| Network Working Group | G. Fairhurst, Ed. |
| Internet-Draft | University of Aberdeen |
| Intended status: Informational | B. Trammell, Ed. |
| Expires: August 10, 2015 | M. Kuehlewind, Ed. |
| ETH Zurich | |
| February 06, 2015 |
Services provided by IETF transport protocols and congestion control mechanisms
draft-ietf-taps-transports-02
This document describes services provided by existing IETF protocols and congestion control mechanisms. It is designed to help application and network stack programmers and to inform the work of the IETF TAPS Working Group.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at http://datatracker.ietf.org/drafts/current/.
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."
This Internet-Draft will expire on August 10, 2015.
Copyright (c) 2015 IETF Trust and the persons identified as the document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
Most Internet applications make use of the Transport Services provided by TCP (a reliable, in-order stream protocol) or UDP (an unreliable datagram protocol). We use the term “Transport Service” to mean the end-to-end service provided to an application by the transport layer. That service can only be provided correctly if information about the intended usage is supplied from the application. The application may determine this information at design time, compile time, or run time, and may include guidance on whether a feature is required, a preference by the application, or something in between. Examples of features of Transport Services are reliable delivery, ordered delivery, content privacy to in-path devices, integrity protection, and minimal latency.
The IETF has defined a wide variety of transport protocols beyond TCP and UDP, including TCP, SCTP, DCCP, MP-TCP, and UDP-Lite. Transport services may be provided directly by these transport protocols, or layered on top of them using protocols such as WebSockets (which runs over TCP) or RTP (over TCP or UDP). Services built on top of UDP or UDP-Lite typically also need to specify additional mechanisms, including a congestion control mechanism (such as a windowed congestion control, TFRC or LEDBAT congestion control mechanism). This extends the set of available Transport Services beyond those provided to applications by TCP and UDP.
Transport protocols can also be differentiated by the features of the services they provide: for instance, SCTP offers a message-based service that does not suffer head-of-line blocking when used with multiple stream, because it can accept blocks of data out of order, UDP-Lite provides partial integrity protection, and LEDBAT can provide low-priority “scavenger” communication.
The following terms are defined throughout this document, and in subsequent documents produced by TAPS describing the composition and decomposition of transport services.
[NOTE: The terminology below was presented at the TAPS WG meeting in Honolulu. While the factoring of the terminology seems uncontroversial, there may be some entities which still require names (e.g. information about the interface between the transport and lower layers which could lead to the availablity or unavailibility of certain transport protocol features). Comments are welcome via the TAPS mailing list.]
This section provides a list of known IETF transport protocol and transport protocol frameworks.
[EDITOR’S NOTE: Contributions to the subsections below are welcome]
TCP is an IETF standards track transport protocol. [RFC0793] introduces TCP as follows: “The Transmission Control Protocol (TCP) is intended for use as a highly reliable host-to-host protocol between hosts in packet-switched computer communication networks, and in interconnected systems of such networks.” Since its introduction, TCP has become the default connection-oriented, stream-based transport protocol in the Internet. It is widely implemented by endpoints and widely used by common applications.
TCP is a connection-oriented protocol, providing a three way handshake to allow a client and server to set up a connection, and mechanisms for orderly completion and immediate teardown of a connection. TCP is defined by a family of RFCs [RFC4614].
TCP provides multiplexing to multiple sockets on each host using port numbers. An active TCP session is identified by its four-tuple of local and remote IP addresses and local port and remote port numbers. The destination port during connection setup has a different role as it is often used to indicate the requested service.
TCP partitions a continuous stream of bytes into segments, sized to fit in IP packets. ICMP-based PathMTU discovery [RFC1191][RFC1981] as well as Packetization Layer Path MTU Discovery (PMTUD) [RFC4821] are supported.
Each byte in the stream is identified by a sequence number. The sequence number is used to order segments on receipt, to identify segments in acknowledgments, and to detect unacknowledged segments for retransmission. This is the basis of TCP’s reliable, ordered delivery of data in a stream. TCP Selective Acknowledgment [RFC2018] extends this mechanism by making it possible to identify missing segments more precisely, reducing spurious retransmission.
Receiver flow control is provided by a sliding window: limiting the amount of unacknowledged data that can be outstanding at a given time. The window scale option [RFC7323] allows a receiver to use windows greater than 64KB.
All TCP senders provide Congestion Control: This uses a separate window, where each time congestion is detected, this congestion window is reduced. A receiver detects congestion using one of three mechanisms: A retransmission timer, detection of loss (interpreted as a congestion signal), or Explicit Congestion Notification (ECN) [RFC3168] to provide early signaling (see [I-D.ietf-aqm-ecn-benefits])
A TCP protocol instance can be extended [RFC4614] and tuned. Some features are sender-side only, requiring no negotiation with the receiver; some are receiver-side only, some are explicitly negotiated during connection setup.
By default, TCP segment partitioning uses Nagle’s algorithm [RFC0896] to buffer data at the sender into large segments, potentially incurring sender-side buffering delay; this algorithm can be disabled by the sender to transmit more immediately, e.g. to enable smoother interactive sessions.
[EDITOR’S NOTE: add URGENT and PUSH flag (note [RFC6093] says SHOULD NOT use due to the range of TCP implementations that process TCP urgent indications differently.) ]
A checksum provides an Integrity Check and is mandatory across the entire packet. The TCP checksum does not support partial corruption protection as in DCCP/UDP-Lite). This check protects from misdelivery of data corrupted data, but is relatively weak, and applications that require end to end integrity of data are recommended to include a stronger integrity check of their payload data.
A TCP service is unicast.
A User/TCP Interface is defined in [RFC0793] providing six user commands: Open, Send, Receive, Close, Status. This interface does not describe configuration of TCP options or parameters beside use of the PUSH and URGENT flags.
In API implementations derived from the BSD Sockets API, TCP sockets are created using the SOCK_STREAM socket type.
The features used by a protocol instance may be set and tuned via this API.
(more on the API goes here)
The transport protocol components provided by TCP are:
[EDITOR’S NOTE: discussion of how to map this to features and TAPS: what does the higher layer need to decide? what can the transport layer decide based on global settings? what must the transport layer decide based on network characteristics?]
[EDITOR’S NOTE: a few sentences describing Multipath TCP [RFC6824] go here. Note that this adds transport-layer multihoming to the components TCP provides]
SCTP [RFC4960] is an IETF standards track transport protocol that provides a bidirectional set of logical unicast meessage streams over a connection-oriented protocol.
Compared to TCP, this protocol and API use messages, rather than a byte-stream. Each stream of messages is independently managed, therefore retransmission does not hold back data sent using other logical streams.
An SCTP Integrity Check is mandatory across the entire packet (it does not support partial corruption protection as in DCCP/UD-Lite).
The SCTP Partial Reliability Extension (SCTP-PR) is defined in [RFC3758].
SCTP supports PLPMTU discovery using padding chunks to construct path probes.
[EDITOR’S NOTE: Michael Tuexen and Karen Nielsen signed up as contributors for these sections.]
An SCTP service is unicast.
PLPMTUD is required for SCTP.
The SCTP API is described in the specifications published in the RFC series.
The transport protocol components provided by SCTP are:
[EDITOR’S NOTE: update this list.]
The User Datagram Protocol (UDP) [RFC0768] [RFC2460] is an IETF standards track transport protocol. It provides a uni-directional minimal message-passing transport that has no inherent congestion control mechanisms or other transport functions. IETF guidance on the use of UDP is provided in [RFC5405]. UDP is widely implemented by endpoints and widely used by common applications.
[EDITOR’S NOTE: Kevin Fall signed up as a contributor for this section.]
UDP is a connection-less datagram protocol, with no connection setup or feature negotiation. The protocol and API use messages, rather than a byte-stream. Each stream of messages is independently managed, therefore retransmission does not hold back data sent using other logical streams.
It provides multiplexing to multiple sockets on each host using port numbers. An active UDP session is identified by its four-tuple of local and remote IP addresses and local port and remote port numbers.
UDP maps each data segement into an IP packet, or a sequence of IP fragemnts.
UDP is connectionless. However, applications send a sequence of messages that constitute a UDP flow. Therefore mechanisms for receiver flow control, congestion control, PathMTU discovery/PLPMTUD, support for ECN, etc need to be provided by upper layer protocols [RFC5405].
PMTU discovery and PLPMTU discovery may be used by upper layer protocols built on top of UDP [RFC5405].
For IPv4 the UDP checksum is optional, but recommended for use in the general Internet [RFC5405]. [RFC2460] requires the use of this checksum for IPv6, but [RFC6935] permits this to be relaxed for specific types of application. The checksum support considerations for omitting the checksum are defined in [RFC6936].
This check protects from misdelivery of data corrupted data, but is relatively weak, and applications that require end to end integrity of data are recommended to include a stronger integrity check of their payload data.
A UDP service may support IPv4 broadcast, multicast, anycast and unicast, determined by the IP destination address.
[RFC0768] describes basic requirements for an API for UDP. Guidance on use of common APIs is provided in [RFC5405].
Many operating systems also allow a UDP socket to be connected, i.e., 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 [RFC5405].
The transport protocol components provided by UDP are:
The Lightweight User Datagram Protocol (UDP-Lite) [RFC3828] is an IETF standards track transport protocol. UDP-Lite provides a bidirectional set of logical unicast or multicast message streams over a datagram protocol. IETF guidance on the use of UDP-Lite is provided in [RFC5405].
[EDITOR’S NOTE: Gorry Fairhurst signed up as a contributor for this section.]
UDP-Lite is a connection-less datagram protocol, with no connection setup or feature negotiation. The protocol use messages, rather than a byte-stream. Each stream of messages is independently managed, therefore retransmission does not hold back data sent using other logical streams.
It provides multiplexing to multiple sockets on each host using port numbers. An active UDP-Lite session is identified by its four-tuple of local and remote IP addresses and local port and remote port numbers.
UDP-Lite fragments packets into IP packets, constrained by the maximum size of IP packet.
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. Applications using UDP-Lite therefore can not make assumptions regarding the correctness of the data received in the insensitive part of the UDP-Lite payload.
As for UDP, mechanisms for receiver flow control, congestion control, PMTU or PLPMTU discovery, support for ECN, etc need to be provided by upper layer protocols [RFC5405].
Examples of use include a class of applications that can derive benefit from having partially-damaged payloads delivered, rather than discarded. One use is to support error tolerate payload corruption when used over paths that include error-prone links, another application is when header integrity checks are required, but payload integrity is provided by some other mechanism (e.g. [RFC6936].
A UDP-Lite service may support IPv4 broadcast, multicast, anycast and unicast.
There is no current API specified in the RFC Series, but guidance on use of common APIs is provided in [RFC5405].
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”. The checksum coverage may also be made visible to the application via the UDP-Lite MIB module [RFC5097].
The transport protocol components provided by UDP-Lite are:
Datagram Congestion Control Protocol (DCCP) [RFC4340] is an IETF standards track bidirectional transport protocol that provides unicast connections of congestion-controlled unreliable messages.
[EDITOR’S NOTE: Gorry Fairhurst signed up as a contributor for this section.]
The DCCP Problem Statement describes the goals that DCCP sought to address [RFC4336]. It is suitable for applications that transfer fairly large amounts of data and that can benefit from control over the trade off between timeliness and reliability [RFC4336].
It offers low overhead, and many characteristics common to UDP, but can avoid “Re-inventing the wheel” each time a new multimedia application emerges. Specifically it includes core functions (feature negotiation, path state management, RTT calculation, PMTUD, etc): This allows applications to use a compatible method defining how they send packets and where suitable to choose common algorithms to manage their functions. Examples of suitable applications include interactive applications, streaming media or on-line games [RFC4336].
DCCP is a connection-oriented datagram protocol, providing a three way handshake to allow a client and server to set up a connection, and mechanisms for orderly completion and immediate teardown of a connection. The protocol is defined by a family of RFCs.
It provides multiplexing to multiple sockets on each host using port numbers. An active DCCP session is identified by its four-tuple of local and remote IP addresses and local port and remote port numbers. At connection setup, DCCP also exchanges the the service code [RFC5595] mechanism to allow transport instantiations to indicate the service treatment that is expected from the network.
The protocol segments data into messages, typically sized to fit in IP packets, but which may be fragemented providing they are less than the A DCCP interface MAY allow applications to request fragmentation for packets larger than PMTU, but not larger than the maximum packet size allowed by the current congestion control mechanism (CCMPS) [RFC4340].
Each message is identified by a sequence number. The sequence number is used to identify segments in acknowledgments, to detect unacknowledged segments, to measure RTT, etc. The protocol may support ordered or unordered delivery of data, and does not itself provide retransmission. There is a Data Checksum option, which contains a strong CRC, lets endpoints detect application data corruption. It also supports reduced checksum coverage, a partial integrity mechanisms similar to UDP-lIte.
Receiver flow control is supported: limiting the amount of unacknowledged data that can be outstanding at a given time.
A DCCP protocol instance can be extended [RFC4340] and tuned. Some features are sender-side only, requiring no negotiation with the receiver; some are receiver-side only, some are explicitly negotiated during connection setup.
DCCP supports negotiation of the congestion control profile, to provide Plug and Play congestion control mechanisms. examples of specified profiles include [RFC4341] [RFC4342] [RFC5662]. All IETF-defined methods provide Congestion Control.
DCCP use a Connect packet to start a session, and permits half-connections that allow each client to choose features it wishes to support. Simultaneous open [RFC5596], as in TCP, can enable interoperability in the presence of middleboxes. The Connect packet includes a Service Code field [RFC5595] designed to allow middle boxes and endpoints to identify the characteristics required by a session. A lightweight UDP-based encapsulation (DCCP-UDP) has been defined [RFC6773] that permits DCCP to be used over paths where it is not natively supported. Support in NAPT/NATs is defined in [RFC4340] and [RFC5595].
Upper layer protocols specified on top of DCCP include: DTLS [RFC5595], RTP [RFC5672], ICE/SDP [RFC6773].
A DCCP service is unicast.
A common packet format has allowed tools to evolve that can read and interpret DCCP packets (e.g. Wireshark).
API charactersitics include: - Datagram transmission. - Notification of the current maximum packet size. - Send and reception of zero-length payloads. - Set the Slow Receiver flow control at areceiver. - Detct a Slow receiver at the sender.
There is no current API specified in the RFC Series.
The transport protocol components provided by DCCP are:
RTP provides an end-to-end network transport service, suitable for applications transmitting real-time data, such as audio, video or data, over multicast or unicast network services, including TCP, UDP, UDP-Lite, DCCP.
[EDITOR’S NOTE: Varun Singh signed up as contributor for this section.]
pseudo transport
[NOTE: A few words on TLS [RFC5246] and DTLS [RFC6347] here, and how they get used by other protocols to meet security goals as an add-on interlayer above transport.]
[EDITOR’S NOTE: No identified contributor for this section yet.]
[EDITOR’S NOTE: No identified contributor for this section yet.]
[EDITOR’S NOTE: this section will drawn from the candidate features provided by protocol components in the previous section – please discuss on taps@ietf.org list]
[EDITOR’S NOTE: Dave Thaler has signed up as a contributor for this section. Michael Welzl also has a beginning of a matrix which could be useful here.]
[EDITOR’S NOTE: The below is a strawman proposal below by Gorry Fairhurst for initial discussion]
The table below summarises protocol mechanisms that have been standardised. It does not make an assessment on whether specific implementations are fully compliant to these specifications.
| Mechanism | UDP | UDP-L | DCCP | SCTP | TCP |
|---|---|---|---|---|---|
| Unicast | Yes | Yes | Yes | Yes | Yes |
| Mcast/IPv4Bcast | Yes(2) | Yes | No | No | No |
| Port Mux | Yes | Yes | Yes | Yes | Yes |
| Mode | Dgram | Dgram | Dgram | Stream | Stream |
| Connected | No | No | Yes | Yes | Yes |
| Data bundling | No | No | No | No | Yes |
| Feature Nego | No | No | Yes | Yes | Yes |
| Options | No | No | Support | Support | Support |
| Data priority | * | * | * | Yes | No |
| Data bundling | No | No | No | No | Yes |
| Reliability | None | None | None | Select | Full |
| Ordered deliv | No | No | No | Stream | Yes |
| Corruption Tol. | No | Support | Support | No | No |
| Flow Control | No | No | Support | Yes | Yes |
| PMTU/PLPMTU | (1) | (1) | Yes | Yes | Yes |
| Cong Control | (1) | (1) | Yes | Yes | Yes |
| ECN Support | (1) | (1) | Yes | No | Yes |
| NAT support | Limited | Limited | Support | TBD | Support |
| Security | DTLS | DTLS | DTLS | DTLS | TLS, AO |
| UDP encaps | N/A | None | Yes | Yes | None |
| RTP support | Support | Support | Support | ? | Support |
Note (1): this feature requires support in an upper layer protocol.
Note (2): this feature requires support in an upper layer protocol when used with IPv6.
This document has no considerations for IANA.
This document surveys existing transport protocols and protocols providing transport-like services. Confidentiality, integrity, and authenticity are among the features provided by those services. This document does not specify any new components or mechanisms for providing these features. Each RFC listed in this document discusses the security considerations of the specification it contains.
[EDITOR’S NOTE: Non-editor contributors of text will be listed here, as noted in the authors section.]
This work is partially supported by the European Commission under grant agreement FP7-ICT-318627 mPlane; support does not imply endorsement.
| [RFC0791] | Postel, J., "Internet Protocol", STD 5, RFC 791, September 1981. |