One document matched: draft-ietf-tsvwg-udp-guidelines-03.xml
<?xml version="1.0" encoding="utf-8"?>
<?xml-stylesheet type="text/xsl" href="rfc2629.xslt" ?>
<!DOCTYPE rfc SYSTEM "rfc2629.dtd">
<?rfc toc="yes"?>
<?rfc strict="yes"?>
<?rfc tocompact="yes"?>
<?rfc compact="yes"?>
<?rfc subcompact="no"?>
<?rfc tocdepth="2"?>
<?rfc symrefs="yes"?>
<?rfc comments="yes" ?>
<!--<?rfc rfcedstyle="yes"?>-->
<?rfc sortrefs="yes" ?>
<!-- noModification3978 -->
<rfc category="bcp" ipr="full3978" docName="draft-ietf-tsvwg-udp-guidelines-03">
<front>
<title abbrev="UDP Usage Guidelines">
UDP Usage Guidelines for Application Designers
</title>
<author initials="L." surname="Eggert" fullname="Lars Eggert">
<organization abbrev="Nokia">
Nokia Research Center
</organization>
<address>
<postal>
<street>P.O. Box 407</street>
<code>00045</code> <city>Nokia Group</city>
<country>Finland</country>
</postal>
<phone>+358 50 48 24461</phone>
<email>lars.eggert@nokia.com</email>
<uri>
http://research.nokia.com/people/lars_eggert/
</uri>
</address>
</author>
<author initials="G." surname="Fairhurst" fullname="Godred Fairhurst">
<organization>University of Aberdeen</organization>
<address>
<postal>
<street>Department of Engineering</street>
<street>Fraser Noble Building</street>
<city>Aberdeen</city>
<code>AB24 3UE</code>
<country>Scotland</country>
</postal>
<email>gorry@erg.abdn.ac.uk</email>
<uri>http://www.erg.abdn.ac.uk/</uri>
</address>
</author>
<date year="2007"/>
<area>Transport Area</area>
<workgroup>Transport Area Working Group</workgroup>
<keyword>UDP</keyword>
<keyword>guidelines</keyword>
<abstract>
<t>
The User Datagram Protocol (UDP) provides a minimal, message-passing transport that has no inherent congestion control mechanisms. Because congestion control is critical to the stable operation of the Internet, applications and upper-layer protocols that choose to use UDP as an Internet transport must employ mechanisms to prevent congestion collapse and establish some degree of fairness with concurrent traffic. This document provides guidelines on the use of UDP for the designers of such applications and upper-layer protocols. Congestion control guidelines are a primary focus, but the document also provides guidance on other topics, including message sizes, reliability, checksums and middlebox traversal.
</t>
</abstract>
</front>
<middle>
<section title="Introduction" anchor="intro">
<t>
The User Datagram Protocol (UDP) <xref target="RFC0768"/> provides a minimal, unreliable, best-effort, message-passing transport to applications and upper-layer protocols (both simply called "applications" in the remainder of this document). Compared to other transport protocols, UDP and its UDP-Lite variant <xref target="RFC3828"/> are unique in that they do not establish end-to-end connections between communicating end systems. UDP communication consequently does not incur connection establishment and teardown overheads and there is no associated end system state. Because of these characteristics, UDP can offer a very efficient communication transport to some applications.
</t>
<t>
A second unique characteristic of UDP is that it provides no inherent congestion control mechanisms. On many platforms, applications can send UDP messages at the line rate of the link interface, which is often much greater than the available path capacity, and doing so contributes to congestion along the path. <xref target="RFC2914"/> describes the best current practice for congestion control in the Internet. It identifies two major reasons why congestion control mechanisms are critical for the stable operation of the Internet:
<list style="numbers">
<t>
The prevention of congestion collapse, i.e., a state where an increase in network load results in a decrease in useful work done by the network.
</t>
<t>
The establishment of a degree of fairness, i.e., allowing multiple flows to share the capacity of a path reasonably equitably.
</t>
</list>
</t>
<t>
Because UDP itself provides no congestion control mechanisms, it is up to the applications that use UDP for Internet communication to employ suitable mechanisms to prevent congestion collapse and establish a degree of fairness. <xref target="RFC2309"/> discusses the dangers of congestion-unresponsive flows and states that "all UDP-based streaming applications should incorporate effective congestion avoidance mechanisms." This is an important requirement, even for applications that do not use UDP for streaming. For example, an application that generates five 1500-byte UDP messages in one second can already exceed the capacity of a 56 Kb/s path. For applications that can operate at higher, potentially unbounded data rates, congestion control becomes vital to prevent congestion collapse and establish some degree of fairness. <xref target="udpguide"/> describes a number of simple guidelines for the designers of such applications.
</t>
<t>
A UDP message is carried in a single IP packet and is hence limited to a maximum payload of 65,487 bytes. The transmission of large IP packets usually requires IP fragmentation, which decreases communication reliability and efficiency and should be avoided. One reason for this decrease in reliability is that many NATs and firewalls do not forward IP fragments; other reasons are documented in <xref target="RFC4963"/>. Some of the guidelines in <xref target="udpguide"/> describe how applications should determine appropriate message sizes.
</t>
<t>
This document provides guidelines to designers of applications that use UDP for unicast transmission. A special class of applications uses UDP for IP multicast transmissions. Congestion control, flow control or reliability for multicast transmissions is more difficult to establish than for unicast transmissions, because a single sender may transmit to multiple receivers across potentially very heterogeneous paths at the same time. Designing multicast applications requires expertise that goes beyond the simple guidelines given in this document. The IETF has defined a reliable multicast framework <xref target="RFC3048"/> and several building blocks to aid the designers of multicast applications, such as <xref target="RFC3738"/> or <xref target="RFC4654"/>.
</t>
</section>
<section title="Terminology" anchor="term">
<t>
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 BCP 14, RFC 2119 <xref target="RFC2119"/>.
</t>
</section>
<section title="UDP Usage Guidelines" anchor="udpguide">
<t>
Internet paths can have widely varying characteristics, including transmission delays, available bandwidths, congestion levels, reordering probabilities, supported message sizes or loss rates. Furthermore, the same Internet path can have very different conditions over time. Consequently, applications that may be used on the Internet MUST NOT make assumptions about specific path characteristics. They MUST instead use mechanisms that let them operate safely under very different path conditions. Typically, this requires conservatively probing the Internet path to establish a transmission behavior that it can sustain and that is reasonably fair to other traffic sharing the path.
</t>
<t>
These mechanisms are difficult to implement correctly. For most applications, the use of one of the existing IETF transport protocols is the simplest method of acquiring the required mechanisms. Consequently, the RECOMMENDED alternative to the UDP usage described in the remainder of this section is the use of an IETF transport protocol such as TCP <xref target="RFC0793"/>, SCTP <xref target="RFC2960"/> or DCCP <xref target="RFC4340"/> with its different congestion control types <xref target="RFC4341"/><xref target="RFC4342"/><xref target="I-D.floyd-dccp-ccid4"/>.
</t>
<t>
If used correctly, these more fully-featured transport protocols are not as "heavyweight" as often claimed. For example, TCP's "Nagle" algorithm <xref target="RFC0896"/> can be disabled, improving communication latency at the expense of more frequent - but still congestion-controlled - packet transmissions. Another example is the TCP SYN Cookie mechanism <xref target="RFC4987"/>, which is available on many platforms. TCP with SYN Cookies does not require a server to maintain per-connection state until the connection is established. TCP also requires the end that closes a connection to maintain the TIME-WAIT state that prevents delayed segments from one connection instance to interfere with a later one. Applications that are aware of and designed for this behavior can shift maintenance of the TIME-WAIT state to conserve resources by controlling which end closes a TCP connection <xref target="FABER"/>. Finally, TCP's built-in capacity-probing and awareness of the maximum transmission unit supported by the path (PMTU) results in efficient data transmission that quickly compensates for the initial connection setup delay, for transfers that exchange more than a few messages.
</t>
<section title="Congestion Control Guidelines" anchor="ccguide">
<t>
If an application or upper-layer protocol chooses not to use a congestion-controlled transport protocol, it SHOULD control the rate at which it sends UDP messages to a destination host, in order to fulfill the requirements of <xref target="RFC2914"/>. It is important to stress that an application SHOULD perform congestion control over all UDP traffic it sends to a destination, independently from how it generates this traffic. For example, an application that forks multiple worker processes or otherwise uses multiple sockets to generate UDP messages SHOULD perform congestion control over the aggregate traffic.
</t>
<t>
The remainder of this section discusses several approaches for this purpose. Not all approaches discussed below are appropriate for all UDP-transmitting applications. <xref target="btguide"/> discusses congestion control options for applications that perform bulk transfers over UDP. Such applications can employ schemes that sample the path over several subsequent RTTs during which data is exchanged, in order to determine a sending rate that the path at its current load can support. Other applications only exchange a few UDP messages with a destination. <xref target="ldrguide"/> discusses congestion control options for such "low data-volume" applications. Because they typically do not transmit enough data to iteratively sample the path to determine a safe sending rate, they need to employ different kinds of congestion control mechanisms.
</t>
<t>
It is important to note that congestion control should not be viewed as an add-on to a finished application. Many of the mechanisms discussed in the guidelines below require application support to operate correctly. Application designers need to consider congestion control throughout the design of their application, similar to how they consider security aspects throughout the design process.
</t>
<t>
Finally, in the past, the IETF has investigated integrated congestion control mechanisms that act on the traffic aggregate between two hosts, i.e., across all communication sessions active at a given time independent of specific transport protocols, such as the Congestion Manager <xref target="RFC3124"/>. Such mechanisms have failed to see deployment, but would otherwise also fulfill the congestion control requirements in <xref target="RFC2914"/>, because they provide congestion control for UDP sessions.
</t>
<section title="Bulk Transfer Applications" anchor="btguide">
<t>
Applications that perform bulk transmission of data to a peer over UDP, i.e., applications that exchange more than a small number of messages per RTT, SHOULD implement TCP-Friendly Rate Control (TFRC) <xref target="RFC3448"/>, window-based, TCP-like congestion control, or otherwise ensure that the application complies with the congestion control principles.
</t>
<t>
TFRC has been designed to provide both congestion control and fairness in a way that is compatible with the IETF's other transport protocols. TFRC is currently being updated <xref target="I-D.ietf-dccp-rfc3448bis"/>, and application designers SHOULD always evaluate whether the latest published specification fits their needs. If an application implements TFRC, it need not follow the remaining guidelines in <xref target="ccguide"/>, because TFRC already addresses them, but SHOULD still follow the remaining guidelines in the subsequent subsections of <xref target="udpguide"/>.
</t>
<t>
Bulk transfer applications that choose not to implement TFRC or TCP-like windowing SHOULD implement a congestion control scheme that results in bandwidth use that competes fairly with TCP within an order of magnitude. <xref target="RFC3551"/> suggests that applications SHOULD monitor the packet loss rate to ensure that it is within acceptable parameters. Packet loss is considered acceptable if a TCP flow across the same network path under the same network conditions would achieve an average throughput, measured on a reasonable timescale, that is not less than that of the UDP flow. The comparison to TCP cannot be specified exactly, but is intended as an "order-of-magnitude" comparison in timescale and throughput.
</t>
<t>
Finally, some bulk transfer applications chose not to implement any congestion control mechanism and instead rely on transmitting across reserved path capacity. This might be an acceptable choice for a subset of restricted networking environments, but is by no means a safe practice for operation in the Internet. When the UDP traffic of such applications leaks out on unprovisioned Internet paths, if can significantly degrade the performance of other traffic sharing the path and even result in congestion collapse. Applications that support an uncontrolled or unadaptive transmission behavior SHOULD NOT do so by default and SHOULD instead require users to explicitly enable this mode of operation.
</t>
</section>
<section title="Low Data-Volume Applications" anchor="ldrguide">
<t>
When applications that exchange only a small number of messages with a destination at any time implement TFRC or one of the other congestion control schemes in <xref target="btguide"/>, the network sees little benefit, because those mechanisms perform congestion control in a way that is only effective for longer transmissions.
</t>
<t>
Applications that exchange only a small number of messages with a destination at any time SHOULD still control their transmission behavior by not sending more than one UDP message per round-trip time (RTT) to a destination. Similar to the recommendation in <xref target="RFC1536"/>, an application SHOULD maintain an estimate of the RTT for any destination with which it communicates. Applications SHOULD implement the algorithm specified in <xref target="RFC2988"/> to compute a smoothed RTT (SRTT) estimate. A lost response from the peer SHOULD be treated as a very large RTT sample, instead of being ignored, in order to cause a sufficiently large (exponential) back-off. When implementing this scheme, applications need to choose a sensible initial value for the RTT. This value SHOULD generally be as conservative as possible for the given application. TCP uses an initial value of 3 seconds <xref target="RFC2988"/>, which is also RECOMMENDED as an initial value for UDP applications. SIP <xref target="RFC3261"/> and GIST <xref target="I-D.ietf-nsis-ntlp"/> use an initial value of 500 ms, and initial timeouts that are shorter than this are likely problematic in many cases. It is also important to note that the initial timeout is not the maximum possible timeout - the RECOMMENDED algorithm in <xref target="RFC2988"/> yields timeout values after a series of losses that are much longer than the initial value.
</t>
<t>
Some applications cannot maintain a reliable RTT estimate for a destination. The first case is applications that exchange too few messages with a peer to establish a statistically accurate RTT estimate. Such applications MAY use a fixed transmission interval that is exponentially backed-off during loss. TCP uses an initial value of 3 seconds <xref target="RFC2988"/>, which is also RECOMMENDED as an initial value for UDP applications. SIP <xref target="RFC3261"/> and GIST <xref target="I-D.ietf-nsis-ntlp"/> use an interval of 500 ms, and shorter values are likely problematic in many cases. As in the previous case, note that the initial timeout is not the maximum possible timeout.
</t>
<t>
A second class of applications cannot maintain an RTT estimate for a destination, because the destination does not send return traffic. Such applications SHOULD NOT send more than one UDP message every 3 seconds, and SHOULD use an even less aggressive rate when possible. The 3-second interval was chosen based on TCP's retransmission timeout when the RTT is unknown <xref target="RFC2988"/>, and shorter values are likely problematic in many cases. Note that the initial timeout interval must be more conservative than in the two previous cases, because the lack of return traffic prevents the detection of packet loss, i.e., congestion events, and the application therefore cannot perform exponential back-off to reduce load.
</t>
<t>
Applications that communicate bidirectionally SHOULD employ congestion control for both directions of the communication. For example, for a client-server, request-response-style application, clients SHOULD congestion control their request transmission to a server, and the server SHOULD congestion-control its responses to the clients. Congestion in the forward and reverse direction is uncorrelated and an application SHOULD independently detect and respond to congestion along both directions.
</t>
</section>
</section>
<section title="Message Size Guidelines" anchor="msguide">
<t>
Because IP fragmentation lowers the efficiency and reliability of Internet communication <xref target="RFC4963"/>, an application SHOULD NOT send UDP messages that result in IP packets that exceed the MTU of the path to the destination. Consequently, an application SHOULD either use the path MTU information provided by the IP layer or implement path MTU discovery itself <xref target="RFC1191"/><xref target="RFC1981"/><xref target="RFC4821"/> to determine whether the path to a destination will support its desired message size without fragmentation.
</t>
<t>
Applications that choose to not adapt their transmit message size SHOULD NOT send UDP messages that exceed the minimum PMTU. The minimum PMTU depends on the IP version used for transmission, and is the lesser of 576 bytes and the first-hop MTU for IPv4 <xref target="RFC1122"/> and 1280 bytes for IPv6 <xref target="RFC2460"/>. To determine an appropriate UDP payload size, applications must subtract IP header and option lengths as well as the length of the UDP header from the PMTU size. Transmission of minimum-sized messages is inefficient over paths that support a larger PMTU, which is a second reason to implement PMTU discovery.
</t>
<t>
Applications that do not send messages that exceed the minimum PMTU of IPv4 or IPv6 need not implement any of the above mechanisms. Note that the presence of tunnels can cause fragmentation even when applications send messages that do not exceed the minimum PMTU, so implementing PMTU discovery will still be beneficial in some cases.
</t>
</section>
<section title="Reliability Guidelines" anchor="reguide">
<t>
Application designers are generally aware that UDP does not provide any reliability. Often, this is a main reason to consider UDP as a transport. Applications that do require reliable message delivery SHOULD implement an appropriate mechanism themselves.
</t>
<t>
UDP also does not protect against message duplication, i.e., an application may receive multiple copies of the same message. Application designers SHOULD verify that their application handles message duplication gracefully, and may consequently need to implement mechanisms to detect duplicates. Even if message reception triggers idempotent operations, applications may want to suppress duplicate messages to reduce load.
</t>
<t>
Finally, the Internet can significantly delay some packets with respect to others, e.g., due to routing transients, intermittent connectivity, or mobility. This can cause message reordering, where UDP messages arrive at the receiver in an order different from the transmission order. Applications that require ordered delivery SHOULD reestablish message ordering themselves. Furthermore, it is important to note that delay spikes can be very large. This can cause reordered packets to arrive many seconds after they were sent. The Internet protocol suite defines the Maximum Segment Lifetime (MSL) as 2 minutes <xref target="RFC0793"/>. This is the maximum delay a packet should experience. Applications SHOULD be robust to the reception of delayed or duplicate packets that are received within this two minute interval.
</t>
</section>
<section title="Checksum Guidelines" anchor="chkguide">
<t>
The UDP header includes an optional, 16-bit ones' complement checksum that provides an integrity check. The UDP checksum provides assurance that the payload was not corrupted in transit. It also verifies that the packet was delivered to the intended destination, because it covers the IP addresses, port numbers and protocol number, and it verifies that the packet is not truncated or padded, because it covers the size field. It therefore protects an application against receiving corrupted payload data in place of, or in addition to, the data that was sent.
</t>
<t>
Applications SHOULD enable UDP checksums, although <xref target="RFC0793"/> permits the option to disable their use. Applications that choose to disable UDP checksums when transmitting over IPv4 therefore MUST NOT make assumptions regarding the correctness of received data and MUST behave correctly when a message is received that was originally sent to a different destination or is otherwise corrupted. The use of the UDP checksum is MANDATORY when applications transmit UDP over IPv6 <xref target="RFC2460"/> and applications consequently MUST NOT disable their use. (The IPv6 header does not have a separate checksum field to protect the IP addressing information.)
</t>
<t>
The UDP checksum provides relatively weak protection from a coding point of view <xref target="RFC3819"/> and, where data integrity is important, application developers SHOULD provide additional checks, e.g., through a Cyclic Redundancy Check (CRC) included with the data to verify the integrity of an entire object/file sent over UDP service.
</t>
<section title="UDP-Lite" anchor="udplite">
<t>
A special class of applications can derive benefit from having partially damaged payloads delivered, rather than discarded, when using paths that include error-prone links. Such applications can tolerate payload corruption and MAY choose to use the Lightweight User Datagram Protocol (UDP-Lite) <xref target="RFC3828"/> variant of UDP instead of basic UDP. Applications that choose to use UDP-Lite instead of UDP MUST still follow the congestion control and other guidelines described for use with UDP in <xref target="ccguide"/>.
</t>
<t>
UDP-Lite changes the semantics of the UDP "payload length" field to that of a "checksum coverage length" field. Otherwise, UDP-Lite is semantically identical to UDP. The interface of UDP-Lite differs from that of UDP by the addition of a single (socket) option that communicates a checksum coverage length value: at the sender, this specifies the intended checksum coverage, with the remaining unprotected part of the payload called the "error insensitive part". If required, an application may dynamically modify this length value, e.g., to offer greater protection to some messages. UDP-Lite always verifies that a packet was delivered to the intended destination, i.e., always verifies the header fields. Errors in the insensitive part will not cause a UDP message to be discarded by the destination. Applications using UDP-Lite therefore MUST NOT make assumptions regarding the correctness of the data received in the insensitive part of the UDP-Lite payload.
</t>
<t>
The sending application SHOULD select the minimum checksum coverage to include all sensitive protocol headers. For example, applications that use the Real-Time Protocol (RTP) <xref target="RFC3550"/> will likely want to protect the RTP header against corruption. Applications, where appropriate, MUST also introduce their own appropriate validity checks for protocol information carried in the insensitive part of the UDP-Lite payload (e.g., internal CRCs).
</t>
<t>
The receiver MUST set a minimum coverage threshold for incoming packets
that is not smaller than the smallest coverage used by the sender. This may
be a fixed value, or may be negotiated by an application. UDP-Lite does not
provide mechanisms to negotiate the checksum coverage between the sender and
receiver.
</t>
<t>
Applications may still experience packet loss, rather than corruption, when
using UDP-Lite. The enhancements offered by UDP-Lite rely upon a link being
able to intercept the UDP-Lite header to correctly identify the
partial-coverage required. When tunnels and/or encryption are used, this can
result in UDP-Lite messages being treated the same as UDP messages, i.e., result in packet loss. Use of IP fragmentation can also prevent special treatment for
UDP-Lite messages, and is another reason why applications SHOULD avoid IP fragmentation <xref target="msguide"/>.
</t>
</section>
</section>
<section title="Middlebox Traversal Guidelines" anchor="natguide">
<t>
Network address translators (NATs) and firewalls are examples of intermediary devices ("middleboxes") that can exist along an end-to-end path. A middlebox typically performs a function that requires it to maintain per-flow state. For connection-oriented protocols, such as TCP, middleboxes snoop and parse the connection-management traffic and create and destroy per-flow state accordingly. For a connectionless protocol such as UDP, this approach is not possible. Consequently, middleboxes may create per-flow state when they see a packet that indicates a new flow, and destroy the state after some period of time during which no packets belonging to the same flow have arrived.
</t>
<t>
Depending on the specific function that the middlebox performs, this behavior can introduce a time-dependency that restricts the kinds of UDP traffic exchanges that will be successful across the middlebox. For example, NATs and firewalls typically define the partial path on one side of them to be interior to the domain they serve, whereas the partial path on their other side is defined to be exterior to that domain. Per-flow state is typically created when the first packet crosses from the interior to the exterior, and while the state is present, NATs and firewalls will forward return traffic. Return traffic arriving after the per-flow state has timed out is dropped, as is other traffic arriving from the exterior.
</t>
<t>
Many applications that use UDP for communication operate across middleboxes without needing to employ additional mechanisms. One example is the DNS, which has a strict request-response communication pattern that typically completes within seconds.
</t>
<t>
Other applications may experience communication failures when middleboxes destroy the per-flow state associated with an application session during periods when the application does not exchange any UDP traffic. Applications SHOULD be able to gracefully handle such communication failures and implement mechanisms to re-establish their UDP sessions.
</t>
<t>
For some applications, such as media transmissions, this re-synchronization is highly undesirable, because it can cause user-perceivable playback artifacts. Such specialized applications MAY send periodic keep-alive messages to attempt to refresh middlebox state. It is important to note that keep-alive messages are NOT RECOMMENDED for general use - they are unnecessary for many applications and can consume significant amounts of system and network resources.
</t>
<t>
An application that needs to employ keep-alives to deliver useful service in the presence of middleboxes SHOULD NOT transmit them more frequently than once every 15 seconds and SHOULD use longer intervals when possible. No common timeout has been specified for per-flow UDP state for arbitrary middleboxes. For NATs, <xref target="RFC4787"/> requires a state timeout of 2 minutes or longer. However, empirical evidence suggests that a significant fraction of the deployed middleboxes unfortunately uses shorter timeouts. The timeout of 15 seconds originates with the Interactive Connectivity Establishment (ICE) protocol <xref target="I-D.ietf-mmusic-ice"/>. Applications that operate in more controlled network environments SHOULD investigate whether the environment they operate in allows them to use longer intervals, or whether it offers mechanisms to explicitly control middlebox state timeout durations, for example, using MIDCOM <xref target="RFC3303"/>, NSIS <xref target="I-D.ietf-nsis-nslp-natfw"/>, STUN <xref target="I-D.wing-behave-nat-control-stun-usage"/> or UPnP <xref target="UPNP"/>.
</t>
<t>
It is important to note that sending keep-alives is not a substitute for implementing robust connection handling. Like all UDP messages, keep-alives can be delayed or dropped, causing middlebox state to time out. In addition, the congestion control guidelines in <xref target="ccguide"/> cover all UDP transmissions by an application, including the transmission of middlebox keep-alives. Congestion control may thus lead to delays or temporary suspension of keep-alive transmission.
</t>
</section>
<section title="Programming Guidelines" anchor="progguide">
<t>
The de facto standard application programming interface (API) for TCP/IP applications is the "sockets" interface <xref target="POSIX"/>. Although this API was developed for UNIX in the early 1980s, a wide variety of non-UNIX operating systems also implements it. The sockets API supports both IPv4 and IPv6 <xref target="RFC3493"/>. The UDP sockets API differs from that for TCP in several key ways. Because application programmers are typically more familiar with the TCP sockets API, the remainder of this section discusses these differences. <xref target="STEVENS"/> provides usage examples of the UDP sockets API.
</t>
<t>
UDP messages may be directly sent and received, without any connection setup. Using the sockets API, applications can receive packets from more than one IP source address on a single UDP socket. Some servers use this to exchange data with more than one remote host through a single UDP socket at the same time. When applications need to ensure that they receive packets from a particular source address, they MUST implement corresponding checks at the application layer or explicitly request that the operating system filter the received packets.
</t>
<t>
Many operating systems also allow a UDP socket to be connected, i.e., allow to bind a UDP socket to a specific pair of addresses and ports. This is similar to the corresponding TCP sockets API functionality. However, for UDP, this is only a local operation that serves to simplify the local send/receive functions and to filter the traffic for the specified addresses and ports. Binding a UDP socket does not establish a connection - UDP does not notify the remote end when a local UDP socket is bound.
</t>
<t>
UDP provides no flow-control. This is another reason why UDP-based applications need to be robust in the presence of packet loss. This loss can also occur within the sending host, when an application sends data faster than the line rate of the outbound network interface. It can also occur on the destination, where receive calls fail to return data when the application issues them too frequently (i.e., when no new data has arrived) or not frequently enough (i.e., such that the receive buffer overflows). Robust flow control mechanisms are difficult to implement, which is why applications that need this functionality SHOULD consider using a full-featured transport protocol.
</t>
<t>
When an application closes a TCP, SCTP or DCCP socket, the transport protocol on the receiving host is required to maintain TIME-WAIT state. This prevents delayed packets from the closed connection instance from being mistakenly associated with a later connection instance that happens to reuse the same IP address and port pairs. The UDP protocol does not implement such a mechanism. Therefore, UDP-based applications need to robust to this case. One application may close a socket or terminate, followed in time by another application receiving on the same port. This later application may then receive packets intended for the first application that were delayed in the network.
</t>
</section>
</section>
<section title="Security Considerations" anchor="seccons">
<t>
UDP does not provide communications security. Applications that need to protect their communications against eavesdropping, tampering, or message forgery SHOULD employ end-to-end security services provided by other IETF protocols.
</t>
<t>
One option of securing UDP communications is with IPsec <xref target="RFC4301"/>, which provides authentication <xref target="RFC4302"/> and encryption <xref target="RFC4303"/> for flows of IP packets. Applications use the Internet Key Exchange (IKE) <xref target="RFC4306"/> to configure IPsec for their sessions. Depending on how IPsec is configured for a flow, it can authenticate or encrypt the UDP headers as well as UDP payloads. In order to be able to use IPsec, an application must execute on an operating system that implements the IPsec protocol suite.
</t>
<t>
Not all operating systems support IPsec. A second option of securing UDP communications is through Datagram Transport Layer Security (DTLS) <xref target="RFC4347"/>. DTLS provides communication privacy by encrypting UDP payloads. It does not protect the UDP headers. Applications can implement DTLS without relying on support from the operating system.
</t>
<t>
Many other options of authenticating or encrypting UDP payloads exist, including other IETF standards, such as S/MIME <xref target="RFC3851"/> or PGP <xref target="RFC2440"/>, as well as many non-IETF protocols. Like congestion control mechanisms, security mechanisms are difficult to design and implement correctly. It is hence RECOMMENDED that applications employ well-known standard security mechanisms such as IPsec or DTLS, rather than inventing their own.
</t>
<t>
In terms of congestion control, <xref target="RFC2309"/> and <xref target="RFC2914"/> discuss the dangers of congestion-unresponsive flows to the Internet. This document provides guidelines to designers of UDP-based applications to congestion-control their transmissions. As such, it does not raise any additional security concerns.
</t>
</section>
<section title="Summary">
<t>
This section summarizes the guidelines made in <xref target="udpguide"/> and <xref target="seccons"/> in a tabular format in <xref target="sumtable"/> for easy referencing.
</t>
<texttable anchor="sumtable" title="Summary of recommendations.">
<ttcol>Section</ttcol>
<ttcol>Recommendation</ttcol>
<c><xref target="udpguide" format="counter"/></c>
<c>MUST accommodate wide range of Internet path conditions</c>
<c/>
<c>SHOULD use a full-featured transport (TCP, SCTP, DCCP)</c>
<c> </c><c> </c>
<c><xref target="ccguide" format="counter"/></c>
<c>SHOULD control rate of transmission</c>
<c/>
<c>SHOULD perform congestion control over all traffic</c>
<c> </c><c> </c>
<c><xref target="btguide" format="counter"/></c>
<c>for bulk transfers,</c>
<c/>
<c>SHOULD consider implementing TFRC</c>
<c/>
<c>else, SHOULD otherwise use bandwidth similar to TCP</c>
<c> </c><c> </c>
<c><xref target="ldrguide" format="counter"/></c>
<c>for non-bulk transfers,</c>
<c/>
<c>SHOULD measure RTT and transmit 1 message/RTT</c>
<c/>
<c>else, SHOULD send at most 1 message every 3 seconds</c>
<c> </c><c> </c>
<c><xref target="msguide" format="counter"/></c>
<c>SHOULD NOT send messages that exceed the PMTU, i.e.,</c>
<c/>
<c>SHOULD discover PMTU or send messages < minimum PMTU</c>
<c> </c><c> </c>
<c><xref target="reguide" format="counter"/></c>
<c>SHOULD handle message loss, duplication, reordering</c>
<c> </c><c> </c>
<c><xref target="chkguide" format="counter"/></c>
<c>SHOULD enable UDP checksum</c>
<c><xref target="udplite" format="counter"/></c>
<c>else, MAY use UDP-Lite with suitable checksum coverage</c>
<c> </c><c> </c>
<c><xref target="natguide" format="counter"/></c>
<c>SHOULD NOT always send middlebox keep-alives</c>
<c/>
<c>MAY use keep-alives when needed (min. interval 15 sec)</c>
<c> </c><c> </c>
<c><xref target="progguide" format="counter"/></c>
<c>MUST check IP source address</c>
<c> </c><c> </c>
<c><xref target="seccons" format="counter"/></c>
<c>SHOULD use standard IETF security protocols when needed</c>
</texttable>
</section>
<section title="IANA Considerations" anchor="ianacons">
<t>
This document raises no IANA considerations.
</t>
</section>
<section title="Acknowledgments" anchor="ack">
<t>
Thanks to Paul Aitken, Mark Allman, Wesley Eddy, Sally Floyd, Philip Matthews, Joerg Ott, Colin Perkins, Pasi Sarolahti, Joe Touch and Magnus Westerlund for their comments on this document.
</t>
<t>
The middlebox traversal guidelines in <xref target="natguide"/> incorporate ideas from Section 5 of <xref target="I-D.ford-behave-app"/> by Bryan Ford, Pyda Srisuresh and Dan Kegel.
</t>
</section>
</middle>
<!-- REFERENCE TEMPLATE
<reference anchor="reference.XXX">
<front>
<title>XXX</title>
<author initials="X." surname="XXX" fullname="XXX">
<organization abbrev="XXX">XXX</organization>
<address>
<postal>
<street>XXX</street>
<city>XXX</city>
<region>XXX</region>
<code>XXX</code>
<country>XXX</country>
</postal>
<phone>XXX</phone>
<facsimile>XXX</facsimile>
<email>XXX</email>
<uri>XXX</uri>
</address>
</author>
<date month="XXX" year="XXX"/>
</front>
<seriesInfo name="XXX" value="XXX"/>
<format type="XXX" target="XXX"/>
</reference>
-->
<back>
<references title='Normative References'>
<?rfc include="reference.RFC.0768" ?>
<?rfc include="reference.RFC.2119" ?>
<?rfc include="reference.RFC.2914" ?>
<?rfc include="reference.RFC.0793" ?>
<?rfc include="reference.RFC.2960" ?>
<?rfc include="reference.RFC.4340" ?>
<?rfc include="reference.RFC.3448" ?>
<?rfc include="reference.RFC.1191" ?>
<?rfc include="reference.RFC.1981" ?>
<?rfc include="reference.RFC.2988" ?>
<?rfc include="reference.RFC.3819" ?>
<?rfc include="reference.RFC.3828" ?>
<?rfc include="reference.RFC.4821" ?>
<?rfc include="reference.RFC.4301" ?>
<?rfc include="reference.RFC.4306" ?>
<?rfc include="reference.RFC.4347" ?>
<?rfc include="reference.RFC.4302" ?>
<?rfc include="reference.RFC.4303" ?>
</references>
<references title='Informative References'>
<?rfc include="reference.RFC.0896" ?>
<?rfc include="reference.RFC.1122" ?>
<?rfc include="reference.RFC.1536" ?>
<?rfc include="reference.RFC.2309" ?>
<?rfc include="reference.RFC.2440" ?>
<?rfc include="reference.RFC.2460" ?>
<?rfc include="reference.RFC.3048" ?>
<?rfc include="reference.RFC.3124" ?>
<?rfc include="reference.RFC.3261" ?>
<?rfc include="reference.RFC.3303" ?>
<?rfc include="reference.RFC.3493" ?>
<?rfc include="reference.RFC.3550" ?>
<?rfc include="reference.RFC.3551" ?>
<?rfc include="reference.RFC.3738" ?>
<?rfc include="reference.RFC.4654" ?>
<?rfc include="reference.RFC.4341" ?>
<?rfc include="reference.RFC.3851" ?>
<?rfc include="reference.RFC.4342" ?>
<?rfc include="reference.RFC.4787" ?>
<?rfc include="reference.RFC.4963" ?>
<?rfc include="reference.RFC.4987" ?>
<?rfc include="reference.I-D.ietf-nsis-nslp-natfw" ?>
<?rfc include="reference.I-D.floyd-dccp-ccid4" ?>
<?rfc include="reference.I-D.wing-behave-nat-control-stun-usage" ?>
<?rfc include="reference.I-D.ietf-mmusic-ice" ?>
<?rfc include="reference.I-D.ietf-nsis-ntlp" ?>
<?rfc include="reference.I-D.ietf-dccp-rfc3448bis" ?>
<?rfc include="reference.I-D.ford-behave-app" ?>
<reference anchor="UPNP">
<front>
<title>Internet Gateway Device (IGD) Standardized Device Control Protocol V 1.0</title>
<author surname="UPnP Forum">
<organization/>
</author>
<date month="November" year="2001"/>
</front>
</reference>
<reference anchor="POSIX">
<front>
<title>Standard for Information Technology - Portable Operating System Interface (POSIX)</title>
<author initials="" surname="IEEE Std. 1003.1-2001">
<organization />
</author>
<date month="December" year="2001"/>
</front>
<seriesInfo name="Open Group Technical Standard: Base Specifications" value="Issue 6, ISO/IEC 9945:2002"/>
</reference>
<reference anchor="STEVENS">
<front>
<title>UNIX Network Programming, The sockets Networking API</title>
<author initials="W. R." surname="Stevens">
<organization />
</author>
<author initials="B." surname="Fenner">
<organization />
</author>
<author initials="A. M." surname="Rudoff">
<organization />
</author>
<date year="2004"/>
</front>
<seriesInfo name="" value="Addison-Wesley"/>
</reference>
<reference anchor="FABER">
<front>
<title>The TIME-WAIT State in TCP and Its Effect on Busy Servers</title>
<author initials="T." surname="Faber">
<organization />
</author>
<author initials="J." surname="Touch">
<organization />
</author>
<author initials="W." surname="Yue">
<organization />
</author>
<date month="March" year="1999"/>
</front>
<seriesInfo name="Proc." value="IEEE Infocom"/>
</reference>
</references>
</back>
</rfc>
| PAFTECH AB 2003-2026 | 2026-04-23 20:48:52 |