TCP Maintenance and Minor L. Eggert Extensions (tcpm)
NECNokia Internet-Draft F. Gont Intended status: Standards Track UTN/FRH Expires: April 25,September 6, 2007 March 5, 2007 October 22, 2006TCP User Timeout Option draft-ietf-tcpm-tcp-uto-04draft-ietf-tcpm-tcp-uto-05 Status of this Memo By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt. The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. This Internet-Draft will expire on April 25,September 6, 2007. Copyright Notice Copyright (C) The Internet Society (2006).IETF Trust (2007). Abstract The TCP user timeout controls how long transmitted data may remain unacknowledged before a connection is forcefully closed. It is a local, per-connection parameter. This document specifies a new TCP option - the TCP User Timeout Option - that allows one end of a TCP connection to advertise its current user timeout for a connection. Thus,value. This information allows the remote TCP may modifyother end to adapt its local user timeout based on knowledge of the peer's user timeout. The TCPuser timeout controls how long transmitted data may remain unacknowledged before a connection is forcefully closed. It is a local, per-connection parameter.accordingly. Increasing the user timeouts allows establishedon both ends of a TCP connectionsconnection allows it to survive extended periods of disconnection.without end-to-end connectivity. Decreasing the user timeouts allows busy servers to explicitly notify their clients that they will maintain the connection state only acrossfor a short periods of disconnection.time without connectivity. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 2. Conventions . . . . . . . . . . . . . . . . . . . . . . . . . 4 3. Operation . . . . . . . . . . . . . . . . . . . . . . . . . . 4 3.1. Changing the Local User Timeout . . . . . . . . . . . . . 6 3.2. UTO Option Reliability Considerations. . . . . . . . . . . . . . . . . . 8 3.3. Option Format . . . . . . . . . . . . . . . . . . . . . . 98 3.4. SpecialReserved Option Values . . . . . . . . . . . . . . . . . . 9 4. Interoperability Issues . . . . . . . . . . . . . . . . . . . 109 4.1. Middleboxes . . . . . . . . . . . . . . . . . . . . . . . 109 4.2. TCP Keep-Alives . . . . . . . . . . . . . . . . . . . . . 10 5. Security Considerations . . . . . . . . . . . . . . . . . . . 1110 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 1211 7. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 1211 8. References . . . . . . . . . . . . . . . . . . . . . . . . . . 1312 8.1. Normative References . . . . . . . . . . . . . . . . . . . 1312 8.2. Informative References . . . . . . . . . . . . . . . . . . 1312 Editorial Comments . . . . . . . . . . . . . . . . . . . . . . . . Appendix A. Alternative solutions . . . . . . . . . . . . . . . . 14 Appendix B.Document Revision History . . . . . . . . . . . . . . 1413 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 1514 Intellectual Property and Copyright Statements . . . . . . . . . . 1716 1. Introduction The Transmission Control Protocol (TCP) specification [RFC0793]defines[RFC0793] defines a local, per-connection "user timeout" parameter that specifies the maximum amount of time that transmitted data may remain unacknowledged before TCP will forcefully close the corresponding connection. Applications can set and change this parameter with OPEN and SEND calls. If a network disconnectionan end-to-end connectivity disruption lasts longer than the user timeout, no acknowledgments will be received for any transmission attempt, including keep-alives [TCP-ILLU],keep-alives, and the TCP connection will close when the user timeout occurs. In the absence of an application-specified user timeout, the TCP specification [RFC0793] defines a default user timeout of 5 minutes. The Host Requirements RFC [RFC1122] refines this definition by introducing two thresholds, R1 and R2 (R2 > R1), onthat control the number of retransmissions ofretransmission attempts for a single segment. It suggests that TCP should notify applications when R1 is reached for a segment, and close the connection oncewhen R2 is reached. [RFC1122] also defines the recommended values for R1 (three retransmissions) and R2 (100 seconds), noting that R2 for SYN segments should be at least 3 minutes. Instead of a single user timeout, some TCP implementations offer finer-grained policies. For example, Solaris supports different timeouts depending on whether a TCP connection is in the SYN-SENT, SYN-RECEIVED, or ESTABLISHED state [SOLARIS-MANUAL]. Although some TCP implementations allow applications to set their local user timeout, there isTCP has no in-protocol mechanism to signal changes into the local user timeout to remote peers.the other end. This causes local changes to be ineffective,ineffective in allowing a connection to survive extended periods without connectivity, because the peerother end will still close the connection after its user timeout expires, even when the host has raised its local user timeout.expires. The ability to suggestinform the remote peer aother end about the local user timeout to be usedfor the connection can improve TCP'sTCP operation in scenarios that are currently not well supported. One example of such scenarios are mobile hosts that change network attachment points based on current location. Such hosts, maybe using MobileIPMobile IP [RFC3344], HIP [RFC4423] or transport-layer mobility mechanisms [I-D.eddy-tcp-mobility], are only intermittently connected to the Internet. In between connected periods, mobile hosts may experience periods of disconnection during which no network service is available.without end-to-end connectivity. Other factors that can cause transient periods of disconnectionconnectivity disruptions are high levels of congestion as well asor link or routing failures inside the network. In scenarios similar to the ones described above,these scenarios, a host may not know exactly when or for how long itconnectivity disruptions will be disconnected from the network,occur, but it might expectbe able to determine an increased likelihood for such events due tobased on past mobility patterns and thus benefit from using longer user timeouts. In other scenarios, the length andtime and duration of a network disconnectionconnectivity disruption may even be predictable. For example, an orbiting node on a non-geostationary satellite might experience disconnectionsconnectivity disruptions due to line-of-sight blocking by other planetary bodies. The disconnection periodstiming of such a nodethese events may be easilycomputable from orbital mechanics. This document specifies a new TCP option - the TCP User Timeout Option (UTO)- that allows one end of a TCP connection to advertise its current localuser timeout parameter. Thus, based on thevalue. This information advertised byallows the remote TCP peer, a TCP may modifyother end to adapt its ownuser timeout accordingly. This allows, for example, mobile hosts to maintain TCP connections across disconnected periods that are longer than their peer's defaultIncreasing the user timeout. A second usetimeouts on both ends of thea TCP User Timeout Option is advertisement of shorter-than-defaultconnection allows it to survive extended periods without end-to-end connectivity. Decreasing the user timeouts. This can allowtimeouts allows busy servers to explicitly notify their clients that they will maintain the connection state associated with established connectionsonly across short periods of disconnection. Use of the TCP User Timeout Option could be triggered either by an API call or by a system-wide toggle. The API could be,for example, a Socket option that would need to be explicitly set by the corresponding application. This option would default to "off". A system-wide toggle would allow a system administrator to enable the use of the TCP User Timeout Option on a system-wide basis, and set the option a desired value. This system-wide toggle would allow the use of the option by application programs that have not been explicitly coded to do so. If such a system-wide toggle were provided, it would default to "off". In all cases, use of the TCP User Timeout Option would depend on an active decision, either by the application programmer (by means of an API call), or by a system administrator (by means ofa system-wide toggle).short time without connectivity. 2. Conventions The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119]. 3. Operation Sending aUse of the TCP User Timeout Option informscan be enabled either on a per- application basis, e.g., through a socket option, or controlled by a system-wide setting. TCP maintains three per-connection state variables to control the remote peeroperation of the current local user timeout for the connection,UTO options, two of which (ENABLED and suggests the TCP peerCHANGEABLE) are new: ENABLED (Boolean) Flag that controls whether UTO options are enabled for a connection. Defaults to adapt its user timeout accordingly. Thefalse. LOCAL_UTO Local user timeout value includedin a TCP User Timeout Option specifieseffect for this connection. This is either the requestedsystem-wide default or an application-specified value. Defaults to the system-wide default. CHANGEABLE (Boolean) Flag that controls whether the local user timeout duringmay be changed based on UTO options received from the synchronized states of a connection (ESTABLISHED, FIN-WAIT-1, FIN-WAIT-2, CLOSE-WAIT, CLOSING, or LAST-ACK). Connections inother states MUST the default timeout values defined in [RFC0793] [RFC1122].end. Defaults to true and becomes false when an application explicitly sets LOCAL_UTO. Note that an exchange of TCP User Timeout OptionsUTO options between peersboth ends of a connection is not a binding negotiation. Transmission of a TCP User Timeout OptionUTO option is an advisorya suggestion that the peerother end consider adapting its localuser timeout. Hosts remain free to adoptThis adaptation only happens if the the other end has explicitly enabled it (CHANGEABLE is true). Before opening a different user timeout, orconnection, an application that wishes to forcefully close or abort connections at any time for any reason, whether or not theyuse custom user timeouts or have suggestedUTO options SHOULD enable their use by setting ENABLED to true. It MAY pick an appropriate local UTO by setting LOCAL_UTO, which is otherwise set to the peersystem default. Finally, the application should determine whether it will allow the local UTO to use them. A host that supportschange based on received UTO options from the TCP User Timeout Option SHOULD include one in each packet that carries a SYN flag.other end. The presence of this optiondefault is not a negotiation of the capability, but simply an advisory message specifying the currently preferred user timeout value. This allows TCPto adoptallow this for connections that do not have a specific user timeout with knowledge ofconcerns, i.e., connections that used byoperate with the peer TCPdefault LOCAL_UTO. If an application explicitly sets LOCAL_UTO, CHANGEABLE MUST become false, to prevent UTO options from the very beginning ofother end to override local application requests. Alternatively, applications MAY set or clear CHANGEABLE directly. Performing these steps before an active or passive open causes UTO options to be exchanged in the data transfer phase. Additionally,SYN and SYN-ACK packets and is a TCP that supportsreliable way to initially exchange and potentially adapt to UTO values. Systems MAY provide system-wide default settings for the User Timeout OptionENABLED, LOCAL_UTO and has sent aCHANGEABLE connection parameters when applications do not initialize them themselves. In addition to exchanging UTO options in the SYN segment as a result of an active OPENsegments, a connection that has enabled UTO options SHOULD include ana UTO option in the first packet that does not have the SYN flag set. This helps to minimize the amount of state information aTCP must keep for connections in non-synchronized states, and is particularly useful when mechanisms such as "SYN cookies" [I-D.ietf-tcpm-syn-flood] are implemented, allowing a newly-established TCP connection to benefit from the information advertised by the UTO option, even if the UTO contained in the initial SYN segment was not recorded. A host that supports the TCP User Timeout OptionUTO options SHOULD include it in the next possible outgoing segment to its peerwhenever it starts using a new user timeout for the connection. This allows the peerother end to adapt its local user timeout for the connection accordingly. When a host that supports the TCP User Timeout Option receives one, it will use the received value to compute the local user timeout for the connection. Generally, hosts should honor requests for changes to the user timeout (see Section 3.1), unless security concerns, resource constraints or external policies indicate otherwise (see Section 5). If so, hosts may use a different user timeout for the connection.A TCP implementation that does not support the TCP User Timeout OptionUTO options MUST silently ignore itthem [RFC1122], thus ensuring interoperability. Hosts MUST impose upper and lower limits on the user timeouts they use.use for a connection. Section 3.1 discusses user timeout limits, and describes a RECOMMENDED scheme to apply them. A TCP User Timeout Option with a value of zero (i.e., "now") is nonsensicallimits and is used for a special purpose, see Section 3.4. Section 3.1discusses potentially problematic effects of otheruser timeout durations.settings. 3.1. Changing the Local User Timeout When a host receives a TCP User Timeout Option, it must decide whether to change the local user timeout of the corresponding connection. Application-requestedIf the CHANGEABLE flag is false, LOCAL_UTO MUST NOT be changed, regardless of the received UTO option. Without this restriction, UTO would modify TCP semantics, because application- requested UTOs could be overridden by peer requests. In this case, they SHOULD, however, notify the application about the user timeout values always take precedence over timeout valuesvalue received from the peer in a TCP User Timeout Option. [anchor3] Consequently,other end. In general, unless the application on the local host has requested a specific user timeoutLOCAL_UTO for the connection, e.g., through the OPEN or SEND calls,CHANGEABLE will be true and hosts SHOULD adjust their local user timeout in response to receiving a TCP User Timeout Option, as described in the remainder of this section. Ifthe local application has requested a specific local user timeout, TCP implementations MUST NOT change it in response to receiving a TCP User Timeout Option. In this case, they SHOULD, however, notify the application about theuser timeout value received fromin response to receiving a UTO option, as described in the peer.remainder of this section. The User Timeout OptionUTO option specifies the user timeout in terms of time units,in seconds or minutes, rather than in terms ofnumber of retransmissions or round- tripround-trip times (RTTs), as in most cases the periods of disconnection have to do with operation and mobility patterns, rather than with the current network conditions.(RTTs). Thus, the TCP User Timeout OptionUTO option allows hosts to exchange user timeout values from 1 second to over 9 hours at a granularity of seconds, and from 1 minute to over 22 days at a granularity of minutes. (An option value of zero is reserved for a special purpose, see Section 3.4.)Very short user timeout values can affect TCP transmissions over high-delay paths. If the user timeout occurs before an acknowledgment for an outstanding segment arrives, possibly due to packet loss, the connection closes. Many TCP implementations default to user timeout values of a few minutes [TCP-ILLU].minutes. Although the TCP User Timeout OptionUTO option allows suggestion of short timeouts, applications advertising them should consider these effects. Long user timeout values allow hosts to tolerate extended periods of disconnection.without end-to-end connectivity. However, they also require hosts to maintain the TCP state information associated with connections for long periods of time. Section 5 discusses the security implications of long timeout values. To protect against these effects, implementations MUST impose limits on the user timeout values they accept and use. The remainder of this section describes a RECOMMENDED scheme to limit user timeouts based on upper and lower limits. Under the RECOMMENDED scheme, and when CHANGEABLE is true, each TCPend SHOULD compute the user timeout (USER_TIMEOUT)LOCAL_UTO for a connection according to this formula: USER_TIMEOUTLOCAL_UTO = min(U_LIMIT, max(LOCAL_UTO, REMOTE_UTO, L_LIMIT)) Each field is to be interpreted as follows: USER_TIMEOUT Resulting user timeout value to be adopted by the local TCP for a connection.U_LIMIT Current upper limit imposed on the user timeout of a connection by the local host. L_LIMIT Current lower limit imposed on the user timeout of a connection by the local host. LOCAL_UTO Current local user timeout of this specific connection.REMOTE_UTO Last "user timeout" value suggested by the remote peer by means ofreceived from the other end in a TCP User Timeout Option. This means that, provided they are within the upper and lower limits, the maximum of current LOCAL_UTO and the two announced values will be adopted for thelast user timeout ofvalue received from the other end will become the new LOCAL_UTO for the connection. The rationale is that choosing the maximum of the two values will let the connection survive longer periods of disconnection.without end-to- end connectivity. If the TCPend that announced the lower of the two user timeout values did so in order to reduce the amount of TCP state information that must be kept on the host, it can, nevertheless, close or abort the connection whenever it wants. [anchor3] It must be noted that the two endpoints of the connection will not necessarily adopt the same user timeout. Enforcing a lower limit (L_LIMIT) prevents connections from closing due to transient network conditions, including temporary congestion, mobility hand-offs and routing instabilities. An upper limit (U_LIMIT) can reduce the effect of resource exhaustion attacks. Section 5discusses5 discusses the details of these attacks. Note that these limits MAY be specified as system-wide constants or at other granularities, such as on per-host, per-user or even per- connection basis. Furthermore, these limits need not be static. For example, they MAY be a function of system resource utilization or attack status and could be dynamically adapted. The Host Requirements RFC [RFC1122] does not impose any limits on the length of the user timeout. However, a time interval of at least 100 seconds is RECOMMENDED. Consequently, the lower limit (L_LIMIT) SHOULD be set to at least 100 seconds when following the RECOMMENDED scheme described in this section. Adopting a user timeout smaller than the current retransmission timeout (RTO) for the connection would likely cause the connection to be aborted unnecessarily. Therefore, the lower limit (L_LIMIT) MUST be larger than the current retransmission timeout (RTO) for the connection. It is worth noting that an upper limit may be imposed on the RTO, provided it is at least 60 seconds [RFC2988]. 3.2. UTO Option Reliability ConsiderationsThe TCP User Timeout Option is an advisory TCP option that does not change processing of subsequent segments. Unlike other TCP options, it need not be exchanged reliably. Consequently, the specification in this sectiondoes not define a reliability handshake for TCP User Timeout OptionUTO option exchanges. When a segment that carries a TCP User Timeout OptionUTO option is lost, the option may never reachother end will simply not have the intended peer.opportunity to update its local UTO. Implementations MAY implement local mechanisms to improve delivery reliability, such as retransmitting the TCP User Timeout Optiona UTO option when they retransmit thea segment that originally carried it, or "attaching" the option to a byte in the stream and retransmitting the option whenever that byte or its ACK are retransmitted. It is important to note that although these mechanisms can improve transmission reliability for the TCP User Timeout Option,UTO options, they do not guarantee delivery (a three-way handshake would be required for this). Consequently, implementations should notMUST NOT assume that UTO options are transmitted reliably. 3.3. Option Format Sending a TCP User Timeout Option informs the other end of the current local user timeout for the connection and suggests that the other end adapt its user timeout accordingly. The user timeout value included in a UTO option contains the local user timeout (LOCAL_UTO) used during the synchronized states of a TCP User Timeout Option is reliably transmitted. 3.3. Option Formatconnection (ESTABLISHED, FIN-WAIT-1, FIN-WAIT-2, CLOSE-WAIT, CLOSING, or LAST-ACK). Connections in other states MUST use the default timeout values defined in [RFC0793] and [RFC1122]. 0 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 = X | Length = 4 |G| User Timeout | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ (One tick mark represents one bit.) Figure 1: Format of the TCP User Timeout Option Figure 1 shows the format of the TCP User Timeout Option. It contains these fields: Kind (8 bits) A TCP option number [RFC0793] to be assigned by IANA upon publication of this document (see Section 6). Length (8 bits) Length of the TCP option in octets [RFC0793]; its value MUST be 4. Granularity (1 bit) Granularity bit, indicating the granularity of the "User Timeout" field. When set (G = 1), the time interval in the "User Timeout" field MUST be interpreted as minutes. Otherwise (G = 0), the time interval in the "User Timeout" field MUST be interpreted as seconds. User Timeout (15 bits) Specifies the user timeout suggestion(LOCAL_UTO) used for this connection. It MUST be interpreted as a 15-bit unsigned integer. The granularity of the timeout (minutes or seconds) depends on the "G" field. 3.4. SpecialReserved Option Values Whenever it is legal to do so according to the specification in the previous sections, TCP implementations MAY send a zero-second TCP User Timeout Option, i.e, with a "User Timeout" field of zero and a "Granularity" of zero. This signals their peers that they support the option, but do not suggest a specific user timeout value at that time. Essentially, a zero-second TCP User Timeout Option acts as a "don't care" value. Thus, the sender SHOULD adapt its local user timeout according to the peer's UTO, and the receiver SHOULD continue using its local user timeout. In order to achieve this, the receiver of a zero-second TCP User Timeout Option SHOULD perform the RECOMMENDED strategy for calculating a new local USER_TIMEOUT described in Section 3.1, with a numeric value of zero seconds for REMOTE_UTO. The sender SHOULD perform the same calculation as described in Section 3.1, with a numeric value of zero seconds for LOCAL_UTO. A zero-minuteAn empty TCP User Timeout Option, i.e., one with a "User Timeout" field of zero and a "Granularity" bit of one,either minutes (1) or seconds (0), is reserved for future use. TCP implementations MUST NOT send it and MUST ignore it upon reception. 4. Interoperability Issues This section discusses interoperability issues related to introducing the TCP User Timeout Option. 4.1. Middleboxes A TCP implementation that does not support the TCP User Timeout Option MUST silently ignore it [RFC1122], thus ensuring interoperability. In a study of the effects of middleboxes on transport protocols, Medina et al. have shown that unknown TCP options are correctly handled bythe vast majority of modern TCP stacks correctly handle unknown TCP options [MEDINA]. In this study, 3% of connections failed when an unknown TCP option appeared in the middle of a connection. Because thesethe number of failures caused by unknown options is small and they are a result of incorrectly implemented TCP stacks that violate existing requirements to ignore unknown options, they do not warrant takingspecial measures to handle these cases. In particular, we domeasures. Thus, this document does not define a separatemechanism to negotiate support of the TCP User Timeout Option onduring the three-way handshake. Stateful firewalls usually reset connections after a period of inactivity. If such a firewall exists along the path between two peers,path, it may close or abort connections regardless of the use of the TCP User Timeout Option. In the future, such firewalls may learn to parse the TCP User Timeout Option and modify their behavior - or the option contents - accordingly. 4.2. TCP Keep-Alives Some TCP implementations, such as the onethose in BSD systems, use a different abort policy for TCP keep-alives than for user data. Thus, the TCP keep-alive mechanism might abort a connection that would otherwise have survived the transient period of disconnection.without connectivity. Therefore, if a TCP peerconnection enables TCPkeep-alives for a connectionthat is also using the TCP User Timeout Option, then the keep-alive timer MUST be set to a value larger than that of the adopted USER TIMEOUT. 5. Security Considerations Lengthening user timeouts has obvious security implications. Flooding attacks cause denial of service by forcing servers to commit resources for maintaining the state of throw-away connections. However, TCP implementations do not become more vulnerable to simple SYN flooding by implementing the TCP User Timeout Option, because user timeouts exchanged during the handshake only affect the synchronized states (ESTABLISHED, FIN-WAIT-1, FIN-WAIT-2, CLOSE-WAIT, CLOSING, LAST-ACK), which simple SYN floods never reach. However, when an attacker completes the three-way handshakes of its throw-away connections it can amplify the effects of resource exhaustion attacks, because the attacked server must maintain the connection state associated with the throw-away connections for longer durations. Because connection state is kept longer, lower- frequency attack traffic, which may be more difficult to detect, can already causeexacerbate resource exhaustion. Several approaches can help mitigate this issue. First, implementations can require prior peer authentication, e.g., using IPsec [RFC4301], before accepting long user timeouts for the peer's connections. Similarly, a host can start to accept long user timeouts for an established connection only after in-band authentication has occurred, for example, after a TLS handshake across the connection has succeeded [RFC2246].[RFC4346]. Although these are arguably the most complete solutions, they depend on external mechanisms to establish a trust relationship. A second alternative that does not depend on external mechanisms would introduce a per-peer limit on the number of connections that may use increased user timeouts. Several variants of this approach are possible, such as fixed limits or shortening accepted user timeouts with a rising number of connections. Although this alternative does not eliminate resource exhaustion attacks from a single peer, it can limit their effects. Reducing the number of high-UTO connections a server supports in the face of an attack turns that attack into a denial-of-service attack against the service of high-UTO connections. Per-peer limits cannot protect against distributed denial of service attacks, where multiple clients coordinate a resource exhaustion attack that uses long user timeouts. To protect against such attacks, TCP implementations could reduce the duration of accepted user timeouts with increasing resource utilization. TCP implementations under attack may be forced to shed load by resetting established connections. Some load-shedding heuristics, such as resetting connections with long idle times first, can negatively affect service for intermittently connected, trusted peers that have suggested long user timeouts. On the other hand, resetting connections to untrusted peers that use long user timeouts may be effective. In general, using the peers' level of trust as a parameter during the load-shedding decision process may be useful. Note that if TCP needs to close or abort connections with a long TCP User Timeout Option to shed load, these connections are still no worse off than without the option. Finally, upper and lower limits on user timeouts, discussed in Section 3.1, can be an effective tool to limit the impact of these sorts of attacks. 6. IANA Considerations This section is to be interpreted according to [RFC2434]. This document does not define any new namespaces. It uses an 8-bit TCP option number maintained by IANA at http://www.iana.org/assignments/tcp-parameters. 7. Acknowledgments The following people have improved this document through thoughtful suggestions: Mark Allman, Caitlin Bestler, David Borman, Bob Braden, Marcus Brunner, Wesley Eddy, Gorry Fairhurst, Abolade Gbadegesin, Ted Faber, Guillermo Gont, Tom Henderson, Joseph Ishac, Jeremy Harris, Phil Karn, Michael Kerrisk, Dan Krejsa, Jamshid Mahdavi, Kostas Pentikousis, Juergen Quittek, Joe Touch, Stefan Schmid, Simon Schuetz, Tim Shepard and Martin Stiemerling. Lars Eggert is partly funded by Ambient Networks, a research project supported by the European Commission under its Sixth Framework Program. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the Ambient Networks project or the European Commission. 8. References 8.1. Normative References [RFC0793] Postel, J., "Transmission Control Protocol", STD 7, RFC 793, September 1981. [RFC1122] Braden, R., "Requirements for Internet Hosts - Communication Layers", STD 3, RFC 1122, October 1989. [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. [RFC2434] Narten, T. and H. Alvestrand, "Guidelines for Writing an IANA Considerations Section in RFCs", BCP 26, RFC 2434, October 1998. 8.2. Informative References [I-D.eddy-tcp-mobility] Eddy, W., "Mobility Support For TCP", draft-eddy-tcp-mobility-00 (work in progress), April 2004. [I-D.ietf-tcpm-syn-flood] Eddy, W., "TCP SYN Flooding Attacks and Common Mitigations", draft-ietf-tcpm-syn-flood-00draft-ietf-tcpm-syn-flood-01 (work in progress), JulyDecember 2006. [MEDINA] Medina, A., Allman, M., and S. Floyd, "Measuring Interactions Between Transport Protocols and Middleboxes", Proc. 4th ACM SIGCOMM/USENIX Conference on Internet Measurement , October 2004. [RFC2246] Dierks, T.[RFC2988] Paxson, V. and C. Allen, "The TLS Protocol Version 1.0",M. Allman, "Computing TCP's Retransmission Timer", RFC 2246, January 1999.2988, November 2000. [RFC3344] Perkins, C., "IP Mobility Support for IPv4", RFC 3344, August 2002. [RFC4301] Kent, S. and K. Seo, "Security Architecture for the Internet Protocol", RFC 4301, December 2005. [RFC4346] Dierks, T. and E. Rescorla, "The Transport Layer Security (TLS) Protocol Version 1.1", RFC 4346, April 2006. [RFC4423] Moskowitz, R. and P. Nikander, "Host Identity Protocol (HIP) Architecture", RFC 4423, May 2006. [SOLARIS-MANUAL] Sun Microsystems, "Solaris Tunable Parameters Reference Manual", Part No. 806-7009-10, 2002. [TCP-ILLU] Stevens, W., "TCP/IP Illustrated, Volume 1: The Protocols", Addison-Wesley , 1994.Editorial Comments [anchor3] Without this, UTO would modify TCP semantics, because application-requested UTOs could be overridden by peer requests. Appendix A. Alternative solutions The same benefits could be obtained through an application-layer mechanism, i.e., exchanging user timeout information via application messages and having the application adjust the user timeouts throughLars: With the TCP API on both sides of a connection. This approach would not require anew TCP option, but would requireCHANGEABLE flag, which prevents changing all application implementations that desire to tolerate extended periodsof disconnection, and in most cases would also require a modification to the corresponding application layer protocol. With the proposed TCP option, application changes may not be necessary at all, or may be restricted to sender- or receiver-side only, and there is no need to modify the correspondingLOCAL_UTO when an application protocol. A different approach to tolerate longer periods of disconnection would be to simply increase the system-wide user timeout on both peers. This approachhas the benefit of not requiring a new TCP option or application changes. However,indicated that it can also significantly increasecares about the amount of connection state a busy server must maintain, because a longer global timeout value would apply to all its connections. The proposed TCP User Timeout Option, onvalue, I think the formula can become LOCAL_UTO = min(U_LIMIT, max(REMOTE_UTO, L_LIMIT)), i.e., we adopt whatever the other hand, allows hostsend suggests, given that it is with in acceptable limits. I didn't want to selectively manage the user timeouts of individual connections, reducing the amount of state they must maintain across disconnected periods.make this change without discussing it first, however. Appendix B.A. Document Revision History To be removed upon publication +----------+--------------------------------------------------------+ | Revision | Comments | +----------+--------------------------------------------------------+ | 05 | Made behavior on when to change/not change the local | | | UTO in response to incoming options consistent through | | | the document. This required some reshuffling of text | | | and also removed the need for the special "don't care" | | | option value. | | 04 | Clarified the results obtained by Medina et al. Added | | | text to suggest inclusion of the UTO in the first | | | non-SYN segment by the TCP that sent a SYN in response | | | to an active OPEN. | | 03 | Corrected use of RFC2119 terminology. Clarified how | | | use of the TCP UTO is triggered. Clarified reason for | | | sending a UTO in the SYN and SYN/ACK segments. | | | Removed discussion of the SO_SNDTIMEO and SO_RCVTIMEO | | | options. Removed text that suggested that a UTO | | | should be sent upon receipt of an UTO from the remoteother | | | peer.end. Required minimum value for the lower limit of | | | the user timeout. Moved alternative solutions to | | | appendix. Miscellaneous editorial changes. | | 02 | Corrected terminology by replacing terms like | | | "negotiate", "coordinate", etc. that were left from | | | pre-WG-document times when the UTO was a more | | | formalized exchange instead of the advisory one it is | | | now. Application-requested UTOs take precedence over | | | ones received from the peer (pointed out by Ted | | | Faber). Added a brief mention of SO_SNDTIMEO and a | | | slightly longer discussion of SO_RCVTIMEO. | | 01 | Clarified and corrected the description of the | | | existing user timeout in RFC793 and RFC1122. Removed | | | distinction between operating during the 3WHS and the | | | established states and introduced zero-second "don't | | | care" UTOs in response to mailing list feedback. | | | Updated references and addressed many other comments | | | from the mailing list. | | 00 | Resubmission of | | | draft-eggert-gont-tcpm-tcp-uto-option-01.txt to the | | | secretariat after WG adoption. Thus, permit | | | derivative works. Updated Lars Eggert's funding | | | attribution. Updated several references. No | | | technical changes. | +----------+--------------------------------------------------------+ Authors' Addresses Lars Eggert NEC Network Laboratories Kurfuerstenanlage 36 Heidelberg 69115 GermanyNokia Research Center P.O. Box 407 Nokia Group 00045 Finland Phone: +49 6221 90511 43 Fax: +49 6221 90511 55+358 50 48 24461 Email: email@example.com@nokia.com URI: http://www.netlab.nec.de/http://research.nokia.com/people/lars_eggert/ Fernando Gont Universidad Tecnologica Nacional / Facultad Regional Haedo Evaristo Carriego 2644 Haedo, Provincia de Buenos Aires 1706 Argentina Phone: +54 11 4650 8472 Email: firstname.lastname@example.org URI: http://www.gont.com.ar/ Full Copyright Statement Copyright (C) The Internet Society (2006).IETF Trust (2007). This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights. This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETYSOCIETY, THE IETF TRUST AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Intellectual Property The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79. Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr. The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at email@example.com. Acknowledgment Funding for the RFC Editor function is provided by the IETF Administrative Support Activity (IASA).