp14> be set tozero0 by an implementation of this specification.</t> <t>Note that the checksum is only present in this option if the use of MPTCP checksumming has been negotiated at the MP_CAPABLE handshake (see <xreftarget="sec_init"/>).target="sec_init" format="default"/>). The presence of the checksum can be inferred from the length of the option. If a checksum ispresent,present but its use had not been negotiated in the MP_CAPABLE handshake, the receiverMUST<bcp14>MUST</bcp14> close the subflow with aRSTRST, as it is not behaving as negotiated. If a checksum is not present when its use has been negotiated, the receiverMUST<bcp14>MUST</bcp14> close the subflow with aRSTRST, as it is considered broken. In both cases, this RSTSHOULD<bcp14>SHOULD</bcp14> be accompaniedwith aby an MP_TCPRST option (<xreftarget="sec_reset"/>)target="sec_reset" format="default"/>) with the reason code fora "MPTCP specifican "MPTCP-specific error".</t> <sectiontitle="Dataanchor="sec_dsn" numbered="true" toc="default"> <name>Data SequenceMapping" anchor="sec_dsn">Mapping</name> <t>The data stream as a whole can be reassembled through the use of thedata sequence mappingData Sequence Mapping components of the DSS option (<xreftarget="tcpm_dsn"/>),target="tcpm_dsn" format="default"/>), which define the mapping from the subflow sequence number to the data sequence number. This is used by the receiver to ensure in-order delivery to the application layer. Meanwhile, the subflow-level sequence numbers (i.e., the regular sequence numbers in the TCP header)have subflow-only relevance.are only relevant to the subflow. It is expected (but not mandated) that SACK <xreftarget='RFC2018'/> istarget="RFC2018" format="default"/> will be used at the subflow level to improve efficiency.</t> <t>Thedata sequence mappingData Sequence Mapping specifies a mapping from the subflow sequence space to the data sequence space. This is expressed in terms of starting sequence numbers for the subflow and the data level, and a length of bytes for which this mapping is valid. This explicit mapping for a range ofdata was chosendata, rather thanper-packet signalingper‑packet signaling, was chosen to assist with compatibility with situations where TCP/IP segmentation or coalescing is undertaken separately from the stack that is generating the data flow (e.g., through the use of TCP segmentation offloading on network interface cards, or by middleboxes such asperformance enhancing proxies).Performance Enhancing Proxies (PEPs) <xref target="RFC3135" format="default"/>). It also allows a single mapping to cover manypackets, whichpackets; this may be useful inbulk transferbulk‑transfer situations.</t> <t>A mapping is fixed, in that the subflow sequence number is bound to the data sequence number after the mapping has been processed. A senderMUST NOT<bcp14>MUST NOT</bcp14> change this mapping after it has been declared; however, the same data sequence number can be mapped to by different subflows for retransmission purposes (see <xreftarget="sec_retransmit"/>).target="sec_retransmit" format="default"/>). This would also permit the same data to be sent simultaneously on multiple subflows for resilience or efficiency purposes, especially in the case of lossy links. Although the detailed specification of such operation is outside the scope of this document, an implementationSHOULD<bcp14>SHOULD</bcp14> treat the first data that is received at a subflow for the data sequence space as the data thatwhichshould be delivered to the application, and anylatersubsequent data for that sequence spaceSHOULD<bcp14>SHOULD</bcp14> be ignored.</t> <t>The data sequence number is specified as an absolute value, whereas the subflow sequence numbering is relative (the SYN at the start of the subflow has a relative subflow sequence number of 0). This is done to allow middleboxes to change theinitial sequence numberInitial Sequence Number (ISN) of a subflow, such as firewalls that undertakeInitial Sequence Number (ISN)ISN randomization.</t> <t>Thedata sequence mappingData Sequence Mapping also contains a checksum of the data that this mapping covers, if the use of checksums has been negotiated at the MP_CAPABLE exchange. Checksums are used to detect if the payload has been adjusted in any way by a non-MPTCP-aware middlebox. If this checksum fails, it will trigger a failure of the subflow, or a fallback to regular TCP, as documented in <xreftarget="sec_fallback"/>,target="sec_fallback" format="default"/>, since MPTCP can no longer reliably know the subflow sequence space at the receiver to builddata sequence mappings.Data Sequence Mappings. Without checksumming enabled, corrupt data may be delivered to the application if a middlebox alters segment boundaries, alters content, or does not deliver all segments covered by adata sequence mapping.Data Sequence Mapping. It is thereforeRECOMMENDED to use<bcp14>RECOMMENDED</bcp14> that checksumming be used, unless it is known that the network path contains no such devices.</t> <t>The checksum algorithm used is the standard TCP checksum <xreftarget="RFC0793"/>,target="RFC0793" format="default"/>, operating over the data covered by this mapping, along with apseudo-headerpseudo‑header as shown in <xreftarget="fig_pseudo"/>.</t> <?rfc needLines='18'?>target="fig_pseudo" format="default"/>.</t> <figurealign="center" anchor="fig_pseudo" title="Pseudo-Headeranchor="fig_pseudo"> <name>Pseudo-Header for DSSChecksum">Checksum</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +--------------------------------------------------------------+ | | | Data Sequence Number (8 octets) | | | +--------------------------------------------------------------+ | Subflow Sequence Number (4 octets) | +-------------------------------+------------------------------+ | Data-Level Length (2 octets) | Zeros (2 octets) | +-------------------------------+------------------------------+ ]]></artwork> </figure> <t>Note that the data sequence number used in the pseudo-header is always the 64-bit value, irrespective of what length is used in the DSS option itself. The standard TCP checksum algorithm has beenchosenchosen, since it will be calculated anyway for the TCP subflow, and if calculated first over the data before adding the pseudo-headers, it only needs to be calculated once. Furthermore, since the TCP checksum is additive, the checksum for a DSN_MAP can be constructed by simply adding together the checksums for the data of each constituent TCPsegment,segment and adding the checksum for the DSSpseudo-header.</t>pseudo‑header.</t> <t>Note that checksumming relies on the TCP subflow containing contiguous data; therefore, a TCP subflowMUST NOT<bcp14>MUST NOT</bcp14> use the Urgent Pointer to interrupt an existing mapping. Further note, however, that if Urgent data is received on a subflow, itSHOULD<bcp14>SHOULD</bcp14> be mapped to the data sequence space and delivered to theapplicationapplication, analogous to Urgent data in regular TCP.</t> <t>To avoid possible deadlock scenarios, subflow-level processing should be undertaken separately fromthatprocessing at the connection level. Therefore, even if a mapping does not exist from the subflow space to thedata-leveldata‑level space, the dataSHOULD<bcp14>SHOULD</bcp14> still be ACKed at the subflow (if it is in-window). This data cannot, however, be acknowledged at the data level (<xreftarget="sec_dataack"/>)target="sec_dataack" format="default"/>) because its data sequence numbers are unknown. ImplementationsMAY<bcp14>MAY</bcp14> hold onto such unmapped data for a shortwhilewhile, in the expectation that a mapping will arrive shortly. Such unmapped data cannot be counted as being within theconnection levelconnection-level receive window because this is relative to the data sequence numbers, so if the receiver runs out of memory to hold this data, it will have to be discarded. If a mapping for that subflow-level sequence space does not arrive within a receive window of data, that subflowSHOULD<bcp14>SHOULD</bcp14> be treated as broken, closed with a RST, and any unmapped data silently discarded.</t> <t>Data sequence numbers are always 64-bitquantities,quantities andMUST<bcp14>MUST</bcp14> be maintained as such in implementations. If a connection is progressing at a slow rate, so protection against wrapped sequence numbers is not required, then an implementationMAY<bcp14>MAY</bcp14> include just the lower 32 bits of the data sequence number in thedata sequence mapping and/orData Sequence Mapping and&wj;/or Data ACK as an optimization, and an implementation can make this choice independently for each packet. An implementationMUST<bcp14>MUST</bcp14> be able to receive and process both 64-bitorand 32-bit sequence number values, but it is not required that an implementationisbe able to send both.</t> <t>An implementationMUST<bcp14>MUST</bcp14> send the full 64-bit data sequence number if it is transmitting at a sufficiently high rate that the 32-bit value could wrap within the Maximum Segment Lifetime (MSL) <xreftarget="RFC7323"/>.target="RFC7323" format="default"/>. The lengths of the DSNs used in these values (which may be different) are declared with flags in the DSS option. ImplementationsMUST<bcp14>MUST</bcp14> accept a 32-bit DSN and implicitly promote it to a 64-bit quantity by incrementing the upper 32 bits of the sequence number each time the lower 32 bits wrap. A sanity checkMUST<bcp14>MUST</bcp14> be implemented to ensure that a wrap occurs at an expected time (e.g., the sequence number jumps from a very high number to a very low number) and is not triggered byout-of-orderout‑of-order packets.</t> <t>As with the standard TCP sequence number, the data sequence number should not start at zero, but at a random value to make blind session hijacking harder. This specification requires setting theinitial data sequence number (IDSN)IDSN of each host to the least significant64 bits64 bits of the SHA-256 hash of the host's key, as described in <xreftarget="sec_init"/>.target="sec_init" format="default"/>. This isrequiredalso required in order for the receiver to know what the expected IDSNis,is and thus determine if any initial connection-level packets are missing; this is particularly relevant if two subflows start transmitting simultaneously.</t><t>A data sequence<t>The mapping provided by a Data Sequence Mapping MUST apply to some or all of the subflow sequence space in the TCP segment that carries the option. It does not need to be included in every MPTCP packet, as long as the subflow sequence space in that packet is covered by a mapping known at the receiver. This can be used to reduce overhead in cases where the mapping is known inadvance; oneadvance. One such case is when there is a single subflow between the hosts, and another is when segments of data are scheduled inlarger than packet-sizedlarger-than-packet-sized chunks.</t> <t>An "infinite" mapping can be used to fall back to regular TCP by mapping the subflow-level data to the connection-level data for the remainder of the connection (see <xreftarget="sec_fallback"/>).target="sec_fallback" format="default"/>). This is achieved by setting the Data-Level Length field of the DSS option to the reserved value of 0. The checksum, in such a case, will also be set tozero.</t>0.</t> </section> <sectiontitle="Data Acknowledgments" anchor="sec_dataack">anchor="sec_dataack" numbered="true" toc="default"> <name>Data Acknowledgments</name> <t>To provide full end-to-end resilience, MPTCP provides a connection-level acknowledgment, to act as a cumulative ACK for the connection as a whole. This is done via the "Data ACK" field of the DSS option (<xreftarget="tcpm_dsn"/>).target="tcpm_dsn" format="default"/>). The Data ACK is analogous to the behavior of the standard TCP cumulative ACK -- indicating how much data has been successfully received (with no holes). Thisis in comparisoncan be compared to the subflow-level ACK, which acts in a fashion analogous to TCP SACK, given that there may still be holes in the data stream at the connection level. The Data ACK specifies the next data sequence number it expects to receive.</t> <t>The Data ACK, as for the DSN, can be sent as the full 64-bitvalue,value or as the lower 32 bits. If data is received with a 64-bit DSN, itMUST<bcp14>MUST</bcp14> be acknowledged with a 64-bit Data ACK. If the DSN received is32 bits,32 bits, an implementation can choose whether to send a 32-bit or 64-bit Data ACK, and an implementationMUST<bcp14>MUST</bcp14> accept either in this situation.</t> <t>The Data ACK proves that the data, and all required MPTCP signaling,hashave been received and accepted by the remote end. One key use of the Data ACK signal is that it is used to indicate the left edge of the advertised receive window. As explained in <xreftarget="sec_rwin"/>,target="sec_rwin" format="default"/>, the receive window is shared by all subflows and is relative to the Data ACK. Because of this, an implementationMUST NOT<bcp14>MUST NOT</bcp14> use the RCV.WND field of a TCP segment at the connection level if it does not also carry a DSS option with a Data ACK field. Furthermore, separating the connection-level acknowledgments from the subflow level allows processing to be done separately, and a receiver has the freedom to drop segments after acknowledgment at the subflowlevel,level -- for example, due to memory constraints when many segments arrive out of order.</t> <t>An MPTCP senderMUST NOT<bcp14>MUST NOT</bcp14> free data from the send buffer until it has been acknowledged by both a Data ACK received on any subflow and at the subflow level by all subflows on which the data was sent. The former condition ensures liveness of theconnectionconnection, and the latter condition ensures liveness and self-consistence of a subflow when data needs to be retransmitted. Note, however, that if some data needs to be retransmitted multiple times over a subflow, there is a risk of blocking thesendingsend window. In this case, the MPTCP sender can decide to terminate the subflow that is behaving badly by sending a RST, using an appropriate MP_TCPRST (<xreftarget="sec_reset"/>)target="sec_reset" format="default"/>) error code.</t> <t>The Data ACKMAY<bcp14>MAY</bcp14> be included in all segments; however, optimizationsSHOULD<bcp14>SHOULD</bcp14> be considered in more advanced implementations, where the Data ACK is present in segments only when the Data ACK value advances, and this behaviorMUST<bcp14>MUST</bcp14> be treated as valid. This behavior ensures that thesendersend buffer is freed, while reducing overhead when the data transfer is unidirectional.</t> </section> <sectiontitle="Closinganchor="sec_close" numbered="true" toc="default"> <name>Closing aConnection" anchor="sec_close">Connection</name> <t>In regular TCP, a FIN announces to the receiver that the sender has no more data to send. In order to allow subflows to operate independently and to keep the appearance of TCP over the wire, a FIN in MPTCP only affects the subflow on which it is sent. This allows nodes to exercise considerable freedom over which paths are in use at any one time. The semantics of a FIN remain as for regular TCP; i.e., it is not until both sides have ACKed each other's FINs that the subflow is fully closed.</t> <t>When an application calls close() on a socket, this indicates that it has no more data to send; for regular TCP, this would result in a FIN on the connection. For MPTCP, an equivalent mechanism isneeded, andneeded; this is referred to as the DATA_FIN.</t> <t>A DATA_FIN is an indication that the sender has no more data to send, and as such it can be used to verify that all data has been successfully received. A DATA_FIN, as with the FIN on a regular TCP connection, is a unidirectional signal.</t> <t>The DATA_FIN is signaled by setting the'F'"F" flag in theData Sequence SignalDSS option (<xreftarget="tcpm_dsn"/>)target="tcpm_dsn" format="default"/>) to 1. A DATA_FIN occupies 1 octet (the final octet) of the connection-level sequence space. Note that the DATA_FIN is included in the Data-LevelLength,Length but not at the subflow level: for example, a segment with a DSN80,value of 80 and a Data-Level Length of 11, with DATA_FIN set, would map 10 octets from the subflow into data sequence space 80-89, and the DATA_FINiswould be DSN 90; therefore, thissegmentsegment, includingDATA_FINDATA_FIN, would be acknowledged with a DATA_ACKof 91.</t>of 91.</t> <t>Note that when the DATA_FIN is not attached to a TCP segment containing data, theData Sequence Signal MUSTDSS <bcp14>MUST</bcp14> have a subflow sequence number of 0, a Data-Level Length of 1, and the data sequence number that corresponds with the DATA_FIN itself. The checksum in this case will only cover the pseudo-header.</t> <t>A DATA_FIN has the same semantics and behavior as a regular TCP FIN, but at the connection level. Notably, it is only DATA_ACKed once all data has been successfully received at the connection level. Note, therefore, that a DATA_FIN is decoupled from a subflow FIN. It is only permissible to combine these signals on one subflow if there is no data outstanding on other subflows. Otherwise, it may be necessary to retransmit data on different subflows. Essentially, a hostMUST NOT<bcp14>MUST NOT</bcp14> close all functioning subflows unless it is safe to do so, i.e., until all outstanding data has beenDATA_ACKed,DATA_ACKed or until the segment with the DATA_FIN flag set is the only outstanding segment.</t> <t>Once a DATA_FIN has been acknowledged, all remaining subflowsMUST<bcp14>MUST</bcp14> be closed with standard FIN exchanges. Both hostsSHOULD<bcp14>SHOULD</bcp14> send FINs on all subflows, as acourtesycourtesy, to allow middleboxes to clean up state even if an individual subflow has failed.It is also encouraged to reduceReducing the timeouts(Maximum Segment Lifetime)(MSL) on subflows at end hosts after receiving aDATA_FIN.DATA_FIN is also encouraged. In particular, any subflows where there is still outstanding data queued (which has been retransmitted on other subflows in order to get the DATA_FIN acknowledged)MAY<bcp14>MAY</bcp14> be closed with a RST with an MP_TCPRST (<xreftarget="sec_reset"/>)target="sec_reset" format="default"/>) error code for "too much outstanding data".</t> <t>A connection is considered closed once both hosts' DATA_FINs have been acknowledged by DATA_ACKs.</t> <t>As specified above, a standard TCP FIN on an individual subflow only shuts down the subflow on which it was sent. If all subflows have been closed with a FINexchange,exchange but no DATA_FIN has been received and acknowledged, the MPTCP connection is treated as closed only after a timeout. This implies that an implementation will have TIME_WAIT states at both the subflow level and the connectionlevelslevel (see <xreftarget="app_fsm"/>).target="app_fsm" format="default"/>). This permits "break-before-make" scenarios where connectivity is lost on all subflows before a new one can bere-established.</t>re‑established.</t> </section> <sectiontitle="Receiver Considerations" anchor="sec_rwin">anchor="sec_rwin" numbered="true" toc="default"> <name>Receiver Considerations</name> <t>Regular TCP advertises a receive window in each packet, telling the sender how much data the receiver is willing to accept past the cumulativeack.ACK. The receive window is used to implement flow control, throttling down fast senders when receivers cannot keep up. </t> <t>MPTCP also uses a unique receive window, shared between the subflows. The idea is to allow any subflow to send data as long as the receiver is willing to accept it. Thealternative,alternative -- maintainingper subflowper-subflow receivewindows,windows -- could end up stalling some subflows while others would not use up their window.</t> <t>The receive window is relative to the DATA_ACK. As in TCP, a receiverMUST NOT<bcp14>MUST NOT</bcp14> shrink the right edge of the receive window (i.e., DATA_ACK + receive window). The receiver will use the data sequence number to tell if a packet should be accepted at the connection level.</t> <t>When deciding to accept packets at the subflow level, regular TCP checks the sequence number in the packet against the allowed receive window. Withmultipath,MPTCP, such a check is done using only the connection-level window. A sanity checkSHOULD<bcp14>SHOULD</bcp14> be performed at the subflow level to ensure that the subflow and mapped sequence numbers meet the following test: SSN - SUBFLOW_ACK <= DSN - DATA_ACK, where SSN is the subflow sequence number of the received packet and SUBFLOW_ACK is the RCV.NXT (next expected sequence number) of the subflow (with the equivalent connection-level definitions for DSN and DATA_ACK).</t> <t>In regular TCP, once a segment is deemed in-window, it is puteitherin either the in-order receive queue orinthe out-of-order queue. In Multipath TCP, the samehappensthing happens, but at the connection level: a segment is placed in theconnection levelconnection-level in-order or out-of-order queue if it is in-window at both the connection level and the subflowlevels.level. The stack still has to remember, for each subflow, which segments were received successfully so that it can ACK them at the subflow level appropriately. Typically, this will be implemented by keepingper subflowper-subflow out-of-order queues (containing only messageheaders,headers -- not the payloads) and remembering the value of the cumulative ACK. </t> <t>It is important for implementers to understand how large areceiverreceive buffer is appropriate. The lower bound for full network utilization is the maximum bandwidth-delay product of any one of the paths. However, this might be insufficient when a packet is lost on a slower subflow and needs to be retransmitted (see <xreftarget="sec_retransmit"/>).target="sec_retransmit" format="default"/>). A tight upper bound would be the maximum round-trip time (RTT) of any path multiplied by the total bandwidth available across all paths. This permits all subflows to continue at full speed while a packet is fast-retransmitted on the maximum RTT path. Even this might be insufficient to maintain full performance in the event of a retransmit timeout on the maximum RTT path.It is for future study to determineDetermining the relationship between retransmission strategies and receive buffersizing.</t>sizing is left for future study.</t> </section> <sectiontitle="Sender Considerations" anchor="sec_sender">anchor="sec_sender" numbered="true" toc="default"> <name>Sender Considerations</name> <t>The sender remembersreceiverreceive window advertisements from the receiver. It should only update its local receive window values when the largest sequence number allowed (i.e., DATA_ACK + receive window)increases,increases on the receipt of a DATA_ACK. This is importantto allow usingfor allowing the use of paths with differentRTTs,RTTs and thus different feedback loops. </t> <t>MPTCP uses a single receive window across all subflows, and if the receive window was guaranteed to be unchangedend-to-end,end to end, a host could always read the most recent receive window value. However, some classes of middleboxes may alter the TCP-level receive window. Typically, these will shrink the offered window, although for short periods of time it may be possible for the window to be larger (however, note that this would not continue for longperiodsperiods, since ultimately the middlebox must keep up with delivering data to the receiver). Therefore, if receive window sizes differ on multiple subflows, when sending data MPTCPSHOULD<bcp14>SHOULD</bcp14> take the largest of the most recent window sizes as the one to use in calculations. This rule is implicit in the requirement not to reduce the right edge of the window.</t> <t>The senderMUST<bcp14>MUST</bcp14> also remember the receive windows advertised by each subflow. The allowed window for subflow i is (ack_i, ack_i + rcv_wnd_i), where ack_i is the subflow-level cumulative ACK of subflow i. This ensures that data will not be sent to a middlebox unless there is enough buffering for the data. </t> <t>Putting the two rules together, we get the following: a sender is allowed to send data segments with data-level sequence numbers between (DATA_ACK, DATA_ACK + receive_window). Each of these segments will be mapped onto subflows, as long as subflow sequence numbers are in the allowed windows for those subflows. Note that subflow sequence numbers do not generally affect flow control if the same receive window is advertised across all subflows. They will perform flow control for those subflows with a smaller advertised receive window. </t> <t>The send bufferMUST,<bcp14>MUST</bcp14>, at a minimum, be as big as the receive buffer, to enable the sender to reach maximum throughput.</t> </section> <sectiontitle="Reliabilityanchor="sec_retransmit" numbered="true" toc="default"> <name>Reliability andRetransmissions" anchor="sec_retransmit">Retransmissions</name> <t>Thedata sequence mappingData Sequence Mapping allows senders to resend data with the same data sequence number on a different subflow. When doing this, a hostMUST<bcp14>MUST</bcp14> still retransmit the original data on the original subflow, in order to preserve thesubflowsubflow's integrity (middleboxes could replay olddata, and/ordata and&wj;/or could reject holes in subflows), and a receiver will ignore these retransmissions. While this is clearly suboptimal, for compatibility reasons this is sensible behavior. Optimizations could be negotiated in future versions of this protocol. Note also that this property would also permit a sender to always send the same data, with the same data sequence number, on multiple subflows, if desired for reliability reasons.</t> <t>This protocol specification does not mandate any mechanisms for handling retransmissions, and much will be dependent upon local policy (as discussed in <xreftarget="sec_policy"/>).target="sec_policy" format="default"/>). One can imagine aggressive connection-levelretransmissionsretransmission policies where every packet lost at the subflow level is retransmitted on a different subflow(hence,(hence wasting bandwidth but possibly reducing application-to-applicationdelays),delays) or conservative retransmission policies where connection-levelretransmitsretransmissions are only used after a few subflow-level retransmission timeouts occur.</t> <t>It is envisaged that a standard connection-level retransmission mechanism would be implemented around a connection-level data queue: all segments that haven't been DATA_ACKed are stored. A timer is set when the head of theconnection-levelconnection level is ACKed at the subflow level butits corresponding datais notACKedDATA_ACKed at the data level. This timer will guard againstfailures inretransmission failures by middleboxes that proactively ACK data.</t> <t>The senderMUST<bcp14>MUST</bcp14> keep data in its send buffer as long as the data has not been acknowledgedatboth (1) at the connection level andon(2) on all subflows on which it has been sent. In this way, the sender can always retransmit the data if needed, on the same subflow or on a different one. A special case is when a subflow fails: the sender will typically resend the data on other working subflows after atimeout,timeout and will keep trying to retransmit the data on the failed subflow too. The sender will declare the subflow failed after a predefined upper bound on retransmissions is reached (whichMAY<bcp14>MAY</bcp14> be lower than the usual TCP limits of theMaximum Segment Life),MSL) or on the receipt of an ICMP error, and only then delete the outstanding data segments. </t> <t>If multiple retransmissionsare triggeredthat indicate that a subflowperforms badly,is performing badly are triggered, thisMAY<bcp14>MAY</bcp14> lead to a host resetting the subflow with a RST. However, additional research is required to understand the heuristics of how and when to reset underperforming subflows. For example, a highly asymmetric path may be misdiagnosed as underperforming. A RST for this purposeSHOULD<bcp14>SHOULD</bcp14> be accompaniedwithby an "Unacceptable performance" MP_TCPRST option (<xreftarget="sec_reset"/>).</t>target="sec_reset" format="default"/>).</t> </section> <sectiontitle="Congestionanchor="sec_cc" numbered="true" toc="default"> <name>Congestion ControlConsiderations" anchor="sec_cc">Considerations</name> <t>Different subflows in an MPTCP connection have different congestion windows. To achieve fairness at bottlenecks and resource pooling, it is necessary to couple the congestion windows in use on each subflow, in order to push most traffic to uncongested links. One algorithm for achieving this is presented in <xreftarget="RFC6356"/>;target="RFC6356" format="default"/>; the algorithm does not achieve perfect resource pooling but is "safe" in that it is readily deployable in the current Internet. Bythis,this we mean that it does not take up more capacity on any one path than if it was a single path flow using only that route, so this ensures fair coexistence with single-path TCP at shared bottlenecks.</t> <t>It is foreseeable that different congestion controllers will be implemented for MPTCP, each aiming to achieve different properties in the resourcepooling/fairness/stabilitypooling / fairness / stability design space, as well as those for achieving different properties in quality of service, reliability, and resilience.</t> <t>Regardless of the algorithm used, the design oftheMPTCPprotocolaims to provide the congestion control implementations with sufficient information totakemake the right decisions; this information includes, for each subflow, which packets were lost and when. </t> </section> <sectiontitle="Subflow Policy" anchor="sec_policy">anchor="sec_policy" numbered="true" toc="default"> <name>Subflow Policy</name> <t>Within a local MPTCP implementation, a host may use any local policy it wishes to decide how to share the traffic to be sent over the available paths.</t> <t>In the typical use case, where the goal is to maximize throughput, all available paths will be used simultaneously for data transfer, using coupled congestion control as described in <xreftarget="RFC6356"/>.target="RFC6356" format="default"/>. It is expected, however, that other use cases will appear.</t> <t>For instance,aone possibility is an'all-or-nothing'"all-or-nothing" approach, i.e., have a second path ready for use in the event of failure of the first path, but alternatives could include entirely saturating one path before using an additional path (the'overflow'"overflow" case). Such choices would be most likely based on the monetary cost oflinks,links but may also be based on properties such as the delay or jitter of links, where stability (of delay or bandwidth) is more important than throughput. Application requirements such as these are discussed in detail in <xreftarget="RFC6897"/>.</t>target="RFC6897" format="default"/>.</t> <t>The ability to make effective choices at the sender requires full knowledge of the path "cost", which is unlikely to be the case. It would be desirable for a receiver to be able to signal their own preferences for paths, since they will often be the multihomedparty,party and may have to pay for metered incoming bandwidth.</t> <t>To enablethis,this behavior, the MP_JOIN option (see <xreftarget="sec_join"/>)target="sec_join" format="default"/>) contains the'B' bit,"B" bit, which allows a host to indicate to its peer that this path should be treated as a backup path to use only in the event of failure of other working subflows (i.e., a subflow where the receiver has indicated that B=1SHOULD NOT<bcp14>SHOULD NOT</bcp14> be used to send data unless there are no usable subflows where B=0).</t> <t>In the event that the available set of paths changes, a host may wish to signal a change in priority of subflows to the peer (e.g., a subflow that was previously set as a backup should now take priority over all remaining subflows). Therefore, the MP_PRIO option, shown in <xreftarget="tcpm_prio"/>,target="tcpm_prio" format="default"/>, can be used to change the'B'"B" flag of the subflow on which it is sent.</t><t>Another use of the MP_PRIO option is to set the 'B' flag on a subflow to cleanly retire its use before closing it and removing it with REMOVE_ADDR <xref target="sec_remove_addr"/>, for example to support make-before-break session continuity, where new subflows are added before the previously used ones are closed.</t> <?rfc needLines='8'?><figurealign="center" anchor="tcpm_prio" title="Changeanchor="tcpm_prio"> <name>Change Subflow Priority (MP_PRIO)Option">Option</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-----+-+ | Kind | Length |Subtype|(rsv)|B| +---------------+---------------+-------+-----+-+ ]]></artwork> </figure> <t>Another use of the MP_PRIO option is to set the "B" flag on a subflow to cleanly "retire" its use before closing it and removing it with REMOVE_ADDR (<xref target="sec_remove_addr" format="default"/>) -- for example, to support make-before-break session continuity, where new subflows are added before the previously used subflows are closed.</t> <t>It should be noted that the backup flag is a request from a data receiver to a data sender only, and the data senderSHOULD<bcp14>SHOULD</bcp14> adhere to these requests. A host cannot assume that the data sender will do so, however, since local policies -- or technical difficulties -- may override MP_PRIO requests. Note also that this signal applies to a single direction, and so the sender of this option could choose to continue using the subflow to send data even if it has signaled B=1 to the other host.</t> </section> </section> <sectiontitle="Addressanchor="sec_pm" numbered="true" toc="default"> <name>Address Knowledge Exchange (PathManagement)" anchor="sec_pm">Management)</name> <t>We use the term "path management" to refer to the exchange of information about additional paths between hosts, which in this design is managed by multiple addresses at hosts. For moredetail ofdetails regarding the architectural thinking behind this design, see the MPTCPArchitecturearchitecture document <xreftarget="RFC6182"/>.</t>target="RFC6182" format="default"/>.</t> <t>This design makes use of two methods of sharing such information, and both can be used on a connection. The first is the direct setup of newsubflows, already describedsubflows (described in <xreftarget="sec_join"/>,target="sec_join" format="default"/>), where the initiator has an additional address. The secondmethod, describedmethod (described in the followingsubsections,subsections) signals addresses explicitly to the other host to allow it to initiate new subflows. The two mechanisms are complementary: the first is implicit and simple, while theexplicitsecond (explicit) is more complex but is more robust. Together,thethese mechanisms allow addresses to change in flight (and thus support operation through NATs, since the source address need not beknown), andknown); they also allow the signaling of previously unknownaddresses,addresses and of addresses belonging to other address families (e.g., both IPv4 and IPv6).</t> <t>Here is an example of typical operation of the protocol:<list style="symbols"> <t>An</t> <ul spacing="normal"> <li>An MPTCP connection is initially set up betweenaddress/portaddress&wj;/port A1 of Host A andaddress/portaddress&wj;/port B1 of HostB. B. If Host A is multihomed and multiaddressed, it can start an additional subflow from its address A2 to B1, by sending a SYN witha Joinan MP_JOIN option from A2 to B1, using B's previously declared token for this connection. Alternatively, if B is multihomed, it can try to set up a new subflow from B2 to A1, using A's previously declared token. In either case, the SYN will be sent to the port already in use for the original subflow on the receivinghost.</t> <t>Simultaneouslyhost.</li> <li>Simultaneously (or after a timeout), an ADD_ADDR option (<xreftarget="sec_add_address"/>)target="sec_add_address" format="default"/>) is sent on an existing subflow, informing the receiver of the sender's alternative address(es). The recipient can use this information to open a new subflow to the sender's additionaladdress.address(es). In our example, A will send the ADD_ADDR option informing B ofaddress/portaddress&wj;/port A2. The mix of using theSYN-basedSYN‑based option and the ADD_ADDR option, including timeouts, is implementation specific and can be tailored to agree with localpolicy.</t> <t>Ifpolicy.</li> <li>If subflow A2-B1 is successfully set up, Host B can use the Address ID in theJoinMP_JOIN option to correlate this source address with the ADD_ADDR option that will also arrive on an existing subflow; now B knows not to open A2-B1, ignoring the ADD_ADDR. Otherwise, if B has not received the A2-B1 MP_JOIN SYN but received the ADD_ADDR, it can try to initiate a new subflow from one or more of its addresses to address A2. This permits new sessions to be opened if one host is behind aNAT.</t> </list>NAT.</li> </ul> <t> Other ways of using the two signaling mechanisms are possible; for instance, signaling addresses in other address families can only be done explicitly using the Add Address (ADD_ADDR) option. </t> <sectiontitle="Address Advertisement" anchor="sec_add_address">anchor="sec_add_address" numbered="true" toc="default"> <name>Address Advertisement</name> <t>TheAdd Address (ADD_ADDR)ADD_ADDR MPTCP option announces additional addresses(and(and, optionally, ports) on which a host can be reached (<xreftarget="tcpm_address"/>).target="tcpm_address" format="default"/>). This option can be used at any time during a connection, depending on when the sender wishes to enable multiple pathsand/orand&wj;/or when paths become available. As with all MPTCP signals, the receiverMUST<bcp14>MUST</bcp14> undertake standard TCP validity checks,e.g.e.g., per <xreftarget="RFC5961"/>,target="RFC5961" format="default"/>, before actingupon it.</t>upon it.</t> <figure anchor="tcpm_address"> <name>Add Address (ADD_ADDR) Option</name> <artwork align="left" name="" type="" alt=""><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-------+---------------+ | Kind | Length |Subtype|(rsv)|E| Address ID | +---------------+---------------+-------+-------+---------------+ | Address (IPv4: 4 octets / IPv6: 16 octets) | +-------------------------------+-------------------------------+ | Port (2 octets, optional) | | +-------------------------------+ | | Truncated HMAC (8 octets, if E=0) | | +-------------------------------+ | | +-------------------------------+ ]]></artwork> </figure> <t>Every address has an Address ID that can be used for uniquely identifying the address within a connection for address removal. The Address ID is also used to identify MP_JOIN options (see <xreftarget="sec_join"/>)target="sec_join" format="default"/>) relating to the same address, even when address translators are in use. The Address IDMUST<bcp14>MUST</bcp14> uniquely identify the address for the sender of the option (within the scope of theconnection), butconnection); the mechanism for allocating such IDs is implementation specific.</t> <t>AlladdressAddress IDs learned via either MP_JOIN or ADD_ADDRSHOULD<bcp14>SHOULD</bcp14> be stored by the receiver in a data structure that gathers all theAddress ID to addressAddress-ID-to-address mappings for a connection (identified by a token pair). In this way, there is a stored mapping between the Address ID, observed source address, and token pair for future processing of control information for a connection. Note that an implementationMAY<bcp14>MAY</bcp14> discard incoming address advertisements atwill,will -- for example,for avoidingto avoid updating mappingstate,state or because advertised addresses are of no use to it (for example, IPv6 addresses when it has IPv4 only). Therefore, a hostMUST<bcp14>MUST</bcp14> treat address advertisements as soft state, and itMAY<bcp14>MAY</bcp14> choose to refresh advertisements periodically. Note also that an implementationMAY<bcp14>MAY</bcp14> choose to cache these address advertisements even if they are not currently relevant but may be relevant in the future, such as IPv4 addresses when IPv6 connectivity is available but IPv4 is awaiting DHCP.</t> <t>This option is shown in <xreftarget="tcpm_address"/>.target="tcpm_address" format="default"/>. The illustration is sized for IPv4 addresses. For IPv6, the length of the address will be16 octets16 octets (instead of 4).</t> <t>The 2 octets that specify the TCP port number to use areoptionaloptional, and their presence can be inferred from the length of the option. Although it is expected that the majority of use cases will use the same port pairs as those used for the initial subflow (e.g., port 80 remains port 80 on all subflows, as does the ephemeral port at the client), there may be cases (such as port-based load balancing) where the explicit specification of a different port is required. If no port is specified, MPTCPSHOULD<bcp14>SHOULD</bcp14> attempt to connect to the specified address on the same port as the port that is already in use by the subflow on which the ADD_ADDR signal was sent; this is discussed in more detail in <xreftarget="heuristics"/>.</t>target="heuristics" format="default"/>.</t> <t>The Truncated HMAC parameter present in thisOptionoption is the rightmost 64 bits of an HMAC, negotiated and calculated in the same way as for MP_JOIN as described in <xreftarget="sec_join"/>.target="sec_join" format="default"/>. For this specification of MPTCP, as there is only one hash algorithm option specified, this will be HMAC as defined in <xreftarget="RFC2104"/>,target="RFC2104" format="default"/>, using the SHA-256 hash algorithm <xreftarget="RFC6234"/>.target="RFC6234" format="default"/>. In the same way as for MP_JOIN, the key for the HMAC algorithm, in the case of the message transmitted by Host A, will be Key-A followed by Key-B, and in the case of Host B, Key-B followed by Key-A. These are the keys that were exchanged in the original MP_CAPABLE handshake. The message for the HMAC is the Address ID, IPAddress,address, andPort whichport that precede the HMAC in the ADD_ADDR option. If the port is not present in the ADD_ADDR option, the HMAC message will nevertheless includetwo2 octets of value zero. The rationale for the HMAC is to prevent unauthorized entities from injecting ADD_ADDR signals in an attempt to hijack a connection. Notethat additionallythat, additionally, the presence of this HMAC prevents the address from being changed in flight unless the key is known by an intermediary. If a host receives an ADD_ADDR option for which it cannot validate the HMAC, itSHOULD<bcp14>SHOULD</bcp14> silently ignore the option.</t> <t>A set of four flagsareis present after the subtype and before the Address ID. Only the rightmost bit- labelled 'E' --- labeled "E" -- is assigned in this specification. The other bits are currentlyunassigned and MUSTunassigned; they <bcp14>MUST</bcp14> be set tozero0 by a sender andMUST<bcp14>MUST</bcp14> be ignored by the receiver.</t> <t>The'E'"E" flag exists to provide reliability for this option. Because this option will often be sent on pure ACKs, there is no guarantee of reliability. Therefore, a receiver receiving a fresh ADD_ADDR option (whereE=0),E=0) will send the same option back to the sender, but not including theHMAC,HMAC and with E=1, to indicate receipt.TheAccording to local policy, the lack of thisechotype of "echo" canbe used byindicate to the initial ADD_ADDR senderto retransmitthat the ADD_ADDRaccordingneeds tolocal policy.</t> <?rfc needLines='11'?> <figure align="center" anchor="tcpm_address" title="Add Address (ADD_ADDR) Option"> <artwork align="left"><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-------+---------------+ | Kind | Length |Subtype|(rsv)|E| Address ID | +---------------+---------------+-------+-------+---------------+ | Address (IPv4 - 4 octets / IPv6 - 16 octets) | +-------------------------------+-------------------------------+ | Port (2 octets, optional) | | +-------------------------------+ | | Truncated HMAC (8 octets, if E=0) | | +-------------------------------+ | | +-------------------------------+ ]]></artwork> </figure>be retransmitted.</t> <t>Due to the proliferation of NATs, it is reasonably likely that one host may attempt to advertise private addresses <xreftarget="RFC1918"/>.target="RFC1918" format="default"/>. It is not desirable to prohibitthis,this behavior, since there may be cases where both hosts have additional interfaces on the same private network, and a hostMAY<bcp14>MAY</bcp14> advertise such addresses. The MP_JOIN handshake to create a new subflow (<xreftarget="sec_join"/>)target="sec_join" format="default"/>) provides mechanisms to minimize security risks. The MP_JOIN message contains a 32-bit token that uniquely identifies the connection to the receiving host. If the token is unknown, the host willreturnrespond with a RST. In the unlikely event that the token is valid at the receiving host, subflow setup will continue, but the HMAC exchange must occur for authentication.ThisThe HMAC exchange willfail,fail and will provide sufficient protection against two unconnected hosts accidentally setting up a new subflow upon the signal of a private address. Further security considerations around the issue of ADD_ADDR messages that accidentally misdirect, or maliciously direct, new MP_JOIN attempts are discussed in <xreftarget="sec_security"/>.</t>target="sec_security" format="default"/>.</t> <t>A host that receives an ADD_ADDR but finds that a connection set up to that IP address and port number is unsuccessfulSHOULD NOT<bcp14>SHOULD NOT</bcp14> perform further connection attempts to thisaddress/portaddress&wj;/port combination for this connection. A sender that wants to trigger a new incoming connection attempt on a previously advertisedaddress/portaddress&wj;/port combination can therefore refresh ADD_ADDR information by sending the option again.</t> <t>A host can therefore send an ADD_ADDR message with analready assignedalready-assigned Address ID, but theAddress MUSTaddress <bcp14>MUST</bcp14> be the same as the address previously assigned to this Address ID. A new ADD_ADDR may have thesame,same port number ordifferent,a different port number. If the port number is different, the receiving hostSHOULD<bcp14>SHOULD</bcp14> try to set up a new subflow to this newaddress/portaddress&wj;/port combination.</t> <t>A host wishing to replace an existing Address IDMUST<bcp14>MUST</bcp14> first remove the existing one (<xreftarget="sec_remove_addr"/>).</t>target="sec_remove_addr" format="default"/>).</t> <t>During normal MPTCP operation, it is unlikely that there will be sufficient TCP option space for ADD_ADDR to be included along with those for data sequence numbering (<xreftarget="sec_dsn"/>).target="sec_dsn" format="default"/>). Therefore, it is expected that an MPTCP implementation will send the ADD_ADDR option on separate ACKs. As discussed earlier, however, an MPTCP implementationMUST NOT<bcp14>MUST NOT</bcp14> treat duplicate ACKs with any MPTCP option, with the exception of the DSS option, as indications of congestion <xreftarget="RFC5681"/>,target="RFC5681" format="default"/>, and an MPTCP implementationSHOULD NOT<bcp14>SHOULD NOT</bcp14> send more than two duplicate ACKs in a row for signaling purposes.</t> </section> <sectiontitle="Remove Address" anchor="sec_remove_addr">anchor="sec_remove_addr" numbered="true" toc="default"> <name>Remove Address</name> <t>If, during the lifetime of an MPTCP connection, a previously announced address becomes invalid (e.g., if the interfacedisappears,disappears or an IPv6 address is no longer preferred), the affected hostSHOULD<bcp14>SHOULD</bcp14> announce this situation so that the peer can remove subflows related to this address. Even if an address is not in use byaan MPTCP connection, if it has been previously announced, an implementationSHOULD<bcp14>SHOULD</bcp14> announce its removal. A hostMAY<bcp14>MAY</bcp14> also choose to announce that a valid IP address should not be used anylonger,longer -- forexampleexample, formake-before-breakmake‑before-break session continuity.</t> <t>This is achieved through the Remove Address (REMOVE_ADDR) option (<xreftarget="tcpm_remove"/>),target="tcpm_remove" format="default"/>), which will remove a previously added address (or list of addresses) from a connection and terminate any subflows currently using that address.</t> <figure anchor="tcpm_remove"> <name>Remove Address (REMOVE_ADDR) Option</name> <artwork align="left" name="" type="" alt=""><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-------+---------------+ | Kind |Length = 3 + n |Subtype|(resvd)| Address ID | ... +---------------+---------------+-------+-------+---------------+ (followed by n-1 Address IDs, if required) ]]></artwork> </figure> <t>For security purposes, if a host receives a REMOVE_ADDR option, it must ensure that the affectedpath(s)path or paths are no longer in use before it instigates closure. The receipt of REMOVE_ADDRSHOULD<bcp14>SHOULD</bcp14> first trigger the sending of a TCP keepalive <xreftarget="RFC1122"/>target="RFC1122" format="default"/> on the path, and if a response isreceivedreceived, the pathSHOULD NOT<bcp14>SHOULD NOT</bcp14> be removed. If the path is found to still be alive, the receiving hostSHOULD<bcp14>SHOULD</bcp14> no longer use the specified address for future connections, but it is the responsibility of the hostwhichthat sent the REMOVE_ADDR to shut down the subflow.TheBefore the address is removed, the requesting hostMAY<bcp14>MAY</bcp14> also use MP_PRIO (<xreftarget="sec_policy"/>)target="sec_policy" format="default"/>) to request that a pathisno longerused, before removal.be used. Typical TCP validity tests on the subflow (e.g., ensuring that sequence and ACK numbers are correct)MUST<bcp14>MUST</bcp14> also be undertaken. An implementation can use indications of these test failures as part of intrusion detection or error logging.</t> <t>The sending and receipt (if no keepalive response was received) of this messageSHOULD<bcp14>SHOULD</bcp14> trigger the sending of RSTs by both hosts on the affected subflow(s) (if possible), as acourtesycourtesy, tocleaning upallow the cleanup of middleboxstate,state before cleaning up any local state.</t> <t>Address removal is undertakenbyaccording to the Address ID, so as to permit the use of NATs and other middleboxes that rewrite source addresses. Iftherean Address ID isno address at the requested ID,not known, the receiver will silently ignore the request.</t> <t>A subflow that is still functioningMUST<bcp14>MUST</bcp14> be closed with a FIN exchange as in regular TCP, rather than using this option. For more information, see <xreftarget="sec_close"/>.</t> <?rfc needLines='8'?> <figure align="center" anchor="tcpm_remove" title="Remove Address (REMOVE_ADDR) Option"> <artwork align="left"><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-------+---------------+ | Kind | Length = 3+n |Subtype|(resvd)| Address ID | ... +---------------+---------------+-------+-------+---------------+ (followed by n-1 Address IDs, if required) ]]></artwork> </figure>target="sec_close" format="default"/>.</t> </section> </section> <sectiontitle="Fast Close" anchor="sec_fastclose">anchor="sec_fastclose" numbered="true" toc="default"> <name>Fast Close</name> <t>Regular TCP has the means of sending areset (RST)RST signal to abruptly close a connection. With MPTCP, a regular RST only has the scope of thesubflow andsubflow; it will only close theconcernedapplicable subflowbutand will not affect the remaining subflows. MPTCP's connection will stay alive at the data level, in order to permit break-before-make handover between subflows. It is therefore necessary to provide an MPTCP-level "reset" to allow the abrupt closure of the whole MPTCPconnection, andconnection; this is done via the MP_FASTCLOSE option.</t> <t>MP_FASTCLOSE is used to indicate to the peer that the connection will be abruptly closed and no data will be accepted anymore. The reasons for triggering an MP_FASTCLOSE are implementation specific. Regular TCP does not allow the sending of a RST while the connection is in a synchronized state <xreftarget="RFC0793"/>.target="RFC0793" format="default"/>. Nevertheless, implementations allow the sending of a RST in thisstate,state if, for example, the operating system is running out of resources. In these cases, MPTCP should send the MP_FASTCLOSE. This option is illustrated in <xreftarget="tcpm_fastclose"/>.</t> <?rfc needLines='12'?>target="tcpm_fastclose" format="default"/>.</t> <figurealign="center" anchor="tcpm_fastclose" title="Fastanchor="tcpm_fastclose"> <name>Fast Close (MP_FASTCLOSE)Option">Option</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-----------------------+ | Kind | Length |Subtype| (reserved) | +---------------+---------------+-------+-----------------------+ | Option Receiver's Key | | (64 bits) | | | +---------------------------------------------------------------+ ]]></artwork> </figure> <t>If Host A wants to force the closure of an MPTCP connection, ithascan do so via twodifferentoptions:<list style="symbols"> <t>Option</t> <ul spacing="normal"> <li>Option A(ACK) :(ACK): Host A sends an ACK containing the MP_FASTCLOSE option on one subflow, containing the key of Host B as declared in the initial connection handshake. On all the other subflows,Host AHost A sends a regular TCP RST to close thesesubflows,subflows and tears them down. Host A now enters FASTCLOSE_WAITstate.</t> <t>Optionstate.</li> <li>Option R(RST) :(RST): Host A sends a RST containing the MP_FASTCLOSE option on all subflows, containing the key of Host B as declared in the initial connection handshake. Host A can tear down the subflows and the connectiondown immediately.</t> </list> </t>immediately.</li> </ul> <t>IfhostHost A decides to force the closure by using Option A and sending an ACK with the MP_FASTCLOSE option, the connection shall proceed as follows:<list style="symbols"> <t>Upon</t> <ul spacing="normal"> <li>Upon receipt of an ACK with MP_FASTCLOSE by Host B, containing the valid key, Host B answers on the same subflow with a TCP RST and tears down all subflows also through sending TCP RST signals. Host B can now close the whole MPTCP connection (it transitions directly to CLOSEDstate).</t> <t>Asstate).</li> <li>As soon as Host A has received the TCP RST on the remaining subflow, it can close this subflow and tear down the whole connection (transition from FASTCLOSE_WAIT state to CLOSEDstates).state). If Host A receives an MP_FASTCLOSE instead of a TCP RST, both hosts attempted fast closure simultaneously. Host A should reply with a TCP RST and tear down theconnection.</t> <t>Ifconnection.</li> <li>If Host A does not receive a TCP RST in reply to its MP_FASTCLOSE after one retransmission timeout (RTO) (the RTO of the subflow where the MP_FASTCLOSE has been sent), itSHOULD<bcp14>SHOULD</bcp14> retransmit the MP_FASTCLOSE.The number of retransmissions SHOULD be limited to avoidTo keep this connection from being retained for a long time,butthe number of retransmissions <bcp14>SHOULD</bcp14> be limited; this limit is implementation specific. ARECOMMENDED<bcp14>RECOMMENDED</bcp14> number is 3. If no TCP RST is received in response, Host ASHOULD<bcp14>SHOULD</bcp14> send a TCP RST with the MP_FASTCLOSE option itself when it releases state in order to clear any remaining state atmiddleboxes.</t> </list> </t> <t>If however hostmiddleboxes.</li> </ul> <t>If, however, Host A decides to force the closure by using Option R and sending a RST with the MP_FASTCLOSE option, Host B will act as follows:Uponupon receipt of a RST with MP_FASTCLOSE, containing the valid key, Host B tears down all subflows by sending a TCP RST.Host BHost B can now close the whole MPTCP connection (it transitions directly to CLOSED state).</t> </section> <sectiontitle="Subflow Reset" anchor="sec_reset">anchor="sec_reset" numbered="true" toc="default"> <name>Subflow Reset</name> <t>An implementation of MPTCP may also need to send a regular TCP RST to force the closure of a subflow. A host sends a TCP RST in order to close a subflow or reject an attempt to open a subflow (MP_JOIN). In order toinformlet the receiving host know why a subflow is being closed or rejected, the TCP RST packetMAY<bcp14>MAY</bcp14> include the MP_TCPRSTOption.option (<xref target="tcpm_reset"/>). The hostMAY<bcp14>MAY</bcp14> use this information to decide, for example, whether it tries to re-establish the subflow immediately, later, or never.</t><?rfc needLines='8'?><figurealign="center" anchor="tcpm_reset" title="TCPanchor="tcpm_reset"> <name>TCP RST Reason (MP_TCPRST)Option">Option</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+-----------------------+ | Kind | Length |Subtype|U|V|W|T| Reason | +---------------+---------------+-------+-----------------------+ ]]></artwork> </figure> <t>The MP_TCPRST option contains a reason code that allows the sender of the option to provide more information about the reason for the termination of the subflow. Using 12 bits of option space, the firstfour bits4 bits are reserved for flags (only one of which is currently defined), and the remaining octet is used to express a reason code for this subflow termination, from which a receiverMAY<bcp14>MAY</bcp14> infer information about the usability of this path.</t> <t>The "T" flag is used by the sender to indicate whether the error condition that is reported is Transient(T("T" bit set to 1) or Permanent(T("T" bit set to 0). If the error condition is considered to be Transient by the sender of the RST segment, the recipient of this segmentMAY<bcp14>MAY</bcp14> try toreestablishre-establish a subflow for this connection over the failed path. The time at which a receiver may try tore-establishre‑establish this subflow isimplementation-specific,implementation specific butSHOULD<bcp14>SHOULD</bcp14> take into account the properties of the failure as defined by thefollowingprovided reason code. If the error condition is considered to bepermanent,Permanent, the receiver of the RST segmentSHOULD NOT<bcp14>SHOULD NOT</bcp14> try toreestablishre‑establish a subflow for this connection over this path. The "U","V""V", and "W" flags are not defined by this specification and are reserved for future use. An implementation of this specificationMUST<bcp14>MUST</bcp14> set these flags to 0, and a receiverMUST<bcp14>MUST</bcp14> ignore them.</t><t>The "Reason" code<t>"Reason" is an 8-bit field that indicates the reason code for the termination of the subflow. The following codes are defined in this document:<list style="symbols"> <t>Unspecified</t> <ul spacing="normal"> <li>Unspecified error (code0x0).0x00). This is the defaulterror implyingerror; it implies that the subflow is no longer available. The presence of this option shows that the RST was generated byaan MPTCP-awaredevice.</t> <t>MPTCP specificdevice.</li> <li>MPTCP-specific error (code 0x01). An error has been detected in the processing of MPTCP options. This is the usual reason code to return in the cases where a RST is being sent to close a subflowfor reasonsbecause of an invalidresponse.</t> <t>Lackresponse.</li> <li>Lack of resources (code 0x02). This code indicates that the sending host does not have enough resources to support the terminatedsubflow.</t> <t>Administrativelysubflow.</li> <li>Administratively prohibited (code 0x03). This code indicates that the requested subflow is prohibited by the policies of the sendinghost.</t> <t>Toohost.</li> <li>Too much outstanding data (code 0x04). This code indicates that there is an excessive amount of data thatneedneeds to be transmitted over the terminated subflow while having already been acknowledged over one or more other subflows. This may occur if a path has been unavailable for a short period and it is more efficient to reset and start again than it is to retransmit the queueddata.</t> <t>Unacceptabledata.</li> <li>Unacceptable performance (code 0x05). This code indicates that the performance of this subflow was too low compared to the other subflows of this Multipath TCPconnection.</t> <t>Middleboxconnection.</li> <li>Middlebox interference (code 0x06). Middlebox interference has been detected over thissubflowsubflow, making MPTCP signaling invalid. For example, this may be sent if the checksum does notvalidate.</t> </list> </t>validate.</li> </ul> </section> <sectiontitle="Fallback" anchor="sec_fallback">anchor="sec_fallback" numbered="true" toc="default"> <name>Fallback</name> <t>Sometimes, middleboxes will exist on a path that could prevent the operation of MPTCP. MPTCP has been designedin orderto cope with many middlebox modifications (see <xreftarget="sec_middleboxes"/>),target="sec_middleboxes" format="default"/>), but there are still some cases where a subflow could fail to operate within the MPTCP requirements.TheseNotably, these cases arenotablythe following: the loss of MPTCP options on apath,path and the modification of payload data. If such an event occurs, it is necessary to "fall back" to the previous, safe operation. This may be either falling back to regular TCP or removing a problematic subflow.</t> <t>At the start of an MPTCP connection (i.e., the first subflow), it is important to ensure that the path is fully MPTCP capable and the necessary MPTCP options can reach each host. The handshake as described in <xreftarget="sec_init"/> SHOULDtarget="sec_init" format="default"/> <bcp14>SHOULD</bcp14> fall back to regular TCP if either of the SYN messagesdodoes not have the MPTCP options: this is the same, and desired, behavior in the case where a host is not MPTCPcapable,capable or the path does not support the MPTCP options. When attempting to join an existing MPTCP connection (<xreftarget="sec_join"/>),target="sec_join" format="default"/>), if a path is not MPTCP capable and the MPTCP options do not get through on the SYNs, the subflow will be closed according to the MP_JOIN logic.</t> <t>There is, however, another corner case that should beaddressed. That is one ofaddressed: the case where MPTCP optionsgettingget through on theSYN,SYN but not on regular packets.This can be resolved ifIf the subflow is the firstsubflow,subflow and thus all data in flight is contiguous, this situation can be resolved by using the followingrules.</t> <t>Arules:</t> <ul spacing="normal"> <li>A senderMUST<bcp14>MUST</bcp14> include a DSS option withdata sequence mappingData Sequence Mapping in every segment until one of the sent segments has been acknowledged with a DSS option containing a Data ACK. Upon reception of the acknowledgment, the sender has the confirmation that the DSS option passes in both directions and may choose to send fewer DSS options than once persegment.</t> <t>If,segment.</li> <li>If, however, an ACK is received for data (not just for the SYN) without a DSS option containing a Data ACK, the sender determines that the path is not MPTCP capable. In the case of this occurring on an additional subflow (i.e., one started with MP_JOIN), the hostMUST<bcp14>MUST</bcp14> close the subflow with a RST, whichSHOULD<bcp14>SHOULD</bcp14> containaan MP_TCPRST option (<xreftarget="sec_reset"/>)target="sec_reset" format="default"/>) with a "Middlebox interference" reasoncode.</t> <t>Incode.</li> <li>In the case of such an ACK being received on the first subflow (i.e., that started with MP_CAPABLE), before any additional subflows are added, the implementationMUST<bcp14>MUST</bcp14> drop out ofanMPTCPmode,mode and fall back to regular TCP. The sender will send one finaldata sequence mapping,Data Sequence Mapping, with the Data-Level Length value of 0 indicating an infinite mapping (to inform the other end in case the path drops options in one direction only), and then revert to sending data on the single subflow without any MPTCPoptions.</t> <t>Ifoptions.</li> <li>If a subflow breaks during operation,e.g.e.g., if it isre-routedrerouted and MPTCP options are no longer permitted, then once this is detected (by the subflow-level receive buffer filling up, since there is no mapping available in order to DATA_ACK this data), the subflowSHOULD<bcp14>SHOULD</bcp14> be treated as broken and closed with a RST, since no data can be delivered to the applicationlayer,layer and no fallback signal can be reliably sent. This RSTSHOULD<bcp14>SHOULD</bcp14> include the MP_TCPRST option (<xreftarget="sec_reset"/>)target="sec_reset" format="default"/>) with a "Middlebox interference" reasoncode.</t>code.</li> </ul> <t>These rules should cover all cases where such a failure couldhappen:happen -- whether it's on the forward or reverse path and whether the server or the client first sends data.</t> <t>Sofarfar, this section has discussed the loss of MPTCP options, eitherinitially,initially or during the course of the connection. As described in <xreftarget="sec_generalop"/>,target="sec_generalop" format="default"/>, each portion of data for which there is a mapping is protected by a checksum, if checksums have been negotiated. This mechanism is used to detect if middleboxes have made any adjustments to the payload (added, removed, or changed data). A checksum will fail if the data has been changed in any way.ThisThe use of a checksum will also detectifwhether the length of data on the subflow is increased or decreased, and this means thedata sequence mappingData Sequence Mapping is no longer valid. The sender no longer knows what subflow-level sequence number the receiver is genuinely operating at (the middlebox will be faking ACKs in return), and it cannot signal any further mappings. Furthermore, in addition to the possibility of payload modifications that are valid at the application layer,thereit isthe possibilitypossible that such modifications could be triggered across MPTCP segment boundaries, corrupting the data. Therefore, all data from the start of the segment that failed the checksumonwardsonward is not trustworthy.</t> <t>Note that if checksum usage has not been negotiated, this fallback mechanism cannot be used unless there is somehigherhigher-layer orlower layerlower‑layer signal to inform the MPTCP implementation that the payload has been tampered with.</t> <t>When multiple subflows are in use, the data in flight on a subflow will likely involve data that is not contiguously part of the connection-level stream, since segments will be spread across the multiple subflows. Due to the problems identified above, it is not possible to determine whatadjustment hasadjustments have been done to the data (notably, any changes to the subflow sequence numbering). Therefore, it is not possible to recover the subflow, and the affected subflow must be immediately closed with aRST, featuringRST that includes an MP_FAIL option (<xreftarget="tcpm_fallback"/>),target="tcpm_fallback" format="default"/>), which defines the data sequence number at the start of the segment (defined by thedata sequence mapping)Data Sequence Mapping) that had the checksum failure. Note that the MP_FAIL option requires the use of the full 64-bit sequence number, even if 32-bit sequence numbers are normally in use in the DSS signals on the path.</t><?rfc needLines='8'?><figurealign="center" anchor="tcpm_fallback" title="Fallbackanchor="tcpm_fallback"> <name>Fallback (MP_FAIL)Option">Option</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------+---------------+-------+----------------------+ | Kind | Length=12 |Subtype| (reserved) | +---------------+---------------+-------+----------------------+ | | | Data Sequence Number (8 octets) | | | +--------------------------------------------------------------+ ]]></artwork> </figure> <t>The receiver of this optionMUST<bcp14>MUST</bcp14> discard all data following the data sequence number specified. Failed dataMUST NOT<bcp14>MUST NOT</bcp14> be DATA_ACKed and so will be retransmitted on other subflows (<xreftarget="sec_retransmit"/>).target="sec_retransmit" format="default"/>). </t> <t>A special case is when there is a single subflow and it fails with a checksum error. If it is known that all unacknowledged data in flight is contiguous (which will usually be the case with a single subflow), an infinite mapping can be applied to the subflow without the need to close it first,andessentiallyturnturning off all further MPTCP signaling. In this case, if a receiver identifies a checksum failure when there is only one path, it will send back an MP_FAIL option on the subflow-level ACK, referring to the data-level sequence number of the start of the segment on which the checksum error was detected. The sender will receivethis, andthis information and, if all unacknowledged data in flight is contiguous, will signal an infinite mapping. This infinite mapping will be a DSS option (<xreftarget="sec_generalop"/>)target="sec_generalop" format="default"/>) on the first new packet, containing adata sequence mappingData Sequence Mapping that acts retroactively, referring to the start of the subflow sequence number of the most recent segment that was known to be delivered intact(i.e.(i.e., was successfully DATA_ACKed). From that pointonwards,onward, data can be altered by a middlebox without affecting MPTCP, as the data stream is equivalent to a regular, legacy TCP session.WhilstWhile in theory paths may only be damaged in onedirection,direction -- and the MP_FAIL signal affects only one direction oftraffic,traffic -- forimplementation simplicity,simplicity of implementation, the receiver of an MP_FAILMUST<bcp14>MUST</bcp14> also respond with an MP_FAIL in the reverse direction and entirely revert to a regular TCP session.</t> <t>In the rare case that the data is not contiguous (which could happen when there is only one subflow but it is retransmitting data from a subflow that has recently been uncleanly closed), the receiverMUST<bcp14>MUST</bcp14> close the subflow with a RST with MP_FAIL. The receiverMUST<bcp14>MUST</bcp14> discard all data that follows the data sequence number specified. The senderMAY<bcp14>MAY</bcp14> attempt to create a new subflow belonging to the sameconnection,connection and, if it chooses to do so,SHOULD<bcp14>SHOULD</bcp14> immediately place the single subflowimmediatelyin single-path mode by setting an infinitedata sequence mapping.Data Sequence Mapping. This mapping will begin from the data-level sequence number that was declared in the MP_FAIL.</t> <t>After a sender signals an infinite mapping, itMUST<bcp14>MUST</bcp14> only use subflow ACKs to clear its send buffer. This is because Data ACKs may become misaligned with the subflow ACKs when middleboxes insert or delete data. Thereceive SHOULDreceiver <bcp14>SHOULD</bcp14> stop generating Data ACKs after it receives an infinitemapping. </t>mapping.</t> <t>When a connection has fallen back with an infinite mapping, only one subflow can send data; otherwise, the receiver would not know how to reorder the data. In practice, this means that all MPTCP subflows will have to be terminated except one. Once MPTCP falls back to regular TCP, itMUST NOT<bcp14>MUST NOT</bcp14> revert to MPTCP later in the connection.</t> <t>It should be emphasized that MPTCP is not attempting to prevent the use of middleboxes that want to adjust the payload. An MPTCP-aware middlebox could provide such functionality by also rewriting checksums.</t> </section> <sectiontitle="Error Handling" anchor="sec_errors">anchor="sec_errors" numbered="true" toc="default"> <name>Error Handling</name> <t>In addition to the fallback mechanismasdescribed above, the standard classes of TCP errors may need to be handled in anMPTCP-specificMPTCP‑specific way. Note that changing semantics -- such as the relevance of a RST -- are covered in <xreftarget="sec_semantics"/>.target="sec_semantics" format="default"/>. Where possible, we do not want to deviate from regular TCP behavior.</t> <t>The following list covers possible errors and the appropriate MPTCP behavior:<list style="symbols"> <t>Unknown</t> <ul spacing="normal"> <li>Unknown token in MP_JOIN (or HMAC failure in MP_JOIN ACK, or missing MP_JOIN in SYN/ACK response): send RST (analogous to TCP's behavior on an unknownport)</t> <t>DSNport)</li> <li>DSN out of window (during normal operation): drop thedata,data; do not send DataACKs</t> <t>RemoveACKs</li> <li>Remove request for unknownaddressAddress ID: silentlyignore</t> </list> </t>ignore</li> </ul> </section> <sectiontitle="Heuristics" anchor="heuristics">anchor="heuristics" numbered="true" toc="default"> <name>Heuristics</name> <t>There are a number of heuristics that are needed for performance or deployment but that are not required for protocol correctness. In this section, we detail such heuristics. Note thatdiscussiondiscussions of buffering and certain sender and receiver window behaviors are presented in Sections <xref target="sec_rwin" format="counter"/> and <xref target="sec_sender" format="counter"/>,as well asand retransmission is discussed in <xreftarget="sec_retransmit"/>.</t>target="sec_retransmit" format="default"/>.</t> <sectiontitle="Port Usage">numbered="true" toc="default"> <name>Port Usage</name> <t>Under typical operation, an MPTCP implementationSHOULD<bcp14>SHOULD</bcp14> use the same ports as the ports that are already in use. In other words, the destination port of a SYN containing an MP_JOIN optionSHOULD<bcp14>SHOULD</bcp14> be the same as the remote port of the first subflow in the connection. The local port for such SYNsSHOULD<bcp14>SHOULD</bcp14> also be the same as the port for the first subflow (and as such, an implementationSHOULD<bcp14>SHOULD</bcp14> reserve ephemeral ports across all local IP addresses), although there may be cases where this is infeasible. This strategy is intended to maximize the probability of the SYN being permitted by a firewall or NAT at the recipient and to avoid confusing anynetwork monitoringnetwork-monitoring software.</t> <t>There may also be cases, however, where a host wishes to signal that a specific port should beused, andused; this facility is provided in the ADD_ADDR option as documented in <xreftarget="sec_add_address"/>.target="sec_add_address" format="default"/>. It is therefore feasible to allow multiple subflows between the same two addresses but using different port pairs, and such a facility could be used to allow load balancing within the network based on 5-tuples (e.g., some ECMP implementations <xreftarget="RFC2992"/>).</t>target="RFC2992" format="default"/>).</t> </section> <sectiontitle="Delayednumbered="true" toc="default"> <name>Delayed Subflow Start and SubflowSymmetry">Symmetry</name> <t>Many TCP connections are short-lived and consist only of a few segments, and so theoverheadsoverhead of using MPTCPoutweighoutweighs any benefits. A heuristic is required, therefore, to decide when to start using additional subflows in an MPTCP connection. Experimental deployments have shown that MPTCP can be applied in a range ofscenariosscenarios, so an implementationiswill likelytoneed to take into account such factorsincludingas the type of traffic being sent and the duration ofsession, andthe session; this informationMAY<bcp14>MAY</bcp14> besignalledsignaled by the application layer.</t> <t>However, for standard TCP traffic, a suggested general-purpose heuristic that an implementationMAY<bcp14>MAY</bcp14> choose to employ is as follows.</t> <t>If a host has data buffered for its peer (which implies that the application has received a request for data), the host opens one subflow for each initial window's worth of data that is buffered.</t> <t>Consideration should also be given to limiting the rate of adding new subflows, as well as limiting the total number of subflows open for a particular connection. A host may choose to vary these values based on its load or knowledge of traffic and path characteristics.</t> <t>Note that this heuristic alone is probably insufficient. Traffic for many common applications, such as downloads, is highlyasymmetricasymmetric, and the host that is multihomed may well be the client that will never fill itsbuffers,buffers and thus never use MPTCP according to this heuristic. Advanced APIs that allow an application to signal its traffic requirements would aid in these decisions.</t> <t>An additional time-based heuristic could be applied, opening additional subflows after a given period of time has passed. This would alleviate the aboveissue,issue and also provide resilience forlow-bandwidthlow‑bandwidth but long-lived applications.</t> <t>Another issue is that both communicating hosts may simultaneously try to set up a subflow between the same pair of addresses. This leads to an inefficient use of resources.</t> <t>If the same ports are used on all subflows, as recommended above, then standard TCPsimultaneous opensimultaneous-open logic should take care of this situation and only one subflow will be established between the address pairs. However, this relies on the same ports being used at both end hosts. If a host does not support TCP simultaneous open, it isRECOMMENDED<bcp14>RECOMMENDED</bcp14> that some element of randomizationisbe applied to the time to wait before opening new subflows, so that only one subflow is created between a given address pair. If, however, hosts signal additional ports to use (for example, for leveraging ECMP on-path), this heuristic is not appropriate.</t> <t>This section has shown some of theconsiderationsfactors that an implementer shouldgiveconsider when developing MPTCP heuristics, but it is not intended to be prescriptive.</t> </section> <sectiontitle="Failure Handling">numbered="true" toc="default"> <name>Failure Handling</name> <t>Requirements for MPTCP's handling of unexpected signalshave beenare given in <xreftarget="sec_errors"/>.target="sec_errors" format="default"/>. There are other failure cases, however, whereahosts can choose appropriate behavior.</t> <t>For example, <xreftarget="sec_init"/>target="sec_init" format="default"/> suggests that a hostSHOULD<bcp14>SHOULD</bcp14> fall back to trying regular TCP SYNs after one or more failures of MPTCP SYNs for a connection. A host may keep a system-wide cache of such information, so that it can back off from using MPTCP, firstly for that particular destinationhost, and eventuallyhost and, eventually, on a whole interface, if MPTCP connections continuefailing.to fail. The duration of such a cache would beimplementation-specific.</t>implementation specific.</t> <t>Another failure could occur when the MP_JOIN handshake fails. <xreftarget="sec_errors"/>target="sec_errors" format="default"/> specifies that an incorrect handshakeMUST<bcp14>MUST</bcp14> lead to the subflow being closed with a RST. A host operating an activeintrusion detectionintrusion-detection system may choose to start blocking MP_JOIN packets from the source host if multiple failed MP_JOIN attempts are seen. From the connection initiator's point of view, if an MP_JOIN fails, itSHOULD NOT<bcp14>SHOULD NOT</bcp14> attempt to connect to the same IP address and port during the lifetime of the connection, unless the other host refreshes the information with another ADD_ADDR option. Note that the ADD_ADDR option is informationalonly,only and does not guarantee that the other host will attempt a connection.</t> <t>In addition, an implementation may learn, over a number of connections, that certain interfaces or destination addresses consistently fail and may default to not trying to use MPTCP forthese. Behavior could also be learned forsuch interfaces or addresses. The behavior of subflows that perform particularly badlyperforming subflowsor subflows that regularly fail duringuse, in order touse could also be learned, so that an implementation can temporarily choose not to use these paths.</t> </section> </section> </section> <sectiontitle="Semantic Issues" anchor="sec_semantics">anchor="sec_semantics" numbered="true" toc="default"> <name>Semantic Issues</name> <t>In order to support multipath operation, the semantics of some TCP components have changed. Toaid clarity,help clarify, this sectioncollectslists these semantic changes as a point of reference.<list style="hanging"> <t hangText="Sequence number:"></t> <dl newline="false" spacing="normal" indent="3"> <dt>Sequence number:</dt> <dd> The (in-header) TCP sequence number is specific to the subflow. To allow the receiver to reorder application data, an additional data-level sequence space is used. In thisdata-leveldata‑level sequence space, the initial SYN and the final DATA_FIN occupy 1 octet of sequence space. This is done to ensure that these signals are acknowledged at the connection level. There is an explicit mapping of data sequence space to subflow sequence space, which is signaled through TCP options in datapackets.</t> <t hangText="ACK:">packets.</dd> <dt>ACK:</dt> <dd> The ACK field in the TCP header acknowledges only the subflow sequencenumber,number -- not the data-level sequence space. ImplementationsSHOULD NOT<bcp14>SHOULD NOT</bcp14> attempt to infer a data-level acknowledgment from the subflow ACKs. This separatessubflow-subflow-level and connection-level processing at an endhost.</t> <t hangText="Duplicate ACK:">host.</dd> <dt>Duplicate ACK:</dt> <dd> A duplicate ACK that includes any MPTCP signaling (with the exception of the DSS option)MUST NOT<bcp14>MUST NOT</bcp14> be treated as a signal of congestion. To limit the chances of non-MPTCP-aware entities mistakenly interpreting duplicate ACKs as a signal of congestion, MPTCPSHOULD NOT<bcp14>SHOULD NOT</bcp14> send more than two duplicate ACKs containing (non-DSS) MPTCP signals in arow.</t> <t hangText="Receive Window:">Therow.</dd> <dt>Receive Window:</dt> <dd>The receive window in the TCP header indicates the amount of free buffer space for the whole data-level connection (as opposed to the amount of space for this subflow) that is available at the receiver.This isThe semantics are the samesemanticsas for regular TCP, but to maintain these semantics the receive window must be interpreted at the sender as relative to the sequence number given in the DATA_ACK rather than the subflow ACK in the TCP header. In this way, the original role of flow controlroleis preserved. Note that some middleboxes may change the receive window, and so a hostSHOULD<bcp14>SHOULD</bcp14> use the maximum value of those recently seen on the constituent subflows for the connection-level receivewindow,window and also needs to maintain a subflow-level window for subflow-levelprocessing.</t> <t hangText="FIN:">processing.</dd> <dt>FIN:</dt> <dd> The FIN flag in the TCP header applies only to the subflow it is senton,on -- not to the whole connection. For connection-level FIN semantics, the DATA_FIN option isused.</t> <t hangText="RST:">used.</dd> <dt>RST:</dt> <dd> The RST flag in the TCP header applies only to the subflow it is senton,on -- not to the whole connection. The MP_FASTCLOSE option provides thefast closeFast Close functionality of a RST at the MPTCP connectionlevel.</t> <t hangText="Address List:">level.</dd> <dt>Address List:</dt> <dd> Address list management (i.e., knowledge of the local and remote hosts' lists of available IP addresses) is handled on a per-connection basis (as opposed to per subflow, per host, or per pair of communicating hosts). This permits the application of per-connection local policy. Adding an address to one connection (either explicitly through anAdd Address message,ADD_ADDR message or implicitly througha Join)an MP_JOIN) has noimplicationimplications for other connections between the same pair ofhosts.</t> <t hangText="5-tuple:">hosts.</dd> <dt>5-tuple:</dt> <dd> The 5-tuple (protocol, local address, local port, remote address, remote port) presented by kernel APIs to the application layer in a non-multipath-aware application is that of the first subflow, even if the subflow has since been closed and removed from the connection. This decision, and other related API issues, are discussed in more detail in <xreftarget="RFC6897"/>.</t> </list> </t>target="RFC6897" format="default"/>.</dd> </dl> </section> <sectiontitle="Security Considerations" anchor="sec_security">anchor="sec_security" numbered="true" toc="default"> <name>Security Considerations</name> <t>As identified in <xreftarget="RFC6181"/>,target="RFC6181" format="default"/>, the addition of multipath capability to TCP will bring with it a number of new classes ofthreat.threats. In order to preventthese,these threats, <xreftarget="RFC6182"/>target="RFC6182" format="default"/> presents a set of requirements for a security solution for MPTCP. The fundamental goal is for the security of MPTCP to be "no worse" than regular TCPtoday, and thetoday. The key security requirementsare: <list style="symbols"> <t>Provideare as follows: </t> <ul spacing="normal"> <li>Provide a mechanism to confirm that the parties in a subflow handshake are the same as the parties in the original connectionsetup.</t> <t>Providesetup.</li> <li>Provide verification that the peer can receive traffic at a new address before using it as part of aconnection.</t> <t>Provideconnection.</li> <li>Provide replay protection, i.e., ensure that a request toadd/removeadd&wj;/remove a subflow is'fresh'.</t> </list>"fresh".</li> </ul> <t> In order to achieve these goals, MPTCP includes a hash-based handshakealgorithmalgorithm, as documented in Sections <xref target="sec_init" format="counter"/> and <xref target="sec_join" format="counter"/>.</t> <t>The security of the MPTCP connection hangs on the use of keys that are shared once at the start of the firstsubflow,subflow and are never sent again over the network (unless used in thefast close mechanism, <xref target="sec_fastclose"/>).Fast Close mechanism (<xref target="sec_fastclose" format="default"/>)). To ease demultiplexing while not giving away any cryptographic material, future subflows use a truncated cryptographic hash of this key as the connection identification "token". The keys are concatenated and used as keys for creating Hash-based Message Authentication Codes (HMACs) used on subflow setup, in order to verify that the parties in the handshake are the same as the parties in the original connection setup. It also provides verification that the peer can receive traffic at this new address. Replay attacks would still be possible when only keys are used; therefore, the handshakes use single-use random numbers (nonces) at both ends -- this ensures that the HMAC will never be the same on two handshakes. Guidance on generating random numbers suitable for use as keys is given in <xreftarget="RFC4086"/>target="RFC4086" format="default"/> and discussed in <xreftarget="sec_init"/>.target="sec_init" format="default"/>. The nonces are valid for the lifetime of the TCP connection attempt. HMAC is also used to secure the ADD_ADDR option, due to the threats identified in <xreftarget="RFC7430"/>.</t>target="RFC7430" format="default"/>.</t> <t>The use of crypto capability bits in the initial connection handshake to negotiate the use of a particular algorithm allows the deployment of additional crypto mechanisms in the future. This negotiation would nevertheless be susceptible to a bid-down attack by an on-path active attacker who could modify the crypto capability bits in the response from the receiver to use a less secure crypto mechanism. The security mechanism presented in this document should therefore protect against all forms of flooding and hijacking attacks discussed in <xreftarget="RFC6181"/>.</t>target="RFC6181" format="default"/>.</t> <t>The version negotiation specified in <xreftarget="sec_init"/>,target="sec_init" format="default"/>, if differing MPTCP versions shared a common negotiation format, would allow an on-path attacker to apply a theoretical bid-down attack. Since the v1 and v0 protocols have a different handshake, such an attack would require that the clienttore-establish the connection usingv0,v0 andthis being supported bythat theserver.server support v0. Note that an on-path attacker would have access to the raw data, negating any other TCP-level security mechanisms.Also a change from RFC6824 has removedAs also noted in <xref target="app_changelog"/>, this document specifies thesubflow identifier fromremoval of the AddrID field <xref target="RFC6824"/> in the MP_PRIO option (<xreftarget="sec_policy"/>), to removetarget="sec_policy" format="default"/>). This change eliminates the possibility of a theoretical attack where a subflow could be placed in "backup" mode by an attacker.</t> <t>During normal operation, regular TCP protection mechanisms (such as ensuring that sequence numbers are in-window) will provide the same level of protection against attacks on individual TCP subflows as the level of protection that exists for regular TCP today. Implementations will introduce additional buffers compared to regular TCP, to reassemble data at the connection level. The application of window sizing will minimize the risk of denial-of-service attacks consuming resources.</t> <t>As discussed in <xreftarget="sec_add_address"/>,target="sec_add_address" format="default"/>, a host may advertise its private addresses, but these might point to different hosts in the receiver's network. The MP_JOIN handshake (<xreftarget="sec_join"/>)target="sec_join" format="default"/>) will ensure that this does not succeed in setting up a subflow to the incorrect host. However, it could still create unwanted TCP handshake traffic. This feature of MPTCP could be a target for denial-of-service exploits, with malicious participants in MPTCP connections encouraging the recipient to target other hosts in the network. Therefore, implementations should consider heuristics (<xreftarget="heuristics"/>)target="heuristics" format="default"/>) at both the sender and receiver to reduce the impact of this.</t> <t>To further protect against malicious ADD_ADDR messages sent by an off-path attacker, the ADD_ADDR includes an HMAC using the keys negotiated during the handshake. This effectively prevents an attacker from diverting an MPTCP connection through an off-path ADD_ADDR injection into the stream.</t> <t>A small security risk could theoretically exist with key reuse, but in order to accomplish a replay attack, both the sender and receiver keys, and the sender and receiver random numbers, in the MP_JOIN handshake (<xreftarget="sec_join"/>)target="sec_join" format="default"/>) would have to match.</t><t>Whilst<t>While this specification defines a "medium" security solution, meeting the criteria specified at the start of this section and in the threat analysis(<xref target="RFC6181"/>),document <xref target="RFC6181" format="default"/>, since attacks only ever get worse, it is likely that a future version of MPTCP would need to be able to support stronger security. There are several ways the security of MPTCP could potentially be improved; some of these would be compatible with MPTCP as defined in this document,whilstwhile others may not be. For now, the best approach is togetgain experience with the current approach, establish what might work, and check that the threat analysis is still accurate.</t> <t>Possible ways of improving MPTCP security couldinclude:<list style="symbols"> <t>defininginclude:</t> <ul spacing="normal"> <li>defining a newMPCTPMPTCP cryptographic algorithm, as negotiated in MP_CAPABLE.A sub-case could be to includeIf an implementation was being deployed in a controlled environment where additionaldeployment assumption,assumptions could be made, such asstateful servers, in orderthe ability for the servers toallowstore state during the TCP handshake, then it may be possible to use amore powerfulstronger cryptographic algorithmtothan would otherwise beused.</t> <t>definingpossible.</li> <li>defining how to secure data transfer with MPTCP,whilstwhile not changing the signaling part of theprotocol.</t> <t>definingprotocol.</li> <li>defining security that requires more option space, perhaps in conjunction with a "long options" proposal for extending the TCPoptionsoption space (such as those surveyed in <xreftarget="TCPLO"/>),target="I-D.ananth-tcpm-tcpoptext" format="default"/>), or perhaps building on the current approach with a second stage ofMPTCP-option-based security.</t> <t>revisitingsecurity based on MPTCP options.</li> <li>revisiting the working group's decision to exclusively use TCP options for MPTCPsignaling,signaling and insteadlooklooking atalso making use ofthe possibility of using TCPpayloads.</t> </list></t>payloads as well.</li> </ul> <t>MPTCP has been designed with several methods available to indicate a new security mechanism, including:<list style="symbols"> <t>available</t> <ul spacing="normal"> <li>available flags in MP_CAPABLE (<xreftarget="tcpm_capable"/>);</t> <t>availabletarget="tcpm_capable" format="default"/>).</li> <li>available subtypes in the MPTCP option (<xreftarget="fig_option"/>);</t> <t>the versiontarget="fig_option" format="default"/>).</li> <li>the Version field in MP_CAPABLE (<xreftarget="tcpm_capable"/>);</t> </list></t>target="tcpm_capable" format="default"/>).</li> </ul> </section> <sectiontitle="Interactionsanchor="sec_middleboxes" numbered="true" toc="default"> <name>Interactions withMiddleboxes" anchor="sec_middleboxes">Middleboxes</name> <t>Multipath TCP was designed to be deployable in the present world. Its design takes into account "reasonable" existing middlebox behavior. In this section, we outline a few representative middlebox-related failure scenarios and show how Multipath TCP handles them. Next, we list the design decisionsmultipathMultipath TCP has made to accommodate the different middleboxes.</t> <t>A primary concern is our use of a new TCP option. Middleboxes should forward packets with unknown options unchanged, yet there are some that don't.These weWe expectwill eitherthese middleboxes to strip options and pass the data, drop packets with new options, copy the same option into multiple segments (e.g., when doing segmentation), or drop options during segment coalescing.</t> <t>MPTCP uses a single new TCP option called "Kind", and all message types are defined by "subtype" values (see <xreftarget="IANA"/>).target="IANA" format="default"/>). This should reduce the chances of only some types of MPTCP options beingpassed, and insteadpassed; instead, the key differing characteristics are differentpaths,paths and the presence of the SYN flag.</t> <t>MPTCP SYN packets on the first subflow of a connection contain the MP_CAPABLE option (<xreftarget="sec_init"/>).target="sec_init" format="default"/>). If this is dropped, MPTCPSHOULD<bcp14>SHOULD</bcp14> fall back to regular TCP. If packets with the MP_JOIN option (<xreftarget="sec_join"/>)target="sec_join" format="default"/>) are dropped, the paths will simply not be used.</t> <t>If a middlebox strips options but otherwise passes the packets unchanged, MPTCP will behave safely. If an MP_CAPABLE option is dropped on either the outgoing path or the return path, the initiating host can fall back to regular TCP, as illustrated in <xreftarget="fig_syn"/>target="fig_syn" format="default"/> and discussed in <xreftarget="sec_init"/>.</t>target="sec_init" format="default"/>.</t> <figure anchor="fig_syn"> <name>Connection Setup with Middleboxes That Strip Options from Packets</name> <artwork align="left" name="" type="" alt=""><![CDATA[ Host A Host B | Middlebox M | | | | | SYN (MP_CAPABLE) | SYN | |-------------------|---------------->| | SYN/ACK | |<------------------------------------| a) MP_CAPABLE option stripped on outgoing path Host A Host B | SYN (MP_CAPABLE) | |-------------------------------------->| | Middlebox M | | | | | SYN/ACK |SYN/ACK (MP_CAPABLE)| |<-----------------|--------------------| b) MP_CAPABLE option stripped on return path ]]></artwork> </figure> <t>Subflow SYNs contain the MP_JOIN option. If this option is stripped on the outgoing path, the SYN will appear to be a regular SYN toHost B. Host B. Depending on whether there is a listening socket on the target port, Host B will replyeitherwith either a SYN/ACK or a RST (subflow connection fails). When Host A receives theSYN/ACKSYN/ACK, it sends a RST because the SYN/ACK does not contain the MP_JOIN option and its token. Either way, the subflow setupfails,fails but otherwise does not affect the MPTCP connection as a whole.</t><figure align="center" anchor="fig_syn" title="Connection Setup with Middleboxes that Strip Options from Packets"> <artwork align="left"><![CDATA[ Host A Host B | Middlebox M | | | | | SYN(MP_CAPABLE) | SYN | |-------------------|---------------->| | SYN/ACK | |<------------------------------------| a) MP_CAPABLE option stripped on outgoing path Host A Host B | SYN(MP_CAPABLE) | |------------------------------------>| | Middlebox M | | | | | SYN/ACK |SYN/ACK(MP_CAPABLE)| |<----------------|-------------------| b) MP_CAPABLE option stripped on return path ]]></artwork> </figure><t>We now examine data flow with MPTCP, assuming that the flow is correctly set up, which implies that the options in the SYN packets were allowed through by the relevant middleboxes. If options are allowed through and there is no resegmentation or coalescing to TCP segments, Multipath TCP flows can proceed without problems.</t> <t>The case when options get stripped on data packetshas beenis discussed inthe Fallback section.<xref target="sec_fallback" format="default"/>. If only some MPTCP options are stripped, behavior is not deterministic. If somedata sequence mappingsData Sequence Mappings are lost, the connection can continue so long as mappings exist for the subflow-level data (e.g., if multiple maps have been sent that reinforce each other). If some subflow-level space is left unmapped, however, the subflow is treated as broken and is closed,throughusing the process described in <xreftarget="sec_fallback"/>.target="sec_fallback" format="default"/>. MPTCP should survive with a loss of some Data ACKs, but performance will degrade as the fraction of stripped options increases. We do not expect such cases to appear in practice, though: most middleboxes will either strip all options or let them all through.</t> <t>We end this section with a list of middlebox classes, their behavior, and the elements in the MPTCP design that allow operation through such middleboxes. Issues surrounding dropping packets with options or stripping options were discussedabove,above and are not included here:<list style="symbols"> <t>NATs <xref target="RFC3022"/></t> <ul spacing="normal"> <li>NATs (Network Address (andPort)port) Translators) <xref target="RFC3022" format="default"/> change the source address (and often the source port) of packets. This means that a host will not know its public-facing address for signaling in MPTCP. Therefore, MPTCP permits implicit address addition via the MP_JOIN option, and the handshake mechanism ensures that connection attempts to private addresses <xreftarget="RFC1918"/>,target="RFC1918" format="default"/>, since they are authenticated, will only set up subflows to the correct hosts. Explicit address removal is undertaken by an Address ID to allow no knowledge of the sourceaddress.</t> <t>Performanceaddress.</li> <li>Performance Enhancing Proxies (PEPs) <xreftarget="RFC3135"/>target="RFC3135" format="default"/> might proactively ACK data to increase performance. MPTCP, however, relies on accurate congestion control signals from the end host, andnon-MPTCP-awarenon‑MPTCP-aware PEPs will not be able to provide such signals. MPTCP will, therefore, fall back to single-pathTCP,TCP or close the problematic subflow (see <xreftarget="sec_fallback"/>).</t> <t>Traffic Normalizerstarget="sec_fallback" format="default"/>).</li> <li>Traffic normalizers <xreftarget="norm"/>target="norm" format="default"/> may not allow holes in sequence numbers, and they may cache packets and retransmit the same data. MPTCP looks like standard TCP on thewire,wire and will not retransmit different data on the same subflow sequence number. In the event of a retransmission, the same data will be retransmitted on the original TCP subflow even if it is additionally retransmitted at the connection level on a differentsubflow.</t> <t>Firewallssubflow.</li> <li>Firewalls <xreftarget="RFC2979"/>target="RFC2979" format="default"/> might performinitial sequence numberInitial Sequence Number (ISN) randomization on TCP connections. MPTCP uses relative sequence numbers indata sequence mappingData Sequence Mappings to cope with this. Like NATs, firewalls will not permit many incoming connections, so MPTCP supports address signaling (ADD_ADDR) so that a multiaddressed host can invite its peer behind the firewall/NAT to connect out to its additionalinterface.</t> <t>Intrusion Detection/Preventioninterface.</li> <li>Intrusion Detection Systems / Intrusion Prevention Systems(IDS/IPS)(IDSs&wj;/IPSs) observe packet streams for patterns and content that could threaten a network. MPTCP may require the instrumentation of additional paths, and an MPTCP-awareIDS/IPSIDS or IPS would need to read MPTCP tokens to correlate data frommutliplemultiple subflows to maintain comparable visibility into all of the traffic between devices. Without such changes, an IDS would get an incomplete view of the traffic, increasing the risk of missing traffic of interest (falsenegatives),negatives) and increasing the chances of erroneously identifying a subflow as a risk due to only seeing partial data (falsepositives).</t> <t>Application-levelpositives).</li> <li>Application-level middleboxes such as content-aware firewalls may alter the payload within asubflow, such assubflow -- for example, rewriting URIs in HTTP traffic. MPTCP will detectthesesuch changes using the checksum and close the affected subflow(s), if there are other subflows that can be used. If all subflows are affected,multipathMPTCP will fall back to TCP, allowing such middleboxes to change the payload. MPTCP-aware middleboxes should be able to adjust the payload and MPTCP metadata in order not to break theconnection.</t> </list>connection.</li> </ul> <t> In addition, all classes of middleboxes may affect TCP traffic in the following ways:<list style="symbols"> <t>TCP</t> <ul spacing="normal"> <li>TCP options may be removed, or packets with unknown options dropped, by many classes of middleboxes. It is intended that the initial SYN exchange, with a TCP option, will be sufficient to identify thepathpath's capabilities. If such a packet does not get through, MPTCP will end up falling back to regularTCP.</t> <t>Segmentation/CoalescingTCP.</li> <li>Segmentation/coalescing (e.g., TCP segmentation offloading) might copy options between packets and might strip some options. MPTCP'sdata sequence mappingData Sequence Mapping includes the relative subflow sequence number instead of using the sequence number in the segment. In this way, the mapping is independent of the packets that carryit.</t> <t>Theit.</li> <li>The receive window may be shrunk by some middleboxes at the subflow level. MPTCP will use the maximum window at the datalevel,level but will also obey subflow-specificwindows.</t> </list> </t> </section> <section anchor="Acknowledgments" title="Acknowledgments"> <!-- <t>The authors were originally supported by Trilogy (http://www.trilogy-project.org), a research project (ICT-216372) partially funded by the European Community under its Seventh Framework Program.</t> <t>Alan Ford was originally supported by Roke Manor Research and later Cisco Systems.</t> --> <t>The authors gratefully acknowledge significant input into this document from Sébastien Barré and Andrew McDonald.</t> <t>The authors also wish to acknowledge reviews and contributions from Iljitsch van Beijnum, Lars Eggert, Marcelo Bagnulo, Robert Hancock, Pasi Sarolahti, Toby Moncaster, Philip Eardley, Sergio Lembo, Lawrence Conroy, Yoshifumi Nishida, Bob Briscoe, Stein Gjessing, Andrew McGregor, Georg Hampel, Anumita Biswas, Wes Eddy, Alexey Melnikov, Francis Dupont, Adrian Farrel, Barry Leiba, Robert Sparks, Sean Turner, Stephen Farrell, Martin Stiemerling, Gregory Detal, Fabien Duchene, Xavier de Foy, Rahul Jadhav, Klemens Schragel, Mirja Kuehlewind, Sheng Jiang, Alissa Cooper, Ines Robles, Roman Danyliw, Adam Roach, Barry Leiba, Alexey Melnikov, Eric Vyncke, and Ben Kaduk.</t>windows.</li> </ul> </section> <section anchor="IANA"title="IANA Considerations">numbered="true" toc="default"> <name>IANA Considerations</name> <t>This document obsoletesRFC6824 and as such<xref target="RFC6824"/>. As such, IANAis requestedhas updated several registries toupdatepoint to this document. In addition, this document creates one new registry. These topics are described in theTCP option spacefollowing subsections.</t> <section anchor="IANA-TCP-Option-Kind" numbered="true" toc="default"> <name>TCP Option Kind Numbers</name> <t>IANA has updated the "TCP Option Kind Numbers" registry to point to this document for Multipath TCP, asfollows:</t> <texttableshown in <xref target="table_tcpo"/>:</t> <table anchor="table_tcpo"title="TCPalign="center"> <name>TCP Option KindNumbers"> <ttcol align="center">Kind</ttcol> <ttcol align="center">Length</ttcol> <ttcol align="center">Meaning</ttcol> <ttcol align="center">Reference</ttcol> <c>30</c> <c>N</c> <c>Multipath TCP (MPTCP)</c> <c>This document</c> </texttable>Numbers</name> <thead> <tr> <th align="center">Kind</th> <th align="center">Length</th> <th align="center">Meaning</th> <th align="center">Reference</th> </tr> </thead> <tbody> <tr> <td align="center">30</td> <td align="center">N</td> <td align="center">Multipath TCP (MPTCP)</td> <td align="center">RFC 8684</td> </tr> </tbody> </table> </section> <section anchor="IANA_subtypes"title="MPTCPnumbered="true" toc="default"> <name>MPTCP OptionSubtypes">Subtypes</name> <t>The 4-bit MPTCP subtypesub-registry ("MPTCPin the "MPTCP Option Subtypes" subregistry under the "Transmission Control Protocol (TCP) Parameters"registry)registry was defined inRFC6824.<xref target="RFC6824"/>. SinceRFC6824 was<xref target="RFC6824"/> is an Experimental RFC and not a Standards Track RFC, and since no further entries have occurred beyond those pointing toRFC6824,<xref target="RFC6824"/>, IANAis requested to replacehas replaced the existing registry with the contents of <xreftarget="table_iana"/>target="table_iana" format="default"/> and with the following explanatory note.</t> <t>Note: This registry specifies the MPTCP Option Subtypes for MPTCP v1, which obsoletes the Experimental MPTCP v0. For the MPTCP v0 subtypes, please refer toRFC6824.</t> <texttable<xref target="RFC6824"/>.</t> <table anchor="table_iana"title="MPTCPalign="center"> <name>MPTCP OptionSubtypes"> <ttcol align="center">Value</ttcol> <ttcol align="center">Symbol</ttcol> <ttcol align="center">Name</ttcol> <ttcol align="center">Reference</ttcol> <c>0x0</c> <c>MP_CAPABLE</c> <c>Multipath Capable</c> <c>This document,Subtypes</name> <thead> <tr> <th align="center">Value</th> <th align="center">Symbol</th> <th align="center">Name</th> <th align="center">Reference</th> </tr> </thead> <tbody> <tr> <td align="center">0x0</td> <td align="center">MP_CAPABLE</td> <td align="center">Multipath Capable</td> <td align="center">RFC 8684, <xreftarget="sec_init"/></c> <c>0x1</c> <c>MP_JOIN</c> <c>Join Connection</c> <c>This document,target="sec_init" format="default"/></td> </tr> <tr> <td align="center">0x1</td> <td align="center">MP_JOIN</td> <td align="center">Join Connection</td> <td align="center">RFC 8684, <xreftarget="sec_join"/></c> <c>0x2</c> <c>DSS</c> <c>Datatarget="sec_join" format="default"/></td> </tr> <tr> <td align="center">0x2</td> <td align="center">DSS</td> <td align="center">Data Sequence Signal (Data ACK anddata sequence mapping)</c> <c>This document, <xref target="sec_generalop"/></c> <c>0x3</c> <c>ADD_ADDR</c> <c>Add Address</c> <c>This document, <xref target="sec_add_address"/></c> <c>0x4</c> <c>REMOVE_ADDR</c> <c>Remove Address</c> <c>This document, <xref target="sec_remove_addr"/></c> <c>0x5</c> <c>MP_PRIO</c> <c>ChangeData Sequence Mapping)</td> <td align="center">RFC 8684, <xref target="sec_generalop" format="default"/></td> </tr> <tr> <td align="center">0x3</td> <td align="center">ADD_ADDR</td> <td align="center">Add Address</td> <td align="center">RFC 8684, <xref target="sec_add_address" format="default"/></td> </tr> <tr> <td align="center">0x4</td> <td align="center">REMOVE_ADDR</td> <td align="center">Remove Address</td> <td align="center">RFC 8684, <xref target="sec_remove_addr" format="default"/></td> </tr> <tr> <td align="center">0x5</td> <td align="center">MP_PRIO</td> <td align="center">Change SubflowPriority</c> <c>This document, <xref target="sec_policy"/></c> <c>0x6</c> <c>MP_FAIL</c> <c>Fallback</c> <c>This document, <xref target="sec_fallback"/></c> <c>0x7</c> <c>MP_FASTCLOSE</c> <c>Fast Close</c> <c>This document, <xref target="sec_fastclose"/></c> <c>0x8</c> <c>MP_TCPRST</c> <c>Subflow Reset</c> <c>This document, <xref target="sec_reset"/></c> <c>0xf</c> <c>MP_EXPERIMENTAL</c> <c>Reserved for private experiments</c> <c></c> </texttable>Priority</td> <td align="center">RFC 8684, <xref target="sec_policy" format="default"/></td> </tr> <tr> <td align="center">0x6</td> <td align="center">MP_FAIL</td> <td align="center">Fallback</td> <td align="center">RFC 8684, <xref target="sec_fallback" format="default"/></td> </tr> <tr> <td align="center">0x7</td> <td align="center">MP_FASTCLOSE</td> <td align="center">Fast Close</td> <td align="center">RFC 8684, <xref target="sec_fastclose" format="default"/></td> </tr> <tr> <td align="center">0x8</td> <td align="center">MP_TCPRST</td> <td align="center">Subflow Reset</td> <td align="center">RFC 8684, <xref target="sec_reset" format="default"/></td> </tr> <tr> <td align="center">0xf</td> <td align="center">MP_EXPERIMENTAL</td> <td align="center">Reserved for Private Use</td> <td align="center"/> </tr> </tbody> </table> <t>Values 0x9 through 0xe are currently unassigned. Option 0xf is reserved for use by private experiments. Its use may be formalized in a future specification. Future assignments in this registry are to be defined by Standards Action as defined by <xreftarget="RFC8126"/>.target="RFC8126" format="default"/>. Assignments consist of the MPTCP subtype's symbolicname andname, its associated value, and a reference to its specification.</t> </section> <section anchor="IANA_handshake"title="MPTCPnumbered="true" toc="default"> <name>MPTCP HandshakeAlgorithms">Algorithms</name> <t>The "MPTCP Handshake Algorithms"sub-registrysubregistry under the "Transmission Control Protocol (TCP) Parameters" registry was defined inRFC6824.<xref target="RFC6824"/>. SinceRFC6824 was<xref target="RFC6824"/> is an Experimental RFC and not a Standards Track RFC, and since no further entries have occurred beyond those pointing toRFC6824,<xref target="RFC6824"/>, IANAis requested to replacehas replaced the existing registry with the contents of <xreftarget="table_crypto"/>target="table_crypto" format="default"/> and with the following explanatory note.</t> <t>Note: This registry specifies the MPTCP Handshake Algorithms for MPTCP v1, which obsoletes the Experimental MPTCP v0. For the MPTCP v0 subtypes, please refer toRFC6824.</t> <texttable<xref target="RFC6824"/>.</t> <table anchor="table_crypto"title="MPTCPalign="center"> <name>MPTCP HandshakeAlgorithms"> <ttcolAlgorithms</name> <thead> <tr> <th align="center">FlagBit</ttcol> <ttcol align="center">Meaning</ttcol> <ttcol align="center">Reference</ttcol> <c>A</c> <c>Checksum required</c> <c>This document,Bit</th> <th align="center">Meaning</th> <th align="center">Reference</th> </tr> </thead> <tbody> <tr> <td align="center">A</td> <td align="center">Checksum required</td> <td align="center">RFC 8684, <xreftarget="sec_init"/></c> <c>B</c> <c>Extensibility</c> <c>This document,target="sec_init" format="default"/></td> </tr> <tr> <td align="center">B</td> <td align="center">Extensibility</td> <td align="center">RFC 8684, <xreftarget="sec_init"/></c> <c>C</c> <c>Dotarget="sec_init" format="default"/></td> </tr> <tr> <td align="center">C</td> <td align="center">Do not attempt to establish new subflows to the sourceaddress.</c> <c>This document,address.</td> <td align="center">RFC 8684, <xreftarget="sec_init"/></c> <c>D-G</c> <c>Unassigned</c> <c></c> <c>H</c> <c>HMAC-SHA256</c> <c>This document,target="sec_init" format="default"/></td> </tr> <tr> <td align="center">D-G</td> <td align="center">Unassigned</td> <td align="center"/> </tr> <tr> <td align="center">H</td> <td align="center">HMAC-SHA256</td> <td align="center">RFC 8684, <xreftarget="sec_join"/></c> </texttable>target="sec_join" format="default"/></td> </tr> </tbody> </table> <t>Note that the meanings of bitsD"D" throughH"H" can be dependent upon bitB,"B", depending on how the Extensibility parameter is defined in future specifications; see <xreftarget="sec_init"/>target="sec_init" format="default"/> for more information.</t> <t>Future assignments in this registry are also to be defined by Standards Action as defined by <xreftarget="RFC8126"/>.target="RFC8126" format="default"/>. Assignments consist of the value of the flags, a symbolic name for the algorithm, and a reference to its specification.</t> </section> <section anchor="IANA_rst"title="MP_TCPRSTnumbered="true" toc="default"> <name>MP_TCPRST ReasonCodes">Codes</name> <t>IANAis requested to createhas created a furthersub-registry,subregistry, "MPTCP MP_TCPRST Reason Codes" under the "Transmission Control Protocol (TCP) Parameters" registry, based on the reason code in the MP_TCPRST (<xreftarget="sec_reset"/>)target="sec_reset" format="default"/>) message. Initial values for this registry are given in <xreftarget="table_rstcodes"/>;target="table_rstcodes" format="default"/>; future assignments are to be defined by Specification Required as defined by <xreftarget="RFC8126"/>.target="RFC8126" format="default"/>. Assignments consist of the value of the code, a short description of its meaning, and a reference to its specification. The maximum value is 0xff.</t> <table anchor="table_rstcodes" align="center"> <name>MPTCP MP_TCPRST Reason Codes</name> <thead> <tr> <th align="center">Code</th> <th align="center">Meaning</th> <th align="center">Reference</th> </tr> </thead> <tbody> <tr> <td align="center">0x00</td> <td align="center">Unspecified error</td> <td align="center">RFC 8684, <xref target="sec_reset" format="default"/></td> </tr> <tr> <td align="center">0x01</td> <td align="center">MPTCP-specific error</td> <td align="center">RFC 8684, <xref target="sec_reset" format="default"/></td> </tr> <tr> <td align="center">0x02</td> <td align="center">Lack of resources</td> <td align="center">RFC 8684, <xref target="sec_reset" format="default"/></td> </tr> <tr> <td align="center">0x03</td> <td align="center">Administratively prohibited</td> <td align="center">RFC 8684, <xref target="sec_reset" format="default"/></td> </tr> <tr> <td align="center">0x04</td> <td align="center">Too much outstanding data</td> <td align="center">RFC 8684, <xref target="sec_reset" format="default"/></td> </tr> <tr> <td align="center">0x05</td> <td align="center">Unacceptable performance</td> <td align="center">RFC 8684, <xref target="sec_reset" format="default"/></td> </tr> <tr> <td align="center">0x06</td> <td align="center">Middlebox interference</td> <td align="center">RFC 8684, <xref target="sec_reset" format="default"/></td> </tr> </tbody> </table> <t>As guidance to theDesignated Expertdesignated expert <xreftarget="RFC8126"/>,target="RFC8126" format="default"/>, assignments should not normally be refused unless codepoint space is becoming scarce,providingprovided that there is a clear distinction from other, already-existingcodes,codes and alsoprovidingprovided that there is sufficient guidance forimplementorsimplementers both sending and receiving these codes.</t><texttable anchor="table_rstcodes" title="MPTCP MP_TCPRST Reason Codes"> <ttcol align="center">Code</ttcol> <ttcol align="center">Meaning</ttcol> <ttcol align="center">Reference</ttcol> <c>0x00</c> <c>Unspecified TCP error</c> <c>This document, <xref target="sec_reset"/></c> <c>0x01</c> <c>MPTCP specific error</c> <c>This document, <xref target="sec_reset"/></c> <c>0x02</c> <c>Lack of resources</c> <c>This document, <xref target="sec_reset"/></c> <c>0x03</c> <c>Administratively prohibited</c> <c>This document, <xref target="sec_reset"/></c> <c>0x04</c> <c>Too much outstanding data</c> <c>This document, <xref target="sec_reset"/></c> <c>0x05</c> <c>Unacceptable performance</c> <c>This document, <xref target="sec_reset"/></c> <c>0x06</c> <c>Middlebox interference</c> <c>This document, <xref target="sec_reset"/></c> </texttable></section> </section> </middle><!-- *****BACK MATTER ***** --><back><references title="Normative References"> &RFC0793; &RFC2104; &RFC2119; &RFC5961; &RFC6234; &RFC8174;<displayreference target="I-D.ananth-tcpm-tcpoptext" to="TCPLO"/> <references> <name>References</name> <references> <name>Normative References</name> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.0793.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2104.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2119.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.5961.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6234.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.8174.xml"/> </references><references title="Informative References"> &RFC1122; &RFC7323; &RFC1918; &RFC2018; &RFC5681; &RFC2979; &RFC2992; &RFC3022; &RFC3135; &RFC4086; &RFC4987; &RFC8126; &RFC6181; &RFC6356; &RFC6897; &RFC6182; &RFC6528; &RFC7413; &RFC7430; &RFC8041;<references> <name>Informative References</name> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.1122.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.7323.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.1918.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2018.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.5681.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2979.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2992.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.3022.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.3135.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.4086.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.4987.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.8126.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6181.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6356.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6897.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6182.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6528.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6824.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.7413.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.7430.xml"/> <xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.8041.xml"/> <!--&TCPLO; draft-ananth-tcpm-tcpoptext-00; Expired-->draft-ananth-tcpm-tcpoptext (Expired) --> <xi:include href="https://www.rfc-editor.org/refs/bibxml3/reference.I-D.ananth-tcpm-tcpoptext.xml"/> <referenceanchor='TCPLO'>anchor="norm" target="https://www.usenix.org/legacy/events/sec01/full_papers/handley/handley.pdf"> <front><title>TCP option space extension</title> <author initials='A' surname='Ramaiah' fullname='Anantha Ramaiah'> <organization /> </author> <date month='March' day='26' year='2012' /> <abstract><t>The document goals are as follows: Firstly, this document summarizes the motivations for extending TCP option space. Secondly, It tries to summarize the various known issues that needs to be taken into account while extending the TCP option space. Thirdly, it briefly provides a short summary of the various TCP option space proposals that has been proposed so far. Some additional proposals which includes variations to the existing proposals are also presented. The goal of this document is to rejuvenate the discussions on this topic and eventually to converge on a scheme for extending TCP option space.</t></abstract> </front> <seriesInfo name='Work in' value='Progress' /> </reference> <reference anchor='norm' target="http://www.usenix.org/events/sec01/full_papers/handley/handley.pdf"><front><title<title abbrev="Network Intrusion Detection: Evasion, Traffic Normalization, and End-to-End Protocol Semantics ">Network Intrusion Detection: Evasion, Traffic Normalization, and End-to-End ProtocolSemantics</title><author initials='M.' surname='Handley' fullname='Mark Handley'><organization>ACIRI</organization></author><author initials='V.' surname='Paxson' fullname='Vern Paxson'><organization>ACIRI</organization></author><author initials='C.' surname='Kreibich' fullname='Christian Kreibich'><organization>Technische Universitat Munchen</organization></author><date year="2001"/></front><seriesInfoSemantics</title> <seriesInfo name="UsenixSecurity" value="2001"/></reference>Security Symposium" value="2001"/> <author initials="M." surname="Handley" fullname="Mark Handley"> <organization>ACIRI</organization> </author> <author initials="V." surname="Paxson" fullname="Vern Paxson"> <organization>ACIRI</organization> </author> <author initials="C." surname="Kreibich" fullname="Christian Kreibich"> <organization>Technische Universitat Munchen</organization> </author> <date month="August" year="2001"/> </front> </reference> <referenceanchor='howhard' target="https://www.usenix.org/conference/nsdi12/how-hard-can-it-be-designing-and-implementing-deployable-multipath-tcp"> <front><titleanchor="howhard" target="https://www.usenix.org/conference/nsdi12/technical-sessions/presentation/raiciu"> <front> <title abbrev="How Hard Can It Be? Designing and Implementing a Deployable Multipath TCP">How Hard Can It Be? Designing and Implementing a Deployable Multipath TCP</title> <seriesInfo name="Usenix Symposium on Networked Systems Design and Implementation" value="2012"/> <authorinitials='C.' surname='Raiciu' fullname='Costin Raiciu'><organization>Universitateainitials="C." surname="Raiciu" fullname="Costin Raiciu"> <organization>Universitatea PolitehnicaBucuresti</organization></author>Bucuresti</organization> </author> <authorinitials='C.' surname='Paasch' fullname='Christoph Paasch'><organization>Universiteinitials="C." surname="Paasch" fullname="Christoph Paasch"> <organization>Universite Catholique deLouvain</organization></author>Louvain</organization> </author> <authorinitials='S.' surname='Barre' fullname='Sebastien Barre'><organization>Universiteinitials="S." surname="Barre" fullname="Sebastien Barre"> <organization>Universite Catholique deLouvain</organization></author>Louvain</organization> </author> <authorinitials='A.' surname='Ford' fullname='Alan Ford'><organization/></author>initials="A." surname="Ford" fullname="Alan Ford"> <organization/> </author> <authorinitials='M.' surname='Honda' fullname='Michio Honda'><organization>Keio University</organization></author>initials="M." surname="Honda" fullname="Michio Honda"> <organization>Keio University</organization> </author> <authorinitials='F.' surname='Duchene' fullname='Fabien Duchene'><organization>Universiteinitials="F." surname="Duchene" fullname="Fabien Duchene"> <organization>Universite Catholique deLouvain</organization></author>Louvain</organization> </author> <authorinitials='O.' surname='Bonaventure' fullname='Olivier Bonaventure'><organization>Universiteinitials="O." surname="Bonaventure" fullname="Olivier Bonaventure"> <organization>Universite Catholique deLouvain</organization></author>Louvain</organization> </author> <authorinitials='M.' surname='Handley' fullname='Mark Handley'><organization>Universityinitials="M." surname="Handley" fullname="Mark Handley"> <organization>University CollegeLondon</organization></author>London</organization> </author> <dateyear="2012" />month="April" year="2012"/> </front><seriesInfo name="Usenix Symposium on Networked Systems Design and Implementation" value="2012"/></reference> <referenceanchor='deployments' target="https://www.ietfjournal.org/multipath-tcp-deployments/"><front><titleanchor="deployments" target="https://www.ietfjournal.org/multipath-tcp-deployments/"> <front> <title abbrev="MPTCP Deployments">Multipath TCPDeployments</title><author initials='O.' surname='Bonaventure' fullname='Olivier Bonaventure'><organization>UniversiteDeployments</title> <seriesInfo name="IETF Journal" value="2016"/> <author initials="O." surname="Bonaventure" fullname="Olivier Bonaventure"> <organization>Universite Catholique deLouvain</organization></author><author initials='S.' surname='Seo' fullname='SungHoon Seo'></author><date day="1"Louvain</organization> </author> <author initials="S." surname="Seo" fullname="SungHoon Seo"/> <date month="November"year="2016"/></front><seriesInfo name="IETF Journal" value="2016"/></reference>year="2016"/> </front> </reference> </references> </references> <sectiontitle="Notesanchor="app_options" numbered="true" toc="default"> <name>Notes on Use of TCPOptions" anchor="app_options">Options</name> <t>The TCP option space is limited due to the length of the Data Offset field in the TCP header (4 bits), which defines the TCP header length in 32-bit words. With the standard TCP header being 20 bytes, this leaves a maximum of 40 bytes for options, and many of these may already be used by options such as timestamp and SACK.</t> <t>Wehaveperformed a brief study on the commonly used TCP options in SYN, data, and pure ACKpackets,packets and found that there is enough room to fit all the optionswe propose usingdiscussed in this document.</t> <t>SYN packets typically include the following options: Maximum Segment Size (MSS) (4 bytes), window scale (3 bytes), SACK permitted(2 bytes),(2 bytes), and timestamp(10 bytes) options. Together these(10 bytes). The sumto 19 bytes.of these options is 19 bytes. Some operating systems appear to pad each option up to a word boundary, thus using 24 bytes (a brief survey suggests that Windows XP and Mac OS X do this, whereas Linux does not). Optimistically, therefore, we have 21 bytesspare,available, or 16 ifit hasoptions have to be word-aligned. In either case, however, the SYN versions ofMultipath Capable (12 bytes)MP_CAPABLE (12 bytes) andJoinMP_JOIN (12 or16 bytes) options16 bytes) will fit in this remaining space.</t> <t>Note that due to the use of a 64-bit data-level sequence space, it is feasible that MPTCP will not require the timestamp option for protection against wrapped sequence numbers(PAWS(per the Protection Against Wrapped Sequences (PAWS) mechanism, as described in <xreftarget="RFC7323"/>),target="RFC7323" format="default"/>), since the data-level sequence space has far less chance of wrapping. Confirmation of the validity of thisoptimisationoptimization is left for further study.</t> <t>TCP data packets typically carry timestamp options in every packet, taking 10 bytes (or1212, with padding). That leaves 30 bytes (or 28, if word-aligned). TheData Sequence Signal (DSS)DSS option varies inlengthlength, depending onwhether(1) whether thedata sequence mapping and DATA_ACKData Sequence Mapping, DATA_ACK, or both are included,and whether(2) whether the sequence numbers in use are 4 or 8octets.octets, and (3) whether the checksum is present. The maximum size of the DSS option is 28 bytes, so even that will fit in the available space. But unless a connection is both bidirectional and high-bandwidth, it is unlikely that all that option space will be required on each DSS option.</t> <t>Within the DSS option, it is not necessary to include thedata sequence mappingData Sequence Mapping and DATA_ACK in each packet, and in many cases it may be possible to alternate their presence (so long as the mapping covers the data being sent in thefollowingsubsequent packet). It would also be possible to alternate between4-4-byte and 8-byte sequence numbers in each option.</t> <t>On subflow and connection setup, an MPTCP option is also set on the third packet (an ACK). These are 20 bytes (forMultipath Capable)MP_CAPABLE) and24 bytes24 bytes (forJoin),MP_JOIN), both of which will fit in the available option space.</t> <t>Pure ACKs in TCP typically contain only timestamps (10 bytes). Here, Multipath TCP typically needs to encode only the DATA_ACK (maximum of 12 bytes). Occasionally, ACKs will contain SACK information. Depending on the number of lost packets, SACK may utilize the entire option space. If a DATA_ACK had to be included, then it is probably necessary to reduce the number of SACK blocks to accommodate the DATA_ACK. However, the presence of the DATA_ACK is unlikely to be necessary in a case where SACK is in use, since until at least some of the SACK blocks have been retransmitted, the cumulative data-level ACK will not be moving forward (or if it does, due to retransmissions on another path, then that path can also be used to transmit the new DATA_ACK).</t> <t>The ADD_ADDR option can be between 16 and 30 bytes, depending onwhether(1) whether IPv4 or IPv6 isused,used andwhether(2) whether or not the port number is present. It is unlikely that such signaling would fit in a data packet (although if there is space, it is fine to include it). It is recommendedto usethat duplicate ACKs not be used withnoany other payload oroptionsoptions, in order to transmit these rare signals. Note that this is the reason for mandating that duplicate ACKs with MPTCP optionsarenot be taken as a signal of congestion.</t> </section> <sectiontitle="TCPanchor="app_tfo" numbered="true" toc="default"> <name>TCP Fast Open andMPTCP" anchor="app_tfo">MPTCP</name> <t>TCP Fast Open (TFO) is an experimental TCP extension, described in <xreftarget="RFC7413"/>,target="RFC7413" format="default"/>, which has been introduced to allow the sending of data one RTT earlier than with regular TCP. This is considered a valuablegaingain, as very short connections are very common, especially for HTTP request/response schemes. It achieves this by sending theSYN-segmentSYN segment together with the application's data and allowing the listener to reply immediately with data after the SYN/ACK. <xreftarget="RFC7413"/>target="RFC7413" format="default"/> secures thismechanism,mechanism by using a new TCP option that includes a cookiewhichthat is negotiated in a preceding connection.</t> <t>When usingTCP Fast OpenTFO in conjunction with MPTCP, there are two key points to take into account, as detailedhereafter.</t>below.</t> <sectiontitle="TFO cookie requestanchor="tfocookie" numbered="true" toc="default"> <name>TFO Cookie Request withMPTCP" anchor="tfocookie">MPTCP</name> <t>When a TFO initiator first connects to a listener, it cannot immediately include data in the SYN for security reasons <xreftarget="RFC7413"/>.target="RFC7413" format="default"/>. Instead, it requests a cookie that will be used in subsequent connections. This is done with the TCP cookie request/response options, ofrespectively2 bytes and 6-18bytesbytes, respectively (depending on the chosen cookie length).</t> <t>TFO and MPTCP can becombinedcombined, provided that the total length of all the options does not exceed the maximum 40 bytes possible in TCP:<list style="symbols"> <t>In</t> <ul spacing="normal"> <li>In the SYN: MPTCP uses a4-bytes long4-byte MP_CAPABLE option. The sum of the MPTCP and TFO optionssum up tois 6 bytes. With typicalTCP-optionsTCP options using up to 19 bytes in the SYN (24 bytes if options are padded at a word boundary), there is enough space to combine the MP_CAPABLE with the TFOCookie Request.</t> <t>Incookie request.</li> <li>In theSYN+ACK:SYN + ACK: MPTCP uses a12-bytes long12-byte MP_CAPABLE option, but now the TFO option can be as long as 18 bytes. Since the maximum option length may be exceeded, it is up to the listener tosolveavoid this problem by using a shorter cookie. As an example, if we consider that 19 bytes are used for classical TCP options, the maximum possible cookie length would beof7 bytes. Notethatthat, for the SYN packet, the same limitation applies to subsequentconnections, for the SYN packetconnections (because the initiator then echoesbackthe cookie back to the listener). Finally, if the security impact of reducing the cookie size is not deemed acceptable, the listener can reduce the amount of space used by otherTCP-optionsTCP options by omitting the TCP timestamps (as outlined in <xreftarget="app_options"/>).</t> </list></t>target="app_options" format="default"/>).</li> </ul> </section> <sectiontitle="Data sequence mappinganchor="tfodata" numbered="true" toc="default"> <name>Data Sequence Mapping underTFO" anchor="tfodata"> <t>MPTCP uses, inTFO</name> <t>In the TCP establishment phase, MPTCP uses a key exchange that is used to generate the Initial Data Sequence Numbers (IDSNs). In particular, the SYN with MP_CAPABLE occupies the first octet ofthedata sequence space. With TFO, one way to handle the data sent together with the SYN would be to consider an implicit DSS mapping that covers that SYN segment (since there is not enough space in the SYN to include a DSS option). The problem with that approach is that if a middlebox modifies the TFO data, this will not be noticed by MPTCP because of the absence of aDSS-checksum.DSS checksum. For example, aTCPTCP‑aware (but notMPTCP)-awareMPTCP-aware) middlebox could insert bytes at the beginning of the stream and adapt the TCP checksum and sequence numbers accordingly. With an implicit mapping, this information would give to the initiator and listener a different viewonof theDSS-mapping, withDSS mapping; there would be no way to detect thisinconsistency asinconsistency, because the DSS checksum is not present.</t> <t>To solvethis,this issue, the TFO data must not be considered part of theData Sequence Numberdata sequence number space: the SYN with MP_CAPABLE still occupies the first octet of data sequence space, but then the first non-TFO data byte occupies the second octet. This guarantees that, if the use ofDSS-checksumthe DSS checksum is negotiated, all data in the data sequence number space is checksummed. We also note that this does not entail a loss of functionality, becauseTFO-dataTFO data is always only sent on the initialsubflowsubflow, before any attempt to create additional subflows.</t> </section> <sectiontitle="Connection establishment examples" anchor="tfoexamples"> <t>The following shows aanchor="tfoexamples" numbered="true" toc="default"> <name>Connection Establishment Examples</name> <t>A few examples of possibleTFO+MPTCP"TFO + MPTCP" establishmentscenarios.</t>scenarios are shown below.</t> <t>Before an initiator can send data together with the SYN, it must request a cookietofrom the listener, as shown in <xreftarget="fig_tfocookie"/>. This is done by simply combining the TFO and MPTCP options.</t> <figure align="center" anchor="fig_tfocookie" title="Cookie request -target="fig_tfocookie" format="default"/>. (Note: The sequence number and length are annotated in <xref target="fig_tfocookie" format="default"/> as Seq(Length) (e.g., "S. 0(0)") and usedhereafteras such in the subsequent figures (e.g., "S 0(20)" in <xref target="fig_tfodata"/>).) This is done by simply combining thefigures.">TFO and MPTCP options.</t> <figure anchor="fig_tfocookie"> <name>Cookie Request</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ initiator listener | | | S Seq=0(Length=0) <MP_CAPABLE>, <TFO cookie request> | |----------------------------------------------------------->--------------------------------------------------------> | | | | S. 0(0) ack 1 <MP_CAPABLE>, <TFO cookie> | |<-----------------------------------------------------------<-------------------------------------------------------- | | | | . 0(0) ack 1 <MP_CAPABLE> | |----------------------------------------------------------->--------------------------------------------------------> | | | ]]></artwork> </figure> <t>Once this is done, the received cookie can be used for TFO, as shown in <xreftarget="fig_tfodata"/>.target="fig_tfodata" format="default"/>. In this example, the initiator first sends 20 bytes in the SYN. The listener immediately replies with 100 bytes following theSYN-ACK uponSYN-ACK, to which the initiator replies with 20 more bytes. Note that the last segment in the figure has a TCP sequence number of 21, while the DSS subflow sequence number is 1 (because the TFO data is not part of the data sequence number space, as explained inSection<xreftarget="tfodata"/>.</t>target="tfodata" format="default"/>.</t> <figurealign="center" anchor="fig_tfodata" title="The listener supports TFO">anchor="fig_tfodata"> <name>The Listener Supports TFO</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ initiator listener | | | S 0(20) <MP_CAPABLE>, <TFO cookie> | |----------------------------------------------------------->--------------------------------------------------------> | | | | S. 0(0) ack 21 <MP_CAPABLE> | |<-----------------------------------------------------------<-------------------------------------------------------- | | | | . 1(100) ack 21 <DSS ack=1 seq=1 ssn=1 dlen=100> | |<-----------------------------------------------------------<-------------------------------------------------------- | | | | . 21(0) ack 1 <MP_CAPABLE> | |----------------------------------------------------------->--------------------------------------------------------> | | | | . 21(20) ack 101 <DSS ack=101 seq=1 ssn=1 dlen=20> | |----------------------------------------------------------->--------------------------------------------------------> | | | ]]></artwork> </figure> <t>In <xreftarget="fig_tfofallback"/>,target="fig_tfofallback" format="default"/>, the listener does not support TFO. The initiator detects that no state is created in the listener (as no data isacked),ACKed) and now sends the MP_CAPABLE in the thirdack,packet, in order for the listener to build its MPTCP context atthenthe end of the establishment. Now, thetfoTFO data, when retransmitted, becomes part of thedata sequence mappingData Sequence Mapping because it is effectively sent (in factre-sent)re‑sent) after the establishment.</t> <figurealign="center" anchor="fig_tfofallback" title="The listener does not support TFO">anchor="fig_tfofallback"> <name>The Listener Does Not Support TFO</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ initiator listener | | | S 0(20) <MP_CAPABLE>, <TFO cookie> | |----------------------------------------------------------->--------------------------------------------------------> | | | | S. 0(0) ack 1 <MP_CAPABLE> | |<-----------------------------------------------------------<-------------------------------------------------------- | | | | . 1(0) ack 1 <MP_CAPABLE> | |----------------------------------------------------------->--------------------------------------------------------> | | | | . 1(20) ack 1 <DSS ack=1 seq=1 ssn=1 dlen=20> | |----------------------------------------------------------->--------------------------------------------------------> | | | | . 0(0) ack 21 <DSS ack=21 seq=1 ssn=1 dlen=0> | |<-----------------------------------------------------------<-------------------------------------------------------- | | | ]]></artwork> </figure> <t>It is also possible that the listener acknowledges only part of the TFO data, as illustrated in <xreftarget="fig_tfopartial"/>.target="fig_tfopartial" format="default"/>. The initiator will simply retransmit the missing data together with aDSS-mapping.</t>DSS mapping.</t> <figurealign="center" anchor="fig_tfopartial" title="Partial data acknowledgement">anchor="fig_tfopartial"> <name>Partial Data Acknowledgment</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ initiator listener | | | S 0(1000) <MP_CAPABLE>, <TFO cookie> | |----------------------------------------------------------->--------------------------------------------------------> | | | | S. 0(0) ack 501 <MP_CAPABLE> | |<-----------------------------------------------------------<-------------------------------------------------------- | | | | . 501(0) ack 1 <MP_CAPABLE> | |----------------------------------------------------------->--------------------------------------------------------> | | | | . 501(500) ack 1 <DSS ack=1 seq=1 ssn=1 dlen=500> | |----------------------------------------------------------->--------------------------------------------------------> | | | ]]></artwork> </figure> </section> </section> <sectiontitle="Control Blocks" anchor="app_tcb">anchor="app_tcb" numbered="true" toc="default"> <name>Control Blocks</name> <t>Conceptually, an MPTCP connection can be represented as an MPTCP protocol control block (PCB) that contains several variables that track the progress and the state of the MPTCP connection and a set of linked TCP control blocks that correspond to the subflows that have been established.</t> <t>RFC 793 <xreftarget="RFC0793"/>target="RFC0793" format="default"/> specifies several state variables. Whenever possible, we reuse the same terminology asRFC 793RFC 793 to describe the state variables that are maintained by MPTCP.</t> <sectiontitle="MPTCPnumbered="true" toc="default"> <name>MPTCP ControlBlock">Block</name> <t>The MPTCP control block contains the followingvariablevariables per connection.</t> <sectiontitle="Authentication and Metadata"> <t><list style="hanging"> <t hangText="Local.Tokennumbered="true" toc="default"> <name>Authentication and Metadata</name> <dl newline="false" spacing="normal" indent="3"> <dt>Local.Token (32bits):">bits):</dt> <dd> This is the token chosen by the local host on this MPTCP connection. The token must be unique among all established MPTCPconnections,connections and is generated from the localkey.</t> <t hangText="Local.Keykey.</dd> <dt>Local.Key (64bits):">bits):</dt> <dd> This is the key sent by the local host on this MPTCPconnection.</t> <t hangText="Remote.Tokenconnection.</dd> <dt>Remote.Token (32bits):">bits):</dt> <dd> This is the token chosen by the remote host on this MPTCP connection, generated from the remotekey.</t> <t hangText="Remote.Keykey.</dd> <dt>Remote.Key (64bits):">bits):</dt> <dd> This is the key chosen by the remote host on this MPTCPconnection</t> <t hangText="MPTCP.Checksum (flag):">connection.</dd> <dt>MPTCP.Checksum (flag):</dt> <dd> This flag is set to true if at least one of the hosts has set theA"A" bit in the MP_CAPABLE options exchanged during connectionestablishment, andestablishment; otherwise, it is set tofalse otherwise.false. If this flag is set, the checksum must be computed in all DSSoptions.</t> </list></t>options.</dd> </dl> </section> <sectiontitle="Sending Side"> <t><list style="hanging"> <t hangText="SND.UNAnumbered="true" toc="default"> <name>Sending Side</name> <dl newline="false" spacing="normal" indent="3"> <dt>SND.UNA (64bits):">bits):</dt> <dd> This is the data sequence number of the next byte to be acknowledged, at the MPTCP connection level. This variable is updated upon reception of a DSS option containing aDATA_ACK.</t> <t hangText="SND.NXTDATA_ACK.</dd> <dt>SND.NXT (64bits):">bits):</dt> <dd> This is the data sequence number of the next byte to be sent. SND.NXT is used to determine the value of the DSN in the DSSoption.</t> <t hangText="SND.WNDoption.</dd> <dt>SND.WND (32bits with RFC 7323, 16 bits otherwise):">bits):</dt> <dd> This is thesendingsend window. 32 bits if the features in RFC 7323 are used; 16 bits otherwise. MPTCP maintains thesendingsend window at the MPTCP connectionlevellevel, and the same window is shared by all subflows. All subflows use the MPTCPconnection levelconnection-level SND.WND to compute the SEQ.WND value that is sent in each transmittedsegment.</t> </list></t>segment.</dd> </dl> </section> <sectiontitle="Receiving Side"> <t><list style="hanging"> <t hangText="RCV.NXTnumbered="true" toc="default"> <name>Receiving Side</name> <dl newline="false" spacing="normal" indent="3"> <dt>RCV.NXT (64bits):">bits):</dt> <dd> This is the data sequence number of the next byte that is expected on the MPTCP connection. This state variable is modified upon reception of in-order data. The value of RCV.NXT is used to specify the DATA_ACK that is sent in the DSS option on allsubflows.</t> <t hangText="RCV.WNDsubflows.</dd> <dt>RCV.WND (32bits with RFC 7323, 16 bits otherwise):">bits):</dt> <dd> This is the connection-level receive window, which is the maximum of the RCV.WND on all thesubflows.</t> </list></t>subflows. 32 bits if the features in RFC 7323 are used; 16 bits otherwise.</dd> </dl> </section> </section> <sectiontitle="TCPnumbered="true" toc="default"> <name>TCP ControlBlocks">Blocks</name> <t>The MPTCP control block also contains a list of the TCP control blocks that are associated with the MPTCP connection.</t> <t>Note that the TCP control block on the TCP subflows does not contain the RCV.WND and SND.WND statevariablesvariables, as these are maintained at the MPTCP connection level and not at the subflow level.</t> <t>Inside each TCP control block, the following state variables are defined.</t> <sectiontitle="Sending Side"> <t><list style="hanging"> <t hangText="SND.UNAnumbered="true" toc="default"> <name>Sending Side</name> <dl newline="false" spacing="normal" indent="3"> <dt>SND.UNA (32bits):">bits):</dt> <dd> This is the sequence number of the next byte to be acknowledged on the subflow. This variable is updated upon reception of each TCP acknowledgment on thesubflow.</t> <t hangText="SND.NXTsubflow.</dd> <dt>SND.NXT (32bits):">bits):</dt> <dd> This is the sequence number of the next byte to be sent on the subflow. SND.NXT is used to set the value of SEG.SEQ upon transmission of the nextsegment.</t> </list></t>segment.</dd> </dl> </section> <sectiontitle="Receiving Side"> <t><list style="hanging"> <t hangText="RCV.NXTnumbered="true" toc="default"> <name>Receiving Side</name> <dl newline="false" spacing="normal" indent="3"> <dt>RCV.NXT (32bits):">bits):</dt> <dd> This is the sequence number of the next byte that is expected on the subflow. This state variable is modified upon reception of in-order segments. The value of RCV.NXT is copied to the SEG.ACK field of the next segments transmitted on thesubflow.</t> <t hangText="RCV.WNDsubflow.</dd> <dt>RCV.WND (32bits with RFC 7323, 16 bits otherwise):"> Thisbits):</dt> <dd>This is the subflow-level receive window that is updated with the window field from the segments received on thissubflow.</t> </list></t>subflow. 32 bits if the features in RFC 7323 are used; 16 bits otherwise.</dd> </dl> </section> </section> </section> <sectiontitle="Finiteanchor="app_fsm" numbered="true" toc="default"> <name>Finite StateMachine" anchor="app_fsm">Machine</name> <t>The diagram in <xreftarget="fig_fsm"/>target="fig_fsm" format="default"/> shows the Finite State Machine for connection-level closure. This illustrates how the DATA_FIN connection-level signal (indicated in the diagram as the DFIN flag on a DATA_ACK) (1) interacts with subflow-levelFINs,FINs and (2) permits"break-before-make"break-before-make handover between subflows.</t> <figurealign="center" anchor="fig_fsm" title="Finiteanchor="fig_fsm"> <name>Finite State Machine for ConnectionClosure">Closure</name> <artworkalign="left"><![CDATA[align="left" name="" type="" alt=""><![CDATA[ +---------+ | M_ESTAB | +---------+ M_CLOSE | | rcv DATA_FIN ------- | | ------- +---------+ snd DATA_FIN / \ snd DATA_ACK[DFIN]+---------++-------+ | M_FIN |<------------------------------------>| M_CLOSE |------------------->|M_CLOSE| | WAIT-1 |--------------------------- | WAIT | +---------+ rcv DATA_FIN \+---------++-------+ | rcv DATA_ACK[DFIN] ------- | M_CLOSE | | -------------- snd DATA_ACK | ------- | | CLOSE all subflows | snd DATA_FIN | V V V +-----------+ +-----------++-----------++----------+ |M_FINWAIT-2| | M_CLOSING || M_LAST-ACK| +-----------+|M_LAST-ACK| +-----------+ +-----------+ +----------+ | rcv DATA_ACK[DFIN] | rcv DATA_ACK[DFIN] | | rcv DATA_FIN -------------- | -------------- | | ------- CLOSE all subflows | CLOSE all subflows | | snd DATA_ACK[DFIN] V delete MPTCP PCB V \ +-----------++---------++--------+ ------------------------>|M_TIMEWAIT|----------------->| M_CLOSED|WAIT|---------------->|M_CLOSED| +-----------++---------++--------+ All subflows in CLOSED ------------ delete MPTCP PCB ]]></artwork> </figure> </section> <sectiontitle="Changesanchor="app_changelog" numbered="true" toc="default"> <name>Changes fromRFC6824" anchor="app_changelog">RFC 6824</name> <t>Thissectionappendix lists the key technical changes betweenRFC6824, specifying<xref target="RFC6824"/>, which specifies MPTCPv0,v0; and this document, which obsoletesRFC6824<xref target="RFC6824"/> and specifies MPTCP v1. Note that this specification is notbackwardsbackward compatible withRFC6824. <list style="symbols"> <t>The<xref target="RFC6824"/>. </t> <ul spacing="normal"> <li>This document incorporates lessonslearntlearned from the various implementations,deploymentsdeployments, and experiments gathered in the documents "Use Cases and Operational Experience with Multipath TCP" <xreftarget="RFC8041"/>target="RFC8041" format="default"/> and the IETF Journal article "Multipath TCP Deployments" <xreftarget="deployments"/>.</t> <t>Connectiontarget="deployments" format="default"/>.</li> <li>Connection initiation, through the exchange of the MP_CAPABLE MPTCP option, is different fromRFC6824.<xref target="RFC6824"/>. The SYN no longer includes the initiator's key,allowingto allow the MP_CAPABLE option on the SYN to be shorter inlength,length and to avoid duplicating the sending of keyingmaterial.</t> <t>Thismaterial.</li> <li>This also ensures reliable delivery of the key on the MP_CAPABLE option by allowing its transmission to be combined with data and thus using TCP'sin-builtbuilt-in reliability mechanism. If the initiator does not immediately have data to send, the MP_CAPABLE option with the keys will be repeated on the first data packet. If the other end is the first to send, then the presence of the DSS option implicitly confirms the receipt of theMP_CAPABLE.</t> <t>InMP_CAPABLE.</li> <li>In the Flags field of MP_CAPABLE,C"C" is now assigned to mean that the sender of this option will not accept additional MPTCP subflows to the source address and port. Thisis animproves efficiencyimprovement,-- forexampleexample, in cases where the sender is behind a strictNAT.</t> <t>InNAT.</li> <li>In the Flags field of MP_CAPABLE,H"H" now indicates the use of HMAC-SHA256 (rather thanHMAC-SHA1).</t> <t>ConnectionHMAC-SHA1).</li> <li>Connection initiation also defines the procedure for version negotiation, for implementations that support both v0(RFC6824)<xref target="RFC6824"/> and v1 (thisdocument).</t> <t>Thedocument).</li> <li>The HMAC-SHA256 (rather than HMAC-SHA1) algorithm is used, asthe algorithmit provides better security. It is used to generate the token in the MP_JOIN and ADD_ADDRmessages,messages and to set theinitial data sequence number.</t> <t>AIDSN.</li> <li>A new subflow-level option exists to signal reasons for sending a RST on a subflow (MP_TCPRST<xref target="sec_reset"/>), which(<xref target="sec_reset" format="default"/>)); this can help an implementation decide whether to attempt laterre-connection.</t> <t>Thereconnection.</li> <li>The MP_PRIO option (<xreftarget="sec_policy"/>),target="sec_policy" format="default"/>), which is used to signal a change of priority for a subflow, no longer includes the AddrID field. Its purpose was to allow the changed priority to be applied on a subflow other than the one it was sent on. However, ithas been realisedwas determined that this could be used by a man-in-the-middle to divert all trafficon toonto its own path, and MP_PRIO does not include a token or other type of securitymechanism.</t> <t>Themechanism.</li> <li>The ADD_ADDR option (<xreftarget="sec_add_address"/>),target="sec_add_address" format="default"/>), which is used to inform the other host about another potential address, is different in several ways. It now includes an HMAC of the added address, for enhanced security. In addition, reliability for the ADD_ADDR option has been added: the IPVer field is replaced with a flag field, and one flag is assigned(E) which("E") that is used as an'Echo'"echo" so a host can indicate that it has received theoption.</t> <t>Anoption.</li> <li>This document describes an additional way of performing a Fast Closeis described,-- by sendingaan MP_FASTCLOSE option on a RST on all subflows. This allows the host to tear down the subflows and the connectionimmediately.</t> <t>Inimmediately.</li> <li>IANA has reserved theIANA registry a newMPTCP option subtypeoption, MP_EXPERIMENTAL, is reservedof value 0xf forprivate experiments. However, thePrivate Use (<xref target="IANA_subtypes"/>). This document doesn't define how to usethe subtype option.</t> <t>Athat value.</li> <li>This document adds a newAppendixappendix (<xref target="app_tfo" format="default"/>), which discusses the usage of boththeMPTCP options andTCP Fast OpenTFO options on the samepacket (<xref target="app_tfo"/>).</t> </list></t>packet.</li> </ul> </section> <section anchor="Acknowledgments" numbered="false" toc="default"> <name>Acknowledgments</name> <t>The authors gratefully acknowledge significant input into this document from <contact fullname="Sebastien Barre"/> and <contact fullname="Andrew McDonald"/>.</t> <t>The authors also wish to acknowledge reviews and contributions from <contact fullname="Iljitsch van Beijnum"/>, <contact fullname="Lars Eggert"/>, <contact fullname="Marcelo Bagnulo"/>, <contact fullname="Robert Hancock"/>, <contact fullname="Pasi Sarolahti"/>, <contact fullname="Toby Moncaster"/>, <contact fullname="Philip Eardley"/>, <contact fullname="Sergio Lembo"/>, <contact fullname="Lawrence Conroy"/>, <contact fullname="Yoshifumi Nishida"/>, <contact fullname="Bob Briscoe"/>, <contact fullname="Stein Gjessing"/>, <contact fullname="Andrew McGregor"/>, <contact fullname="Georg Hampel"/>, <contact fullname="Anumita Biswas"/>, <contact fullname="Wes Eddy"/>, <contact fullname="Alexey Melnikov"/>, <contact fullname="Francis Dupont"/>, <contact fullname="Adrian Farrel"/>, <contact fullname="Barry Leiba"/>, <contact fullname="Robert Sparks"/>, <contact fullname="Sean Turner"/>, <contact fullname="Stephen Farrell"/>, <contact fullname="Martin Stiemerling"/>, <contact fullname="Gregory Detal"/>, <contact fullname="Fabien Duchene"/>, <contact fullname="Xavier de Foy"/>, <contact fullname="Rahul Jadhav"/>, <contact fullname="Klemens Schragel"/>, <contact fullname="Mirja Kühlewind"/>, <contact fullname="Sheng Jiang"/>, <contact fullname="Alissa Cooper"/>, <contact fullname="Ines Robles"/>, <contact fullname="Roman Danyliw"/>, <contact fullname="Adam Roach"/>, <contact fullname="Eric Vyncke"/>, and <contact fullname="Ben Kaduk"/>.</t> </section> </back> </rfc>